Post on 05-Jan-2016
transcript
Implementation and Analysis of Directional Discrete Cosine
Transform in H.264 for Baseline Profile
Shreyanka Subbarayappa
Electrical Engineering Graduate Student
The University of Texas at Arlington
Advisor
Dr. K. R. Rao, EE Dept, UTA
Committee Members
Dr. W. Alan Davis, EE Dept, UTA
Dr Kambiz Alavi, EE Dept, UTA
AgendaAgendaIntroduction to the field of researchMotivation for the researchOverview of H.264Overview of DDCTImage Quality measuresH.264 JM 18.0 settingsExperimental ResultsConclusionsFuture workReferences
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 2
Introduction to the field of Introduction to the field of researchresearch
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 3
IntroductionIntroduction Importance of video Need for compression
◦ High bandwidth requirements◦ Remove inherent redundancy
Need for standardization◦ Ensures interoperability
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 4
Year
Coding
Efficiency
Network
awareness
Complexity20052005
20102010
19991999
19941994
MPEG4MPEG4
H.264H.264
19921992MPEG1MPEG1
Video Conferencing
H.26H.2633
20032003
Mobile Phone
Hand PC
Mobile TV
SVCHDTV
MPEG2MPEG2
H.265/HECH.265/HEC/ NGVC/ NGVC
VC-1
NEED FOR IMAGE OR
VIDEO COMPRESION
2011-2013
Lossless or Lossy Lossless or Lossy CompressionCompression
Lossless compression
◦ There is no information loss, and the image can be reconstructed exactly the same as the original
◦ Applications: Medical imagery, Archiving
Lossy compression
◦ Information loss is tolerable.
◦ Applications: commercial distribution (DVD) and rate constrained environment where lossless methods cannot provide enough compression ratio
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 5
Motivation for the researchMotivation for the research
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 6
Motivation for implementing DDCT in Motivation for implementing DDCT in H.264H.264
Choice of codecs◦ Prevalence of H.264
Need for DDCT in H.264◦ New concept in the
transform domain◦ Better coding gain◦ Better image quality◦ Implemented in the other
upcoming standards like H.265
◦ Larger applications for H.264 in communication fields, data storage and streaming.
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 .7
Broadcast
Streaming
Content Server
Internet
Link
Mobile
Storage
H.264
ISO media file format
Overview of H.264Overview of H.264
April 16, 2012Implementation and Analysis of Directional Discrete Cosine Transform in H.264 8
H.264: OverviewH.264: Overview
Latest block-oriented motion-compensation-based codec.
Good video quality at substantially lower bit rates.
Better rate-distortion performance and compression efficiency than MPEG-2 [42].
Simple syntax specifications, very flexible.Network friendly.Wide variety of applications such as video
broadcasting, video streaming, video conferencing, D-Cinema, HDTV.
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 9
H.264 – Encoder [1]H.264 – Encoder [1]
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 10
H.264 – Decoder [1]H.264 – Decoder [1]
April 16, 2012Implementation and Analysis of Directional Discrete
Cosine Transform in H.264 11
Profiles in H.264 [1]Profiles in H.264 [1]
April 16, 2012Implementation and Analysis of Directional Discrete
Cosine Transform in H.264 12
13 Implementation and Analysis of Directional Discrete
Cosine Transform in H.264April 16, 2012
Tools introduced in FRExts and their classification under the new high profiles
Overview of D-DCTOverview of D-DCT
April 16, 2012Implementation and Analysis of Directional Discrete
Cosine Transform in H.264 14
OutlineOutline
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 15
Problem:- To replace the DCT-like transform for intra prediction residuals in AVC and the associated zigzag scan pattern-Solution: Directional Discrete Cosine Transform (DDCT)
DDCT:- The transforms- Properties
Implementation - Transform- Quantization-Scanning pattern
Complexity- Computation - Memory
Conventional DCT [3]Conventional DCT [3]
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 16
- The 2-D discrete cosine transform (DCT) of a square or a rectangular block shape is used for almost all block-based transform schemes for image and video coding.
- Implemented separately through two 1-D transforms, one along the vertical direction and another along the horizontal direction.- The conventional DCT seems to be the best choice for image blocks in which vertical and/or horizontal edges are dominating.
+
(M 1D-DCT s of Length N)
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 17
Forward 2D DCT (NXM) [3]
Inverse 2D DCT (NXM) [3]
x(n,m) = Samples in the 2D data domain.
