Research ArticleFPGA Implementation of Digital Images Watermarking SystemBased on Discrete Haar Wavelet Transform
Mohamed Ali Hajjaji 1 Mohamed Gafsi 12
Abdessalem Ben Abdelali 1 and Abdellatif Mtibaa12
1Universite de Monastir Laboratoire dElectronique et de Microelectronique LR99ES30 5000 Monastir Tunisia2Universite de Monastir Ecole Nationale drsquoIngenieurs de Monastir 5000 Monastir Tunisia
Correspondence should be addressed to Mohamed Ali Hajjaji daly fsmyahoofr
Received 14 August 2018 Revised 11 November 2018 Accepted 9 December 2018 Published 3 January 2019
Academic Editor Stelvio Cimato
Copyright copy 2019 Mohamed Ali Hajjaji et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited
In this paper we propose a novel and efficient hardware implementation of an image watermarking system based on the HaarDiscreteWavelet Transform (DWT) DWT is used in image watermarking to hide secret pieces of information into a digital contentwith a good robustnessThemain advantage of Haar DWT is the frequencies separation into four subbands (LL LH HL and HH)which can be treated independently This permits ensuring a better compromise between robustness and visibility factors A FieldProgrammable Gate Array (FPGA) that is based on a very large scale integration architecture of the watermarking algorithm isdeveloped to accelerate media authentication A hardware cosimulation strategy using the Matlab-Xilinx system generator (XSG)was applied to prove the validity of the suggested implementationThe hardware cosimulation results show the effectiveness of thedeveloped architecture in terms of visibility and robustness against several attacks The proposed hardware system presents also ahigh performance in terms of the operating speed
1 Introduction
Digital watermarking is a technique of hiding informationon a digital support such as images video or audio forauthentication control copyright protection integrity veri-fication etc The hidden information is called a watermarkand the marked documents are named watermarked dataDistortion caused by the hidden watermark on the hostdata should be made as low as possible The watermarkedand original images must be perceptually equivalent so thatthe embedded watermark can remain imperceptible by aHumanVisual System (HVS)The Peak Signal to Noise Ratio(PSNR) parameter is used for the imperceptibility measureEven if the distortion caused by the watermark is smallit can be undesirable in some image types such as themedical and military ones For these types of applicationsthe PSNR value must be greater than 40 dB [1] In this casewatermarking in the transform domain is recommended Infact transform spaces such as Discrete Cosine Transform
(DCT) Discrete Wavelet Transform (DWT) and KarhunenLoeve Transform (KLT) provide a special authentication tohost images They are especially used in telemedicine e-healthcare legal domains telesurgery etc
The performance of a watermarking system is generallysubject to the following requirements
(i) Imperceptibility The watermark should not affect thequality of the original image after any watermarking oper-ation Cox et al [2] defined the imperceptibility as a visualsimilarity between the original and watermarked imagesThewatermark has to be inserted in away that it still is completelyinvisible to HVS [3] Indeed the insertion process must notdamage the host image However not only the image but alsothe watermark should not be distorted This latter must beinvisible but also easy to extract
(ii) Capacity The ability of a watermarking system refers tothe ratio of the amount of data to be hidden according to
HindawiSecurity and Communication NetworksVolume 2019 Article ID 1294267 17 pageshttpsdoiorg10115520191294267
2 Security and Communication Networks
the size of the host document [4] Sometimes the size of thewatermark is limited just to 1 bit
(iii) Robustness Robustness is the resistance of the watermarksystem against intentional transformations on a watermarkedimage [5]These transformations can be of a given geometrictype such as rotation and cropping and they include all typesof image degradation caused by lossy compression high-passfilter low-pass filter etc
To these requirements we can add the computationalcomplexity In fact execution time can be an importantfactor for many applications Watermarking algorithms witha low computation cost can be used to reduce the executiontimeHowever this can highly affect the system performanceElsewhere the algorithm can be adapted for hardware imple-mentation to accelerate the processing while maintainingthe techniques effectiveness [6] In the related literaturesoftware implementation of the watermarking algorithmsis largely applied in contrast to hardware implementationdespite the performance that can be achieved by applyingthis type of development [7] In a software implementationthe algorithmrsquos operations are performed as a code runningon a microprocessor [8] The main drawback of this typeof implementation [8] is the limited means for improvingthe system speed and the hardware performances Althoughit might be faster to implement an algorithm in softwarethere are a few compelling reasons for a move to hard-ware implementation In this kind of implementation thealgorithmrsquos operations are fully implemented in a custom-designed circuitry This investigates great advantages such ashardware area and consumption decrease and mainly speedincrease [7ndash9]
In the given literature a number of hardware designs forconventional watermarking algorithms have been reportedThe Very Large Scale Integration (VLSI) architecture for aconventional watermarking algorithm in the spatial domainproposed by Gerimella et al [10] might be considered as anoteworthy early work Later Mohanty et al [11] proposeda watermarking hardware architecture that can insert twovisible watermarks into digital images using a spatial domainwatermarking technique Mohanty et al [12] put forward aVLSI architecture that could insert invisible or visible water-marks into digital images in the DCT domain Mohanty et al[13] developed two versions (low-power high-performance)of watermarking hardware module The DC component andthe three low frequency components are considered forinsertion in the DCT domain Maity et al [14] suggested afastWalsh transform (FWT) based on a Spread Spectrum (SS)image watermarking scheme that would serve for authen-tication in data transmission In [15] Korrapati Rajitha etal proposed an FPGA implementation of a watermarkingsystem using the Xilinx System Generator (XSG) Insertionand extraction of information were applied in the spatialdomain In [16] Rohollah Mazrae Khoshki et al put forwarda hardware implementation of a watermarking system basedon DCT Their work was developed using Matlab-Simulinkfollowed by Altera DSP Builder (integrated with SimulinkEmbedded coder) for Auto-Code generation In [17] RahateKunal B et al suggested a hardware implementation of a
fragile watermarking system operating in the spatial domainTheir proposed watermarking scheme was imperceptible androbust against geometric attacks but fragile against filteringand compression Hirak Kumar Maitya et al [6] put forwarda hardware implementation of reversible watermarking inthe spatial domain by using a reversible contrast mappingtechniqueThe principal advantage of the proposed work wasthe operation frequency (more than 9876 MHz) In [18]Sakthivel and SM et al put forward a VLSI architecture of adigital imagewatermarking systemTheir embedding processwas based on the Pixel Value Search Algorithm (PVSA)applied in the spatial domain The system was implementedusing verilog Hardware Description Language (HDL) andthe Altera Quartus-II 110 tool with Matlab R-2012b Thepresented results showed that the proposed system was nothighly fast with an average quality of the watermarked imageand the extracted watermark resulting in different attacks In[19] Manas N et al suggested a hardware implementationof a watermarking algorithm based on phase congruencyand singular value decomposition Their idea consisted inembedding watermark data in the host image using the Sin-gular Value Decomposition (SVD) in the congruency phasemapping points applied in the spatial domain Their systemwas implemented using the Xilinx ISE 143 tool and a Virtex5 FPGA device In [6] Hirak M et al proposed an FPGAimplementation of an image watermarking algorithm usingReversible Contrast Mapping (RCM) in the spatial domainThe implemented algorithm and the resulting architecturewere relatively simple In [20] Karthigai kumara P et al putforward an FPGA implementation of an image watermarkingsystem using the XSG tool Their suggested system consistsin embedding a binary watermark in the discrete waveletdomain of a host image The main disadvantage of theproposed system is that the corresponding hardware designconsumed a lot of hardware resources despite that the systemused only the DWT tool
After this review of the existing work that addresses thehardware implementation of watermarking systems we cannote that the majority of their present inefficiency is in termsof hardware performances or in terms of robustness of thehardware design against attacks Many of them are applied inthe spatial domainwith some time very simple techniques tobe implemented as well as a lack of hardware speed efficiencyHowever hiding confidential data in the spatial domain isgenerally vulnerable against hackers In this work we suggesta novel and efficient hardware implementation of a water-marking system based onHaar DWTWe aim at developing awatermarking system that ensures high performance in termsof hardware efficiency with high imperceptibility (PSNR) androbustness (Normalized Cross-Correlation NC)The systemis designed using the XSG tool and synthesized for XilinxVirtex-5 FPGA of the ML507 platform A comparison withexisting watermarking systems will be undergone to show theeffectiveness of the proposed module in terms of hardwareperformances with the high imperceptibility and robustnessagainst several attacks
The rest of the paper is organized as follows In Section 2a description of the different steps of the adopted watermark-ing algorithm is given In Section 3 we describe the hardware
Security and Communication Networks 3
2-D DWT at second level
Watermarked image
Key
WatermarkOriginal image
IDWT 2-D
X
XOR
+Wrsquo
LH1LL2
HL2 HH2
HL1 HH1
LH2
Figure 1 The insertion step
design of the watermarking system The implementationresults and the performance evaluation of the developedwatermarking system are presented in Section 4
2 Description of Watermarking Algorithm
Watermarking systems of digital images are composed oftwo main parts insertion and detection [21] The diffusionprocess includes the attacks applied to watermarked images
21 Insertion Step As illustrated in Figure 1 the proposedsystem is an additive scheme The watermark insertion isexpressed by
119875119900119903119894 = i119905ℎ original coefficient
119882119894 = i119905ℎ bit of the watermark
(1)
In the insertion phase our system requires four datainputs
(i) The original image (I) that will contain the data to bepreserved and protected
Key
Extractedwatermark
XOR
Thresholding
Ciphered watermark
Watermarked imageOriginal Original iimagemage Watermarked imageWatermarked imageOriginal Original iimagemageriginal riginal
Second level 2D-DWT
Watermarked LH2Original LH2
Original image
Subtraction
Figure 2 The extraction step
(ii) The watermark (W) which represents the informa-tion to be inserted (a binary information)
(iii) The key (C) which is a binary sequence to be mixedwith the watermark for its protection
(iv) The visibility factor (120572) which is themarking strengthin the image This coefficient must be adequatelychosen to maintain a best compromise betweenrobustness and imperceptibility factors of the scheme
After the second level of decomposition using the 2DHaar DWTwe obtain four subbands of 18 of the input imagesize (Figure 1) approximation (LL2 band low frequencies)and details (horizontal (LH2) vertical (HL2) and diagonal(HH2)) In our adopted method we opt for inserting thewatermark in the LH2 subband which includes the mediumfrequencies In the end of this phase 2D IDWT is applied toconstruct the watermarked image
22 Extraction Step As depicted in Figure 2 the extractionstep consists in following the same steps as in the insertionphase The 2D Haar DWT is applied at the second level of
4 Security and Communication Networks
the decomposition After that the watermark is recovered byusing the following equation
1198821015840 (119894) = [ [119871119867119905 (119894) minus 119871119867119900 (119894)]120572 oplus 119862 (119894)]119882119894119905ℎ 119871119867119905 119882119886119905119890119903119898119886119903119896119890119889 119904119906119887 minus 119887119886119899119889
119871119867119900 119874119903119894119892119894119899119886119897 119904119906119887 minus 119887119886119899119889(2)
The watermarked image may be subject to alterations causedby attacks Indeed a thresholding phase is necessary for theproper extraction of the watermark Equation (3) is appliedto set the value of the watermark
119894119891Ρ119886119897 (i) ge S 997904rArr W (i) = 1119890119897119904119890 997904rArr W (i) = 0
with Pal (i) ith is the difference between watermarked and the original coefficient
S Threshold value determined empirically
(3)
3 Hardware Design ofthe Watermarking System
Xilinx Company proposes an Integrator Design Environment(IDE) for FPGA under the Matlab tool This IDE is aiming toincrease the abstraction level of the hardware design and tominimize the manual intervention of the HDL code genera-tion [22]This tool is named XSG it is a high-level design toolthat allows using the MathWorks Simulink environment inthe design of digital circuits dedicated to Xilinx FPGAs [23]It is used for hardware system generation simulation andvalidation throughout the hardware cosimulation technique
The structure of a system is created in the Simulinkmodeling environment using a specific library offered byXilinx All the designing steps for the implementation onFPGA including synthesis placement and routing areautomatically performed to generate an FPGA programmingfile
The designer starts with creating the system model inSimulink Next ldquoSysgenrdquo automatically generates the bit-stream to program the FPGA Intermediate steps which aresynthesis placement and routing are performed by interme-diate tools Figure 3 describes the XSG based design flow
In our design the acquisition and display of input andoutput images are performed using the Matlab tool Atthis phase data are presented in a double-precision float-ing number The processing algorithm is implemented byusing XSG blocks In the XSG design boolean and fixed-point formats are used for data representation To adaptthe representation differences between the XSG design andthe Matlab software part Xilinx offers a simple interfacingutilizing predefined ldquoGateway-Inrdquo and ldquoGateway-Outrdquo blocksprovided in the Xilinx Blockset Library The global designof the watermarking system is divided into two principalmodules insertion and extraction
31 Insertion Module As shown in Figure 4 the globaldesign of the insertion module is composed of two mainblocks The first one corresponds to the decomposition and
reconstruction of the DWT at the second level The secondone corresponds to the insertion step
311 2D Haar DWT
(a) Decomposition Step of the 2D DWT of Haar The one-dimensional decomposition is obtained by applying theequations of the decomposition ldquoArdquo for approximations andldquoMrdquo for details
A = X(2n) + X(2n+1)2 X Input signal
M = Y(2n) minus Y(2n+1)2 Y Output signal(4)
As shown in Figure 5 the Haar wavelet decomposition intwo dimensions is mainly performed in two stages The firststage consists in applying (1) and (2) along lines This allowsobtaining two subbands generally denoted as L and HThena transposition is made in order to reach the second stagewhich consists in applying the same equations on columnsSo the four subbands named LL LH HL and HH will beobtained
Figure 6 gives the various parts of the 2D Haar DWTglobal design
(i) Preprocessing Subsystem The preprocessing subsystemallows the preparation of the input data for accelerating thewavelet computing The idea consists in decomposing theentire image into four components so separating the evenand odd pixels from each even and odd image line Thisprocess allows performing the wavelet steps in one go Thedesign is presented in Figure 7
(ii) Calculation of the Subsystem This subsystem computesthe coefficient of wavelet fieldThus it receives and processesthe outputs of the preprocessing subsystem in order toproduce four outputs which are LL LH HL and HHcoefficients Obviously as shown in Figure 8 the calculation
Security and Communication Networks 5
MATLAB
mdl
Simulink
m
ISIM ModelSIMWaveform
Test benchISE DS
Model Simulation
Implementation
Synthesis
FPGA
bit
Exte
rnal
Si
mul
ator
HW in the loop HW CoSIM
Figure 3 XSG based design flow
SystemGenerator
Ol ln1
OW ln2
pretreatment 2_2D-DWT-HAAR
Insertion
[A]
[B]
[C]
[D]
[A]
[B]
[C]
[D]
Go1
Go2
Go3
Go4
Fro1
Fro2
Fro3
Fro42D-IDWT-HAAR
W_l To W
Figure 4 Part 1 XSG blocks for the insertion phase
is done by the addition subtraction and multiplicationblocks
(iii) Storage Subsystem After wavelet computing a storagestage is required hence we present the objective of ldquoStoragerdquosubsystem However to accelerate the writeread of data
we have opted for using internal RAM blocks The Storagesubsystem design is presented in Figure 9
(b) Inverse Transformation of 2D DWT of Haar The principleof calculating the coefficients of the original image is depictedin Figure 10 From four subbands (LL LH HH and HL) the
6 Security and Communication Networks
X11 X12 X18
X21
X81 X82 X88
L11
L21
L14
L81 L84
H11 H14
H21
H81 H84
LL11 LL14
LL41
HL11 HL14 HH11
LL44
LH11 LH14
LH44LH41
HH14
HL41 HL44 HH41 HH44
Tran