XC2 (k, l) = Coefficients in the 2D-DCT domain
Limitations of conventional DCT
• It is not very efficient when the conventional DCT is applied to an image block in which other directional edges dominate.• When the first 1-D DCT (vertical or horizontal) is applied, the nonzero coefficients are not well aligned across different columns (or rows). Consequently, the second 1-D DCT may produce more nonzero coefficients
Modes Of DDCT [4]Modes Of DDCT [4]
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 18
Mode 0 – VerticalMode 1 – HorizontalMode 2 – DC (Ignored for DDCT modes)Mode 3 – Diagonal down leftMode 4 – Diagonal down right
Mode 5 – Vertical rightMode 6 – Horizontal downMode 7 – Vertical leftMode 8 – Horizontal up
Mode 3 DDCT- Diagonal Down LeftMode 3 DDCT- Diagonal Down Left
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 19
Step 1: (X00, X01,…… ,X32, X33)- Pixels in the 2-D spatial domain.
Step 2: 1D- DCT is performed for the 4X4block in diagonal down-left position with lengths L=1, 2, 3, 4, 3, 2, 1.
(A,B,C,……O,P)- coefficients in the DCT domain.
Step 3: The coefficients of step2 after 1D DCT are arranged vertically as shown in the figure.Apply Horizontal 1D- DCT for lengths L=7, 5, 3 and 1 and arranged in the same pattern
STEP 1 STEP 2 STEP 3
PIXELS Coefficients
Coefficients
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 20
STEP 4 STEP 5
Step 4: Apply Horizontal 1D- DCT for lengths L=7, 5, 3 and 1, the coefficients are arranged in the same pattern as shown in the figure step 4.
Step 5: After Step 4, move all 2D (4X4) Directional DCT coefficients to the left.Implement quantization followed by 2D VLC for compression/coding along zig-zag scan.This scanning helps to increase the runlength of zero (transform) coefficients leading to
reduced bit rate in the 2D-VLC coding (similar to JPEG [12]).
CoefficientsCoefficients
Mode 4 DDCT- Diagonal Down Mode 4 DDCT- Diagonal Down RightRight
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 21
STEP 1 STEP 2 STEP 3
Step 1: X00, X01, ….., X33 are the pixels in the 2D spatial domain.
Step 2: 1D DCT is performed for the 4X4 block in diagonal down-right position with lengths L= 1, 2, 3, 4, 3, 2 and 1.
Step 3: The coefficients of step 2 after 1 D DCT are arranged vertically in the same pattern as shown in step 3. Then apply horizontal 1 D DCT for lengths L = 7, 5, 3 and 1 and arrange in the same pattern.
CoefficientsCoefficientsPIXELS
Implementation and Analysis of Directional Discrete Cosine Transform in H.264 22
STEP 4 STEP 5
Step 4: Apply horizontal 1 D DCT for lengths L= 7, 5, 3 and 1. The coefficients are arranged the same pattern as shown in step4.
Step 5: After step 4, move all 2D (4X4) directional DCT coefficients to the left. Implement quantization followed by 2D VLC for compression/coding zigzag scan.
April 16, 2012
Coefficients
Mode 5 DDCT- Vertical RightMode 5 DDCT- Vertical Right
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 23
STEP 1 STEP 2 STEP 3
Step 1: X00, X01, ….., X33 are the pixels in the 2D spatial domain.
Step 2: 1D DCT is performed for the 4X4 block in vertical-right position with lengths L= 2,4,4,4,2.
Step 3: The coefficients of step 2 after 1 D DCT are arranged vertically in the same pattern as shown in step 3. Then apply horizontal 1 D DCT for lengths L = 5, 5, 3 and 3and arrange in the same pattern.
CoefficientsCoefficients
PIXELS
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 24
STEP 4 STEP 5
Step 4: Apply horizontal 1 D DCT for lengths L= 5, 5, 3 and 3. The coefficients are arranged the same pattern as shown in step 4.
Step 5: After step 4, move all 2D (4X4) Directional DCT coefficients to the left. Implement quantization followed by 2D VLC for compression/coding zigzag scan as shown in step 5. This scanning helps to increase the run-length of zero (transform) coefficients leading to reduce bit rate in 2D-VLC coding (similar to JPEG).
Coefficients
Mode 6 DDCT- Horizontal downMode 6 DDCT- Horizontal down
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 25
STEP 1 STEP 2 STEP 3
Step 1: X00, X01, ….., X33 are the pixels in the 2D spatial domain.
Step 2: 1D DCT is performed for the 4X4 block in Horizontal down position with lengths L= 2, 4, 4, 4 and 2.