sform
atio
n
on th
e row
s
Original Matrix sized 8lowast8
LL0 and LH0 HL0 and HH0
Transformation on columns
L and H M
atrix
Figure 5 Principle of the 2D Haar DWT
Pre-processing
Calculation
Input monochrome image
Storage
Output
Figure 6 Different blocks of the proposed architectures of the 2DHaar DWT
first step is to calculate L and HThen in the second step theoriginal pixels are calculated
The process of calculation is as follows(i) We begin with the computation of the L subband
coefficients This is done by browsing at the same timethe two LL and LH subbands along the columns using thefollowing equations
119871 (2 times 119894) = 119871119871 (119895) + 119871119867 (119895)2
119871 (2 times 119894 minus 1) = 119871119871 (119895) minus 119871119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = number of pixels in a column
(5)
(ii) Thereafter with HL and HH we calculate the coeffi-cients of band HThis is achieved by the following equations
119867(2 times 119894) = 119867119871 (119895) + 119867119867 (119895)2
119867 (2 times 119894 minus 1) = 119867119871 (119895) minus 119867119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = nombre de pixels drsquoun colonnes
(6)
(iii) At this stage the original pixels are calculatedbrowsing at the same time the two bands L and H along linesusing the following equations
119875 119900119903119894 (2 times 119894 minus 1) = 119871 (119894) minus 119867 (119894)2
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
2 Security and Communication Networks
the size of the host document [4] Sometimes the size of thewatermark is limited just to 1 bit
(iii) Robustness Robustness is the resistance of the watermarksystem against intentional transformations on a watermarkedimage [5]These transformations can be of a given geometrictype such as rotation and cropping and they include all typesof image degradation caused by lossy compression high-passfilter low-pass filter etc
To these requirements we can add the computationalcomplexity In fact execution time can be an importantfactor for many applications Watermarking algorithms witha low computation cost can be used to reduce the executiontimeHowever this can highly affect the system performanceElsewhere the algorithm can be adapted for hardware imple-mentation to accelerate the processing while maintainingthe techniques effectiveness [6] In the related literaturesoftware implementation of the watermarking algorithmsis largely applied in contrast to hardware implementationdespite the performance that can be achieved by applyingthis type of development [7] In a software implementationthe algorithmrsquos operations are performed as a code runningon a microprocessor [8] The main drawback of this typeof implementation [8] is the limited means for improvingthe system speed and the hardware performances Althoughit might be faster to implement an algorithm in softwarethere are a few compelling reasons for a move to hard-ware implementation In this kind of implementation thealgorithmrsquos operations are fully implemented in a custom-designed circuitry This investigates great advantages such ashardware area and consumption decrease and mainly speedincrease [7ndash9]
In the given literature a number of hardware designs forconventional watermarking algorithms have been reportedThe Very Large Scale Integration (VLSI) architecture for aconventional watermarking algorithm in the spatial domainproposed by Gerimella et al [10] might be considered as anoteworthy early work Later Mohanty et al [11] proposeda watermarking hardware architecture that can insert twovisible watermarks into digital images using a spatial domainwatermarking technique Mohanty et al [12] put forward aVLSI architecture that could insert invisible or visible water-marks into digital images in the DCT domain Mohanty et al[13] developed two versions (low-power high-performance)of watermarking hardware module The DC component andthe three low frequency components are considered forinsertion in the DCT domain Maity et al [14] suggested afastWalsh transform (FWT) based on a Spread Spectrum (SS)image watermarking scheme that would serve for authen-tication in data transmission In [15] Korrapati Rajitha etal proposed an FPGA implementation of a watermarkingsystem using the Xilinx System Generator (XSG) Insertionand extraction of information were applied in the spatialdomain In [16] Rohollah Mazrae Khoshki et al put forwarda hardware implementation of a watermarking system basedon DCT Their work was developed using Matlab-Simulinkfollowed by Altera DSP Builder (integrated with SimulinkEmbedded coder) for Auto-Code generation In [17] RahateKunal B et al suggested a hardware implementation of a
fragile watermarking system operating in the spatial domainTheir proposed watermarking scheme was imperceptible androbust against geometric attacks but fragile against filteringand compression Hirak Kumar Maitya et al [6] put forwarda hardware implementation of reversible watermarking inthe spatial domain by using a reversible contrast mappingtechniqueThe principal advantage of the proposed work wasthe operation frequency (more than 9876 MHz) In [18]Sakthivel and SM et al put forward a VLSI architecture of adigital imagewatermarking systemTheir embedding processwas based on the Pixel Value Search Algorithm (PVSA)applied in the spatial domain The system was implementedusing verilog Hardware Description Language (HDL) andthe Altera Quartus-II 110 tool with Matlab R-2012b Thepresented results showed that the proposed system was nothighly fast with an average quality of the watermarked imageand the extracted watermark resulting in different attacks In[19] Manas N et al suggested a hardware implementationof a watermarking algorithm based on phase congruencyand singular value decomposition Their idea consisted inembedding watermark data in the host image using the Sin-gular Value Decomposition (SVD) in the congruency phasemapping points applied in the spatial domain Their systemwas implemented using the Xilinx ISE 143 tool and a Virtex5 FPGA device In [6] Hirak M et al proposed an FPGAimplementation of an image watermarking algorithm usingReversible Contrast Mapping (RCM) in the spatial domainThe implemented algorithm and the resulting architecturewere relatively simple In [20] Karthigai kumara P et al putforward an FPGA implementation of an image watermarkingsystem using the XSG tool Their suggested system consistsin embedding a binary watermark in the discrete waveletdomain of a host image The main disadvantage of theproposed system is that the corresponding hardware designconsumed a lot of hardware resources despite that the systemused only the DWT tool
After this review of the existing work that addresses thehardware implementation of watermarking systems we cannote that the majority of their present inefficiency is in termsof hardware performances or in terms of robustness of thehardware design against attacks Many of them are applied inthe spatial domainwith some time very simple techniques tobe implemented as well as a lack of hardware speed efficiencyHowever hiding confidential data in the spatial domain isgenerally vulnerable against hackers In this work we suggesta novel and efficient hardware implementation of a water-marking system based onHaar DWTWe aim at developing awatermarking system that ensures high performance in termsof hardware efficiency with high imperceptibility (PSNR) androbustness (Normalized Cross-Correlation NC)The systemis designed using the XSG tool and synthesized for XilinxVirtex-5 FPGA of the ML507 platform A comparison withexisting watermarking systems will be undergone to show theeffectiveness of the proposed module in terms of hardwareperformances with the high imperceptibility and robustnessagainst several attacks
The rest of the paper is organized as follows In Section 2a description of the different steps of the adopted watermark-ing algorithm is given In Section 3 we describe the hardware
Security and Communication Networks 3
2-D DWT at second level
Watermarked image
Key
WatermarkOriginal image
IDWT 2-D
X
XOR
+Wrsquo
LH1LL2
HL2 HH2
HL1 HH1
LH2
Figure 1 The insertion step
design of the watermarking system The implementationresults and the performance evaluation of the developedwatermarking system are presented in Section 4
2 Description of Watermarking Algorithm
Watermarking systems of digital images are composed oftwo main parts insertion and detection [21] The diffusionprocess includes the attacks applied to watermarked images
21 Insertion Step As illustrated in Figure 1 the proposedsystem is an additive scheme The watermark insertion isexpressed by
119875119900119903119894 = i119905ℎ original coefficient
119882119894 = i119905ℎ bit of the watermark
(1)
In the insertion phase our system requires four datainputs
(i) The original image (I) that will contain the data to bepreserved and protected
Key
Extractedwatermark
XOR
Thresholding
Ciphered watermark
Watermarked imageOriginal Original iimagemage Watermarked imageWatermarked imageOriginal Original iimagemageriginal riginal
Second level 2D-DWT
Watermarked LH2Original LH2
Original image
Subtraction
Figure 2 The extraction step
(ii) The watermark (W) which represents the informa-tion to be inserted (a binary information)
(iii) The key (C) which is a binary sequence to be mixedwith the watermark for its protection
(iv) The visibility factor (120572) which is themarking strengthin the image This coefficient must be adequatelychosen to maintain a best compromise betweenrobustness and imperceptibility factors of the scheme
After the second level of decomposition using the 2DHaar DWTwe obtain four subbands of 18 of the input imagesize (Figure 1) approximation (LL2 band low frequencies)and details (horizontal (LH2) vertical (HL2) and diagonal(HH2)) In our adopted method we opt for inserting thewatermark in the LH2 subband which includes the mediumfrequencies In the end of this phase 2D IDWT is applied toconstruct the watermarked image
22 Extraction Step As depicted in Figure 2 the extractionstep consists in following the same steps as in the insertionphase The 2D Haar DWT is applied at the second level of
4 Security and Communication Networks
the decomposition After that the watermark is recovered byusing the following equation
1198821015840 (119894) = [ [119871119867119905 (119894) minus 119871119867119900 (119894)]120572 oplus 119862 (119894)]119882119894119905ℎ 119871119867119905 119882119886119905119890119903119898119886119903119896119890119889 119904119906119887 minus 119887119886119899119889
119871119867119900 119874119903119894119892119894119899119886119897 119904119906119887 minus 119887119886119899119889(2)
The watermarked image may be subject to alterations causedby attacks Indeed a thresholding phase is necessary for theproper extraction of the watermark Equation (3) is appliedto set the value of the watermark
119894119891Ρ119886119897 (i) ge S 997904rArr W (i) = 1119890119897119904119890 997904rArr W (i) = 0
with Pal (i) ith is the difference between watermarked and the original coefficient
S Threshold value determined empirically
(3)
3 Hardware Design ofthe Watermarking System
Xilinx Company proposes an Integrator Design Environment(IDE) for FPGA under the Matlab tool This IDE is aiming toincrease the abstraction level of the hardware design and tominimize the manual intervention of the HDL code genera-tion [22]This tool is named XSG it is a high-level design toolthat allows using the MathWorks Simulink environment inthe design of digital circuits dedicated to Xilinx FPGAs [23]It is used for hardware system generation simulation andvalidation throughout the hardware cosimulation technique
The structure of a system is created in the Simulinkmodeling environment using a specific library offered byXilinx All the designing steps for the implementation onFPGA including synthesis placement and routing areautomatically performed to generate an FPGA programmingfile
The designer starts with creating the system model inSimulink Next ldquoSysgenrdquo automatically generates the bit-stream to program the FPGA Intermediate steps which aresynthesis placement and routing are performed by interme-diate tools Figure 3 describes the XSG based design flow
In our design the acquisition and display of input andoutput images are performed using the Matlab tool Atthis phase data are presented in a double-precision float-ing number The processing algorithm is implemented byusing XSG blocks In the XSG design boolean and fixed-point formats are used for data representation To adaptthe representation differences between the XSG design andthe Matlab software part Xilinx offers a simple interfacingutilizing predefined ldquoGateway-Inrdquo and ldquoGateway-Outrdquo blocksprovided in the Xilinx Blockset Library The global designof the watermarking system is divided into two principalmodules insertion and extraction
31 Insertion Module As shown in Figure 4 the globaldesign of the insertion module is composed of two mainblocks The first one corresponds to the decomposition and
reconstruction of the DWT at the second level The secondone corresponds to the insertion step
311 2D Haar DWT
(a) Decomposition Step of the 2D DWT of Haar The one-dimensional decomposition is obtained by applying theequations of the decomposition ldquoArdquo for approximations andldquoMrdquo for details
A = X(2n) + X(2n+1)2 X Input signal
M = Y(2n) minus Y(2n+1)2 Y Output signal(4)
As shown in Figure 5 the Haar wavelet decomposition intwo dimensions is mainly performed in two stages The firststage consists in applying (1) and (2) along lines This allowsobtaining two subbands generally denoted as L and HThena transposition is made in order to reach the second stagewhich consists in applying the same equations on columnsSo the four subbands named LL LH HL and HH will beobtained
Figure 6 gives the various parts of the 2D Haar DWTglobal design
(i) Preprocessing Subsystem The preprocessing subsystemallows the preparation of the input data for accelerating thewavelet computing The idea consists in decomposing theentire image into four components so separating the evenand odd pixels from each even and odd image line Thisprocess allows performing the wavelet steps in one go Thedesign is presented in Figure 7
(ii) Calculation of the Subsystem This subsystem computesthe coefficient of wavelet fieldThus it receives and processesthe outputs of the preprocessing subsystem in order toproduce four outputs which are LL LH HL and HHcoefficients Obviously as shown in Figure 8 the calculation
Security and Communication Networks 5
MATLAB
mdl
Simulink
m
ISIM ModelSIMWaveform
Test benchISE DS
Model Simulation
Implementation
Synthesis
FPGA
bit
Exte
rnal
Si
mul
ator
HW in the loop HW CoSIM
Figure 3 XSG based design flow
SystemGenerator
Ol ln1
OW ln2
pretreatment 2_2D-DWT-HAAR
Insertion
[A]
[B]
[C]
[D]
[A]
[B]
[C]
[D]
Go1
Go2
Go3
Go4
Fro1
Fro2
Fro3
Fro42D-IDWT-HAAR
W_l To W
Figure 4 Part 1 XSG blocks for the insertion phase
is done by the addition subtraction and multiplicationblocks
(iii) Storage Subsystem After wavelet computing a storagestage is required hence we present the objective of ldquoStoragerdquosubsystem However to accelerate the writeread of data
we have opted for using internal RAM blocks The Storagesubsystem design is presented in Figure 9
(b) Inverse Transformation of 2D DWT of Haar The principleof calculating the coefficients of the original image is depictedin Figure 10 From four subbands (LL LH HH and HL) the
6 Security and Communication Networks
X11 X12 X18
X21
X81 X82 X88
L11
L21
L14
L81 L84
H11 H14
H21
H81 H84
LL11 LL14
LL41
HL11 HL14 HH11
LL44
LH11 LH14
LH44LH41
HH14
HL41 HL44 HH41 HH44
Tran
sform
atio
n
on th
e row
s
Original Matrix sized 8lowast8
LL0 and LH0 HL0 and HH0
Transformation on columns
L and H M
atrix
Figure 5 Principle of the 2D Haar DWT
Pre-processing
Calculation
Input monochrome image
Storage
Output
Figure 6 Different blocks of the proposed architectures of the 2DHaar DWT
first step is to calculate L and HThen in the second step theoriginal pixels are calculated
The process of calculation is as follows(i) We begin with the computation of the L subband
coefficients This is done by browsing at the same timethe two LL and LH subbands along the columns using thefollowing equations
119871 (2 times 119894) = 119871119871 (119895) + 119871119867 (119895)2
119871 (2 times 119894 minus 1) = 119871119871 (119895) minus 119871119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = number of pixels in a column
(5)
(ii) Thereafter with HL and HH we calculate the coeffi-cients of band HThis is achieved by the following equations
119867(2 times 119894) = 119867119871 (119895) + 119867119867 (119895)2
119867 (2 times 119894 minus 1) = 119867119871 (119895) minus 119867119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = nombre de pixels drsquoun colonnes
(6)
(iii) At this stage the original pixels are calculatedbrowsing at the same time the two bands L and H along linesusing the following equations
119875 119900119903119894 (2 times 119894 minus 1) = 119871 (119894) minus 119867 (119894)2
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 3
2-D DWT at second level
Watermarked image
Key
WatermarkOriginal image
IDWT 2-D
X
XOR
+Wrsquo
LH1LL2
HL2 HH2
HL1 HH1
LH2
Figure 1 The insertion step
design of the watermarking system The implementationresults and the performance evaluation of the developedwatermarking system are presented in Section 4
2 Description of Watermarking Algorithm
Watermarking systems of digital images are composed oftwo main parts insertion and detection [21] The diffusionprocess includes the attacks applied to watermarked images
21 Insertion Step As illustrated in Figure 1 the proposedsystem is an additive scheme The watermark insertion isexpressed by
119875119900119903119894 = i119905ℎ original coefficient
119882119894 = i119905ℎ bit of the watermark
(1)