Step 3: The coefficients of step 2 after 1 D DCT are arranged vertically in the same pattern as shown in step 3. Then apply horizontal 1 D DCT for lengths L = 5, 5, 3 and 3 and arrange in the
same pattern.
Coefficients CoefficientsPIXELS
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 26
STEP 4 STEP 5
Step 4: Apply horizontal 1 D DCT for lengths L= 5, 5, 3 and 3. The coefficients are arranged the same pattern as shown in step 4.
Step 5: After step 4, move all 2D (4X4) directional DCT coefficients to the left. Implement quantization followed by 2D VLC for compression/coding zigzag scan as shown in step 5. This
scanning helps to increase the run-length of zero (transform) coefficients leading to reduce bit rate in 2D-VLC coding (similar to JPEG).
Coefficients
Mode 7 DDCT- Vertical LeftMode 7 DDCT- Vertical Left
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 27
STEP 1 STEP 2 STEP 3
Step 1 X00, X01, ….., X33 are the pixels in the 2D spatial domain.
Step 2: 1D DCT is performed for the 4X4 block in Vertical left position with lengths L= 2,4,4,4 and 2 as shown in step 3.
Step 3: The coefficients of step 2 after 1 D DCT are arranged vertically in the same pattern as shown in step 3. Then apply horizontal 1 D DCT for lengths L = 5, 5, 3 and 3 and arrange in the
same pattern.
Coefficients
CoefficientsPIXELS
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 28
STEP 4 STEP 5
Step 4: Apply horizontal 1 D DCT for lengths L= 5, 5, 3 and 3. The coefficients are arranged the same pattern.
Step 5: After step 4, move all 2D (4X4) Directional DCT coefficients to the left. Implement quantization followed by 2D VLC for compression/coding zigzag scan as shown in step 3.
Coefficients
Mode 8 DDCT- Horizontal UpMode 8 DDCT- Horizontal Up
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 29
STEP 1 STEP 2 STEP 3
Step 1: X00, X01, ….., X33 are the pixels in the 2D spatial domain.
Step 2: 1D DCT is performed for the 4X4 block in Horizontal Up position with lengths L= 2, 4, 4, 4 and 2 as shown step 2.
Step 3: The coefficients of step 2 after 1 D DCT are arranged vertically in the same pattern as shown in step 3. Then apply horizontal 1 D DCT for lengths L = 5, 5, 3 and 3 and arrange in the same pattern.
Coefficients CoefficientsPIXELS
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 30
STEP 4 STEP 5
Step 4: Apply horizontal 1 D DCT for lengths L= 5, 5, 3 and 3. The coefficients are arranged the same pattern as shown in step 4.
Step 5: After step 4, move all 2D (4X4) directional DCT coefficients to the left. Implement quantization followed by 2D VLC for compression/coding zigzag scan as shown in step 5. This scanning helps to increase the run-length of zero (transform)
coefficients leading to reduce bit rate in 2D-VLC coding (similar to JPEG [4]).
Coefficients
Obtaining Mode 3 from Mode 4Obtaining Mode 3 from Mode 4
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 31
Rotate pixels by –pi/2 ( counterclock wise by 90°) to get Mode 4
Pixels
Pixels
Pixels
STEP BY STEP MODE CHANGESTEP BY STEP MODE CHANGE
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 32
MODE 3
-π/2=
MODE 4Pixels Pixels Coefficients
Step 1: Rotate the pixels by –π/2
Step 2: Perform 1-D DCT with Length = 1, 2, 3, 4, 3, 2, 1
Step 3: We get the coefficients of mode 3 (Diagonal Down Left) from mode 4 (Diagonal Down Right) as shown in figure 3.
Obtaining Mode 6 from Mode 5Obtaining Mode 6 from Mode 5
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 33
Rotate pixels by reflecting across the diagonal axis to get Mode 6
Pixels
Pixels
Pixels
Obtaining Mode 7 from Mode 5Obtaining Mode 7 from Mode 5
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 34
Rotate pixels by reflecting across the horizontal axis to get Mode 7
Pixels
Pixels
Pixels
Obtaining Mode 8 from Mode 5Obtaining Mode 8 from Mode 5
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 35
Rotate pixels by pi/2 (clockwise by 90°) to get Mode 8
Pixels
Pixels
Pixels
Eigen or Basis ImagesEigen or Basis Images
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 36
Mapping of a 2D data array into a 2D DCT domain implies decomposing the 2D data array into the basis images of the DCT.