In the insertion phase our system requires four datainputs
(i) The original image (I) that will contain the data to bepreserved and protected
Key
Extractedwatermark
XOR
Thresholding
Ciphered watermark
Watermarked imageOriginal Original iimagemage Watermarked imageWatermarked imageOriginal Original iimagemageriginal riginal
Second level 2D-DWT
Watermarked LH2Original LH2
Original image
Subtraction
Figure 2 The extraction step
(ii) The watermark (W) which represents the informa-tion to be inserted (a binary information)
(iii) The key (C) which is a binary sequence to be mixedwith the watermark for its protection
(iv) The visibility factor (120572) which is themarking strengthin the image This coefficient must be adequatelychosen to maintain a best compromise betweenrobustness and imperceptibility factors of the scheme
After the second level of decomposition using the 2DHaar DWTwe obtain four subbands of 18 of the input imagesize (Figure 1) approximation (LL2 band low frequencies)and details (horizontal (LH2) vertical (HL2) and diagonal(HH2)) In our adopted method we opt for inserting thewatermark in the LH2 subband which includes the mediumfrequencies In the end of this phase 2D IDWT is applied toconstruct the watermarked image
22 Extraction Step As depicted in Figure 2 the extractionstep consists in following the same steps as in the insertionphase The 2D Haar DWT is applied at the second level of
4 Security and Communication Networks
the decomposition After that the watermark is recovered byusing the following equation
1198821015840 (119894) = [ [119871119867119905 (119894) minus 119871119867119900 (119894)]120572 oplus 119862 (119894)]119882119894119905ℎ 119871119867119905 119882119886119905119890119903119898119886119903119896119890119889 119904119906119887 minus 119887119886119899119889
119871119867119900 119874119903119894119892119894119899119886119897 119904119906119887 minus 119887119886119899119889(2)
The watermarked image may be subject to alterations causedby attacks Indeed a thresholding phase is necessary for theproper extraction of the watermark Equation (3) is appliedto set the value of the watermark
119894119891Ρ119886119897 (i) ge S 997904rArr W (i) = 1119890119897119904119890 997904rArr W (i) = 0
with Pal (i) ith is the difference between watermarked and the original coefficient
S Threshold value determined empirically
(3)
3 Hardware Design ofthe Watermarking System
Xilinx Company proposes an Integrator Design Environment(IDE) for FPGA under the Matlab tool This IDE is aiming toincrease the abstraction level of the hardware design and tominimize the manual intervention of the HDL code genera-tion [22]This tool is named XSG it is a high-level design toolthat allows using the MathWorks Simulink environment inthe design of digital circuits dedicated to Xilinx FPGAs [23]It is used for hardware system generation simulation andvalidation throughout the hardware cosimulation technique
The structure of a system is created in the Simulinkmodeling environment using a specific library offered byXilinx All the designing steps for the implementation onFPGA including synthesis placement and routing areautomatically performed to generate an FPGA programmingfile
The designer starts with creating the system model inSimulink Next ldquoSysgenrdquo automatically generates the bit-stream to program the FPGA Intermediate steps which aresynthesis placement and routing are performed by interme-diate tools Figure 3 describes the XSG based design flow
In our design the acquisition and display of input andoutput images are performed using the Matlab tool Atthis phase data are presented in a double-precision float-ing number The processing algorithm is implemented byusing XSG blocks In the XSG design boolean and fixed-point formats are used for data representation To adaptthe representation differences between the XSG design andthe Matlab software part Xilinx offers a simple interfacingutilizing predefined ldquoGateway-Inrdquo and ldquoGateway-Outrdquo blocksprovided in the Xilinx Blockset Library The global designof the watermarking system is divided into two principalmodules insertion and extraction
31 Insertion Module As shown in Figure 4 the globaldesign of the insertion module is composed of two mainblocks The first one corresponds to the decomposition and
reconstruction of the DWT at the second level The secondone corresponds to the insertion step
311 2D Haar DWT
(a) Decomposition Step of the 2D DWT of Haar The one-dimensional decomposition is obtained by applying theequations of the decomposition ldquoArdquo for approximations andldquoMrdquo for details
A = X(2n) + X(2n+1)2 X Input signal
M = Y(2n) minus Y(2n+1)2 Y Output signal(4)
As shown in Figure 5 the Haar wavelet decomposition intwo dimensions is mainly performed in two stages The firststage consists in applying (1) and (2) along lines This allowsobtaining two subbands generally denoted as L and HThena transposition is made in order to reach the second stagewhich consists in applying the same equations on columnsSo the four subbands named LL LH HL and HH will beobtained
Figure 6 gives the various parts of the 2D Haar DWTglobal design
(i) Preprocessing Subsystem The preprocessing subsystemallows the preparation of the input data for accelerating thewavelet computing The idea consists in decomposing theentire image into four components so separating the evenand odd pixels from each even and odd image line Thisprocess allows performing the wavelet steps in one go Thedesign is presented in Figure 7
(ii) Calculation of the Subsystem This subsystem computesthe coefficient of wavelet fieldThus it receives and processesthe outputs of the preprocessing subsystem in order toproduce four outputs which are LL LH HL and HHcoefficients Obviously as shown in Figure 8 the calculation
Security and Communication Networks 5
MATLAB
mdl
Simulink
m
ISIM ModelSIMWaveform
Test benchISE DS
Model Simulation
Implementation
Synthesis
FPGA
bit
Exte
rnal
Si
mul
ator
HW in the loop HW CoSIM
Figure 3 XSG based design flow
SystemGenerator
Ol ln1
OW ln2
pretreatment 2_2D-DWT-HAAR
Insertion
[A]
[B]
[C]
[D]
[A]
[B]
[C]
[D]
Go1
Go2
Go3
Go4
Fro1
Fro2
Fro3
Fro42D-IDWT-HAAR
W_l To W
Figure 4 Part 1 XSG blocks for the insertion phase
is done by the addition subtraction and multiplicationblocks
(iii) Storage Subsystem After wavelet computing a storagestage is required hence we present the objective of ldquoStoragerdquosubsystem However to accelerate the writeread of data
we have opted for using internal RAM blocks The Storagesubsystem design is presented in Figure 9
(b) Inverse Transformation of 2D DWT of Haar The principleof calculating the coefficients of the original image is depictedin Figure 10 From four subbands (LL LH HH and HL) the
6 Security and Communication Networks
X11 X12 X18
X21
X81 X82 X88
L11
L21
L14
L81 L84
H11 H14
H21
H81 H84
LL11 LL14
LL41
HL11 HL14 HH11
LL44
LH11 LH14
LH44LH41
HH14
HL41 HL44 HH41 HH44
Tran
sform
atio
n
on th
e row
s
Original Matrix sized 8lowast8
LL0 and LH0 HL0 and HH0
Transformation on columns
L and H M
atrix
Figure 5 Principle of the 2D Haar DWT
Pre-processing
Calculation
Input monochrome image
Storage
Output
Figure 6 Different blocks of the proposed architectures of the 2DHaar DWT
first step is to calculate L and HThen in the second step theoriginal pixels are calculated
The process of calculation is as follows(i) We begin with the computation of the L subband
coefficients This is done by browsing at the same timethe two LL and LH subbands along the columns using thefollowing equations
119871 (2 times 119894) = 119871119871 (119895) + 119871119867 (119895)2
119871 (2 times 119894 minus 1) = 119871119871 (119895) minus 119871119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = number of pixels in a column
(5)
(ii) Thereafter with HL and HH we calculate the coeffi-cients of band HThis is achieved by the following equations
119867(2 times 119894) = 119867119871 (119895) + 119867119867 (119895)2
119867 (2 times 119894 minus 1) = 119867119871 (119895) minus 119867119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = nombre de pixels drsquoun colonnes
(6)
(iii) At this stage the original pixels are calculatedbrowsing at the same time the two bands L and H along linesusing the following equations
119875 119900119903119894 (2 times 119894 minus 1) = 119871 (119894) minus 119867 (119894)2
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
4 Security and Communication Networks
the decomposition After that the watermark is recovered byusing the following equation
1198821015840 (119894) = [ [119871119867119905 (119894) minus 119871119867119900 (119894)]120572 oplus 119862 (119894)]119882119894119905ℎ 119871119867119905 119882119886119905119890119903119898119886119903119896119890119889 119904119906119887 minus 119887119886119899119889
119871119867119900 119874119903119894119892119894119899119886119897 119904119906119887 minus 119887119886119899119889(2)
The watermarked image may be subject to alterations causedby attacks Indeed a thresholding phase is necessary for theproper extraction of the watermark Equation (3) is appliedto set the value of the watermark
119894119891Ρ119886119897 (i) ge S 997904rArr W (i) = 1119890119897119904119890 997904rArr W (i) = 0
with Pal (i) ith is the difference between watermarked and the original coefficient
S Threshold value determined empirically
(3)
3 Hardware Design ofthe Watermarking System
Xilinx Company proposes an Integrator Design Environment(IDE) for FPGA under the Matlab tool This IDE is aiming toincrease the abstraction level of the hardware design and tominimize the manual intervention of the HDL code genera-tion [22]This tool is named XSG it is a high-level design toolthat allows using the MathWorks Simulink environment inthe design of digital circuits dedicated to Xilinx FPGAs [23]It is used for hardware system generation simulation andvalidation throughout the hardware cosimulation technique
The structure of a system is created in the Simulinkmodeling environment using a specific library offered byXilinx All the designing steps for the implementation onFPGA including synthesis placement and routing areautomatically performed to generate an FPGA programmingfile
The designer starts with creating the system model inSimulink Next ldquoSysgenrdquo automatically generates the bit-stream to program the FPGA Intermediate steps which aresynthesis placement and routing are performed by interme-diate tools Figure 3 describes the XSG based design flow
In our design the acquisition and display of input andoutput images are performed using the Matlab tool Atthis phase data are presented in a double-precision float-ing number The processing algorithm is implemented byusing XSG blocks In the XSG design boolean and fixed-point formats are used for data representation To adaptthe representation differences between the XSG design andthe Matlab software part Xilinx offers a simple interfacingutilizing predefined ldquoGateway-Inrdquo and ldquoGateway-Outrdquo blocksprovided in the Xilinx Blockset Library The global designof the watermarking system is divided into two principalmodules insertion and extraction
31 Insertion Module As shown in Figure 4 the globaldesign of the insertion module is composed of two mainblocks The first one corresponds to the decomposition and
reconstruction of the DWT at the second level The secondone corresponds to the insertion step
311 2D Haar DWT
(a) Decomposition Step of the 2D DWT of Haar The one-dimensional decomposition is obtained by applying theequations of the decomposition ldquoArdquo for approximations andldquoMrdquo for details
A = X(2n) + X(2n+1)2 X Input signal
M = Y(2n) minus Y(2n+1)2 Y Output signal(4)
As shown in Figure 5 the Haar wavelet decomposition intwo dimensions is mainly performed in two stages The firststage consists in applying (1) and (2) along lines This allowsobtaining two subbands generally denoted as L and HThena transposition is made in order to reach the second stagewhich consists in applying the same equations on columnsSo the four subbands named LL LH HL and HH will beobtained
Figure 6 gives the various parts of the 2D Haar DWTglobal design
(i) Preprocessing Subsystem The preprocessing subsystemallows the preparation of the input data for accelerating thewavelet computing The idea consists in decomposing theentire image into four components so separating the evenand odd pixels from each even and odd image line Thisprocess allows performing the wavelet steps in one go Thedesign is presented in Figure 7
(ii) Calculation of the Subsystem This subsystem computesthe coefficient of wavelet fieldThus it receives and processesthe outputs of the preprocessing subsystem in order toproduce four outputs which are LL LH HL and HHcoefficients Obviously as shown in Figure 8 the calculation
Security and Communication Networks 5
MATLAB
mdl
Simulink
m
ISIM ModelSIMWaveform
Test benchISE DS
Model Simulation
Implementation
Synthesis
FPGA
bit
Exte
rnal
Si
mul
ator
HW in the loop HW CoSIM
Figure 3 XSG based design flow
SystemGenerator
Ol ln1
OW ln2
pretreatment 2_2D-DWT-HAAR
Insertion
[A]
[B]
[C]
[D]
[A]
[B]
[C]
[D]
Go1
Go2
Go3
Go4
Fro1
Fro2
Fro3
Fro42D-IDWT-HAAR
W_l To W
Figure 4 Part 1 XSG blocks for the insertion phase
is done by the addition subtraction and multiplicationblocks
(iii) Storage Subsystem After wavelet computing a storagestage is required hence we present the objective of ldquoStoragerdquosubsystem However to accelerate the writeread of data
we have opted for using internal RAM blocks The Storagesubsystem design is presented in Figure 9
(b) Inverse Transformation of 2D DWT of Haar The principleof calculating the coefficients of the original image is depictedin Figure 10 From four subbands (LL LH HH and HL) the
6 Security and Communication Networks
X11 X12 X18
X21
X81 X82 X88
L11
L21
L14
L81 L84
H11 H14
H21
H81 H84
LL11 LL14
LL41
HL11 HL14 HH11
LL44
LH11 LH14
LH44LH41
HH14
HL41 HL44 HH41 HH44
Tran
sform
atio
n
on th
e row
s
Original Matrix sized 8lowast8
LL0 and LH0 HL0 and HH0
Transformation on columns
L and H M
atrix
Figure 5 Principle of the 2D Haar DWT
Pre-processing
Calculation
Input monochrome image
Storage
Output
Figure 6 Different blocks of the proposed architectures of the 2DHaar DWT
first step is to calculate L and HThen in the second step theoriginal pixels are calculated
The process of calculation is as follows(i) We begin with the computation of the L subband
coefficients This is done by browsing at the same timethe two LL and LH subbands along the columns using thefollowing equations
119871 (2 times 119894) = 119871119871 (119895) + 119871119867 (119895)2
119871 (2 times 119894 minus 1) = 119871119871 (119895) minus 119871119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = number of pixels in a column
(5)
(ii) Thereafter with HL and HH we calculate the coeffi-cients of band HThis is achieved by the following equations
119867(2 times 119894) = 119867119871 (119895) + 119867119867 (119895)2
119867 (2 times 119894 minus 1) = 119867119871 (119895) minus 119867119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = nombre de pixels drsquoun colonnes
(6)
(iii) At this stage the original pixels are calculatedbrowsing at the same time the two bands L and H along linesusing the following equations
119875 119900119903119894 (2 times 119894 minus 1) = 119871 (119894) minus 119867 (119894)2
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 5
MATLAB
mdl
Simulink
m
ISIM ModelSIMWaveform
Test benchISE DS
Model Simulation
Implementation
Synthesis
FPGA
bit
Exte
rnal
Si
mul
ator
HW in the loop HW CoSIM
Figure 3 XSG based design flow
SystemGenerator
Ol ln1
OW ln2
pretreatment 2_2D-DWT-HAAR
Insertion
[A]
[B]
[C]
[D]
[A]
[B]
[C]
[D]
Go1
Go2
Go3
Go4
Fro1
Fro2
Fro3
Fro42D-IDWT-HAAR
W_l To W
Figure 4 Part 1 XSG blocks for the insertion phase
is done by the addition subtraction and multiplicationblocks
(iii) Storage Subsystem After wavelet computing a storagestage is required hence we present the objective of ldquoStoragerdquosubsystem However to accelerate the writeread of data
we have opted for using internal RAM blocks The Storagesubsystem design is presented in Figure 9
(b) Inverse Transformation of 2D DWT of Haar The principleof calculating the coefficients of the original image is depictedin Figure 10 From four subbands (LL LH HH and HL) the
6 Security and Communication Networks
X11 X12 X18
X21
X81 X82 X88
L11
L21
L14
L81 L84
H11 H14
H21
H81 H84
LL11 LL14
LL41
HL11 HL14 HH11
LL44
LH11 LH14
LH44LH41
HH14
HL41 HL44 HH41 HH44
Tran
sform
atio
n
on th
e row
s
Original Matrix sized 8lowast8
LL0 and LH0 HL0 and HH0
Transformation on columns
L and H M
atrix
Figure 5 Principle of the 2D Haar DWT
Pre-processing
Calculation
Input monochrome image
Storage
Output
Figure 6 Different blocks of the proposed architectures of the 2DHaar DWT
first step is to calculate L and HThen in the second step theoriginal pixels are calculated
The process of calculation is as follows(i) We begin with the computation of the L subband
coefficients This is done by browsing at the same timethe two LL and LH subbands along the columns using thefollowing equations
119871 (2 times 119894) = 119871119871 (119895) + 119871119867 (119895)2
119871 (2 times 119894 minus 1) = 119871119871 (119895) minus 119871119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = number of pixels in a column
(5)
(ii) Thereafter with HL and HH we calculate the coeffi-cients of band HThis is achieved by the following equations
119867(2 times 119894) = 119867119871 (119895) + 119867119867 (119895)2
119867 (2 times 119894 minus 1) = 119867119871 (119895) minus 119867119867 (119895)2
119894 119895 = 1 997888rarr 1198732 119873 = nombre de pixels drsquoun colonnes
(6)
(iii) At this stage the original pixels are calculatedbrowsing at the same time the two bands L and H along linesusing the following equations
119875 119900119903119894 (2 times 119894 minus 1) = 119871 (119894) minus 119867 (119894)2
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
(iv) After this last step we have the original pixels Finallythe pixels are organized to reform the input image
For the implementation of the IDWT of Haar with XSGtools we propose the subsystem shown in Figure 10 Thusthe subsystem processes the coefficients of wavelet field inorder to acquire the original data Hence the computing ofthe original data is donewith addition and subtraction blocksAlso we use other logic blocks for data control and shaping
312 Hiding Watermark on the Host Image As presented inFigure 11 the second step is about the insertion system At
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 7
1
1
2
3
4
P1
P2
P3
P4
Figure 7 Block diagram of the ldquopre-processingrdquo subsystem
1
2
3
4
in1
in2Reg1
Reg2
Reg3
TDD1
TDD2
TDD3
TDD4
LL
LH
HL
HH
1
2
3
4
Figure 8 Block diagram of calculation sub-system
this step the totality of the watermark is embedded in LH2(second horizontal subband) The watermark is scrambledby a secret key generated by the ldquoLFSRrdquo block Afterwardthe ldquoDSP48 macrordquo block is used to carry out the additionof the scrambled watermark multiplied by the ldquo120572rdquo visibilityfactor
The inputs of the ldquoDSP48 macrordquo block are respectivelyLH2 alpha and the scrambled watermark Its output is thewatermarked LH2
32 Extraction Module The extraction step is the last phaseof the watermarking system which aims to extract the
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
8 Security and Communication Networks
Figure 9 Block diagram of storage subsystem
Output
Inputs
Figure 10 Block diagram of different subsystems of inverse Haar DWT
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 9
Input
Output
Figure 11 Watermark insertion model
inserted data Figure 12 represents the global design ofthe extraction system At this step the same procedure isreversely used
The main difference relative to the insertion step isthe extraction block Figure 13 presents the design of theextraction block We obtain the original and watermarkedsubbands After that a subtraction is applied to extract themodifiedwatermark namedWrsquoThe latter is stocked in FIFOFinally by using the thresholding the final watermark isextracted
In this section we start by presenting the hardware imple-mentation results of the adopted system Some examples ofthe cosimulation results of the generated hardware block willbe present The efficiency of the proposed system is thendiscussed according to the PSNR value between the originaland the watermarked image and the NC value between theoriginal and the detected watermark against several attacksA comparison with some existing works will be described inthe following
41 Cosimulation Results After the validation of the adoptedalgorithm by the software simulation we proceed to theimplementation on a Xilinx platform The configuration fileis obtained automatically by following the necessary stepsto convert the design into an FPGA synthesizable module(Figure 14)The target device selected for thiswork is Virtex-5FPGA on the ML507 platform
The hardware implementation of the insertion andextraction steps on the ML 507 target generates the resultsof the FPGA resource consumer in Table 1 The RegisterTransfer Level (RTL) diagrams of the insertion and extractionsystems are presented in Figure 15
For the validation of our study we considered an ordinaryimage base known as the image ldquoCameramanrdquo ldquoLenardquoldquoBarbarardquo etc In Figure 16 we present some implementationresults of the adopted watermarking system on the ldquoCamera-manrdquo image with a variation of the value 120572 (equal to 3 6 10and 20) However we notice that the increase in the visibilityfactor leads to the loss of the psychovisual quality of thewatermarked image It should be noted that in the absence ofattacks the watermark is well extracted from which we canconclude that the implemented system gives results similar tothose obtained by software implementation
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
10 Security and Communication Networks
Figure 12 Part 2 XSG blocks for extraction phase
Figure 13 Extraction Block
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 11
(a)
(b)
Figure 14 Hardware block generation of insertion and extraction steps
Table 1 Consumed hardware resources in insertion and extraction steps resources
Resources Insertion Step Extraction StepUsed Available Percentage Used Available Percentage
Number ofregister slices 1536 32640 4 619 32640 2
Number of sliceLUT 2092 32640 6 1002 32640 3
Number of usedlogic blocks
2494 32640 8 532 32640 1
Number of DSP48 1 48 2 1 48 2Number of BRAM 9 148 6 12 148 8
Maximum frequency =224 MHz Maximum frequency =232 MHz
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
12 Security and Communication Networks
(a)
(b)
Figure 15 RTL schematic of insertion and extraction steps
42 Performance Evaluation against Several Attacks of Imple-mented System Following the literature themain constraintsof the watermarked scheme are imperceptibility and robust-ness factors The first one is named PSNR and presented in(8) PSNR is accepted if its value is greater than 30 dB [24]The second one named NC is presented in (9)TheNC valueis accepted if its value is greater than 07 [25]
119860 times 119861 timesmax (119868(119894119895))2sum119860119894=1sum119861119895=1 (119868(119894119895) minus 1198681015840(119894119895))2
]]]
I(ij) I1015840(ij) pixel values of the pixel (i j)in the host and watermarked image
AB Width and Height of the host image
(8)
119873119862 = sum119898sum119899 (119883(119898119899) minus 119883) times (119884(119898119899) minus 119884)radic(sum119898sum119899 (119883(119898119899) minus 119883)2) times (sum119898sum119899 (119884(119898119899) minus 119884)2)
X and Y Mean values of the coefficients of two matrices
(9)
After several empirical tests applied to ordinary imageswith and without attacks we have found that for 120572 equal to3 the adopted algorithm ensures a maximum compromisebetween robustness and imperceptibility factors
Figure 17 shows the results of the hardware cosimulationof ordinary images in the absence of attacks It can beconcluded that the values of PSNR are well acceptable withrespect to the previous work and with respect to resultsobtained for the software implementation
Our algorithm implemented on the hardware is morerobust against other types of attacks After applying several
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 13
=20=10
=6=3Original Image
10
=6 33=3Original Image
Original Watermark
Extracted Watermark
Figure 16 Hardware co-simulation results of insertion system
PSNR= 484715 PSNR =482478 PSNR=484715 PSNR=483581
Figure 17 Hardware co-simulation results of watermarked ordinary images without attacks
attacks we extract the watermark and we compare it to theoriginal one The main goal is to ensure that the extractedwatermark is not modified by attacks It is important toget an NC value close to 1 and a good PSNR value Therobustness against diverse types of attacks such as JPEG 2000attacks impulsive noisemedian filter cropping flipping andstretching is among the important watermarking constraints
After attacking the 6 types of ordinary images we attemptto extract our watermark and calculate the NC value Ouraim is to conclude on the degree of robustness of our schemeagainst diverse attacks Tables 2 3 4 and 5 and Figure 18 showthe experimental results relative to the NC and PSNR valuesbetween the host and extracted watermark after applyingattacks
43 Discussion of the Proposed Scheme In this section wecompare the obtained results of the suggested system withresults relative to the systems cited in the related worksection For this comparison we consider the most typicaland recent related papers [6 18ndash20] The latter representalmost the most important works addressing watermarkingsystems hardware design with interesting results First forpsychovisual quality of the original and watermarked images
our hardware implementation provides very good resultscompared to the software implementation ones
As provided in Table 6 in the absence of any type ofattack PSNR for the image ldquoLenardquo is equal to 484715which represents a better result than those aforementionedalgorithms Among the most serious attacks we apply theJPEG attack The obtained result shows that the proposedscheme is very effective against this kind of attacks In factthe results presented inTable 2 show that froma compressionrate equal to 50 the NC value is greater than 07 Comparedwith previous work (Table 6) we note that our implementedsystem gives better results
The evaluation of our implemented method againstimpulsive noises shows very promising results as presented inTable 3 In fact the recovery of the watermark is greater than07 for a density equal to 001 Beyond this value the recoveryof the watermark is not acceptable Indeed our implementedapproach has proven its robustness against this type of attackand compared to other works our implemented system givesbetter results
Also we test our system against the median filterThe testis evaluated with various sized windows (from [3times3] until[9times9]) (Table 4) The detection by correlation between the
14 Security and Communication Networks
Table 2 NC and PSNR values for watermarked and attacked images by JPEG-2000 compression
Ratio () PSNR NCCameraman Mandrill Lena Pepper Cameraman Mandrill Lena Pepper
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Salt amp pepper 005Ref [18] 231951 0918Ref [19] 2633 0847
Proposed algorithm 2823 0969
Cropping 25Ref [18] --- ---Ref [19] 2907 0847
Proposed algorithm 1809 0410
extracted and inserted watermarks has shown that our imple-mented system is robust against median-filter attacks (NCis greater than 07 for a window size coefficient less than orequal to [5x5]) and keeps the visual appearance of the imageafter watermarking As illustrated in Table 6 in general theproposed architecture gives relatively good results
The proposed architecture gives also acceptable results(NC greater than 07) against geometric attacks such asthe flapping and stretching of the watermarked images Thelast attack applied on the proposed system is the so-calledldquocroppingrdquo attack Note that for a window lower than or equalto 25 of the size of the watermarked image the NC value is
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
less than 07 Compared to previous works we note that this isalso an acceptable result for our architecture
The hardware performances of the proposed systemhave been evaluated relatively to the operating frequencythe FPGA resources occupancy rate According to Table 7broadly the proposed architecture gives better results Thehighest operating frequency reported in previous work is1838 MHz [19] However for our algorithm the maximumoperating frequency is 224MHz Compared to [20] we notedthat even if the proposed architecture is slower it presents abetter hardware resources occupation rate
5 Conclusion
In this paper a novel and efficient hardware implementationof an image watermarking system based on the Haar DiscreteWavelet Transform has been developed The performanceof the proposed hardware implementation in terms of pro-cessing latency has been evaluated and compared to otherprevious work The XSG tool has been used for systemdevelopment The utilization of this tool has a big benefit interms of conception time since the same design has beenfirstly used for the software validation and then for hardwaresystem generation A hardware cosimulation strategy usingthe XSG was applied to prove the validity of the proposedimplementation The hardware cosimulation results showedthe effectiveness of the developed architecture in terms ofvisibility and robustness against several attacks
Data Availability
Theobtained results used to support the findings of this studyare included within the article
Conflicts of Interest
The authors declare no conflicts of interest
Authorsrsquo Contributions
All authors helped in conceiving the experiments MohamedAli Hajjaji designed and performed the experiments At thesame time Mohamed Ali Hajjaji and Mohamed Gafsi wrotethemain part of the paper Abdellatif Mtibaa and AbdessalemBen Abdelali contributed to interpreting the results andrevising and writing of the paper
References
[1] M A Hajjaji E-B Bourennane A Ben Abdelali and AMtibaa ldquoCombining Haar wavelet and Karhunen Loeve trans-forms for medical images watermarkingrdquo BioMed ResearchInternational vol 2014 Article ID 313078 15 pages 2014
[2] I J Cox J Kilian F T Leighton and T Shamoon ldquoSecurespread spectrumwatermarking for multimediardquo IEEE Transac-tions on Image Processing vol 6 no 12 pp 1673ndash1687 1997
[3] Y Zhang ldquoBlind watermark algorithm based on HVS and RBFneural network in DWT domainrdquo WSEAS Transactions onComputers vol 8 no 1 pp 174ndash183 2009
[4] R O Preda andDN Vizireanu ldquoA robust digital watermarkingscheme for video copyright protection in the wavelet domainrdquoMeasurement vol 43 no 10 pp 1720ndash1726 2010
[5] W-B Lee and T-H Chen ldquoA public verifiable copy protectiontechnique for still imagesrdquoThe Journal of Systems and Softwarevol 62 no 3 pp 195ndash204 2002
[6] HKMaity andS PMaity ldquoFPGA implementation of reversiblewatermarking in digital images using reversible contrast map-pingrdquo The Journal of Systems and Software vol 96 pp 93ndash1042014
[7] S P Mohanty N Ranganathan and R K Namballa ldquoVLSIimplementation of invisible digital watermarking algorithmstowards the development of a secure JPEG encoderrdquo in Proceed-ings of the 2003 IEEE Workshop on Signal Processing SystemsSIPS 2003 pp 183ndash188 August 2003
[8] N J Mathai D Kundur and A Sheikholeslami ldquoHardwareimplementation perspectives of digital video watermarkingalgorithmsrdquo IEEE Transactions on Signal Processing vol 51 no4 pp 925ndash938 2003
[9] A Basu T S Das S K Sarkar et al ldquoFPGA prototype of visualinformation hidingrdquo in Proceedings of the 2010 Annual IEEEIndia Conference Green Energy Computing and Communica-tion INDICON 2010 pp 1ndash4 2011
[10] A Garimella M V V Satyanarayan R S Kumar P SMurugesh andU CNiranjan ldquoVLSI Impementation of OnlineDigital Watermarking Techniques With Difference Encodingfor the 8-bit Gray Scale Imagesrdquo in Proceedings of the Interna-tional Conference on VLSI Design pp 792ndash796 2003
[11] S P Mohanty N Ranganathan and R K Namballa ldquoA VLSIarchitecture for visible watermarking in a secure still digitalcamera (S 2DC) designrdquo IEEE Transactions on Very Large ScaleIntegration (VLSI) Systems vol 13 no 8 pp 1002ndash1011 2005
[12] S P Mohanty N Ranganathan and K Balakrishnan ldquoA dualvoltage-frequency VLSI chip for image watermarking in DCTdomainrdquo IEEE Transactions on Circuits and Systems II ExpressBriefs vol 53 no 5 pp 394ndash398 2006
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 17
[13] S P Mohanty O B Adamo and E Kougianos ldquoVLSI Architec-ture of an Invisible Watermarking Unit for a Biometric-BasedSecurity System in a Digital Camerardquo in Proceedings of the 2007Digest of Technical Papers International Conference onConsumerElectronics pp 1-2 Las Vegas NV USA 2007
[14] M Santi A Banerjee A Abhijit and K Malay ldquoVLSI Designof Spread Spectrum ImageWatermarkingrdquo in Proceedings of the13th National Conference on Communication NCC 2007 IITKanpur India 2007
[15] K Rajitha U R Nelakuditi V N Mandhala and T-H KimldquoFPGA implementation of watermarking scheme using XSGrdquoInternational Journal of Security and Its Applications vol 9 no1 pp 89ndash96 2015
[16] R M Khoshki ldquoHardware Based Implementation of an ImageWatermarking Systemrdquo International Journal of AdvancedResearch in Computer and Communication Engineering vol 3no 5 2014
[17] B Rahate Kunal A S Bhalchandra and S S Agrawal ldquoVLSIImplementation of Digital Image Watermarkingrdquo InternationalJournal of Engineering ResearchampTechnology (IJERT) vol 2 no6 2013
[18] SM Sakthivel andA Ravi Sankar ldquoA real timewatermarking ofgrayscale images without altering itrsquos contentrdquo in Proceedings ofthe 2015 International Conference on VLSI Systems ArchitectureTechnology and Applications VLSI-SATA 2015 IEEE January2015
[19] M R Nayak J Bag S Sarkar and S K Sarkar ldquoHardwareimplementation of a novel water marking algorithm basedon phase congruency and singular value decomposition tech-niquerdquo International Journal of Electronics and Communica-tions 2017
[20] P Karthigai kumara and K B Anumolb ldquoFPGA Implemen-tation of High Speed Low Area DWT Based Invisible ImageWatermarking Algorithmrdquo in Proceedings of the InternationalConference on Communication Technology and System DesignProcedia Engineering Elsevier 2011
[21] M A Hajjaji A Ben Abdellali N Farhani M Gafsi and AMtibaa ldquoReal time implementation of numerical watermarkingsystem using Xilinx system generatorrdquo in Proceedings of the 16thInternational Conference on Sciences and Techniques of Auto-matic Control and Computer Engineering STA 2015 pp 404ndash409 2016
[22] R Hmida A Ben Abdelali and A Mtibaa ldquoHardware imple-mentation and validation of a traffic road sign detection andidentification systemrdquo Journal of Real-Time Image Processingpp 1ndash18 2016
[23] M Gafsi S Ajili M A Hajjaji and A Mtibaa ldquoXSG for hard-ware implementation of a robust watermarking systemrdquo inProceedings of the 17th International Conference on Sciences andTechniques of Automatic Control and Computer EngineeringSTA 2016 pp 117ndash122 December 2016
[24] P M Naini Digital Watermarking Using MATLAB INTECHOpen Access Publisher pp 465-480 2011
[25] H Guan Z Zeng J Liu and S Zhang ldquoA novel robust digitalimage watermarking algorithm based on two-level DCTrdquo inProceedings of the 2014 International Conference on InformationScience Electronics and Electrical Engineering ISEEE 2014 pp1804ndash1809 April 2014