Computation of basis images for diagonal down left (a) The original 4X4 block with diagonal down left computation (b) The 1 D DCT of coefficients for lengths 7, 5, 3 and 1 for basis image (0, 0) (c) The 1 D DCT of coefficients for lengths 7, 5, 3 and 1 for basis image (0,1) (d) The 1 D DCT of coefficients for lengths 7, 5, 3 and 1 for basis image (3,3)
Mode 3 Diagonal Down Left Eigen Image (1,1) matrix Mode 3 Diagonal Down Left Eigen Image (1,1) matrix Computation for 4X4 blockComputation for 4X4 block
Step 1: Horizontal 1D-DCT for length =7, 5, 3, 1
Step 2: Coefficients of 1D-DCT for length =7, 5, 3, 1
Step 3: Put back the coefficients in the block form
Step 4: 1 D DCT of diagonal down left with lengths = 1, 2, 3, 4, 3, 2 and 1
Step 5: Putting back the coefficients of step 4 1 D-DCT we get (1,1) basis image for 4X4 block
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 38
MODE 3 - Diangonal down left basis images for 4X4 block of an image
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 39
MODE 3 - Diangonal down left basis images for 8X8 block of an image
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 40
MODE 0 or 1 – Vertical or Horizontal basis images for 8X8 block of an image
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 41
MODE 5 – Vertical right basis images for 8X8 block of an image
Computation of DDCT for an imageComputation of DDCT for an image
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 42
Image Quality MeasuresImage Quality Measures
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 43
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 44
•Criteria to evaluate compression quality
•Two types of quality measures Objective quality measure- PSNR, MSE Structural quality measure- SSIM [29]
• SSIM emphasizes that the human visual system is highly adapted to extract structural information from visual scenes. Therefore, structural similarity measurement should provide a good approximation to perceptual image quality.
•MSE and PSNR for a NxM pixel image are defined as
M
m
N
n
nmynmxNM
MSE1 1
2,,*
1
MSE
LPSNR
2
10log10
where x is the original image and y is the reconstructed image. M and N are the width and height of an image and ‘L’ is the maximum pixel value in the NxM pixel image.
Image Quality Measures
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 45
• The SSIM index is defined as a product of luminance (l), contrast (c) and structural (s) comparison functions.
where , α>0, β>0 and γ >0 are parameters used to adjust the relative importance of the three components
where μ is the mean intensity, and σ is the standard deviation as a round estimate of the signal contrast. C1 and C2 are constants.
H.264 JM18.0 [24] settingsH.264 JM18.0 [24] settings
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 46
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 47
4.4.2 Encoder Configuration in JM 18.0
• FramesToBeEncoded = 1 #Number of Frames to be coded• ProfileIDC = 66 # Profile IDC (66 = baseline, 77 = main, 88 = extended; FREXT Profiles: 100 = High, 110= High 10, 122= High 4:2:2, 244 = High 4:4:4, 44= CAVLC 4:4:4 Intra, 118 = Multiview High Profile,128 = Stereo High Profile)• IntraProfile = 1 # Activate Intra Profile for FRExt (0: false, 1: true) #(e.g. ProfileIDC = 110, IntraProfile = 1 => High 10 Intra Profile)• Transform8X8Mode = 0 # (0: only 4X4 transform, 1: allow using 8X8 transform additionally, 2: only 8X8 transform•Transform 16X16Mode=0 #(0: no 16X16 mode, 1: allow 16X16 mode)• Input YUV file: foreman_qcif.yuv• Output H.264 bitstream: test.264• Output YUV file: test_rec.yuv• YUV format: YUV 4:2:0• Frames to be encoded: 1• Frequency used for encoded bitstream: 30.00 fps• DistortionSSIM = 1 # Compute SSIM distortion (0: disable/default, 1: enabled)
Experimental Results and Experimental Results and GraphsGraphs
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 48
QCIF and CIFQCIF and CIF
49 Implementation and Analysis of Directional
Discrete Cosine Transform in H.264April 16, 2012
• CIF (Common Intermediate Format), is a format used to standardize the horizontal and vertical resolutions in pixels for sequences in video signals, commonly used in video teleconferencing systems.
•The CIF "image sizes" were specifically chosen to be multiples of macroblocks (i.e. 16 × 16 pixels) due to the way that discrete cosine transform based video compression/decompression is handled. So, by example, a CIF-size image (352 × 288) corresponds to 22 × 18 macroblocks
•QCIF means "Quarter CIF". To have one fourth of the area as "quarter" implies the height and width of the frame are halved. QCIF-size image is 176 x 144.
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 50
Bit Rate (kbps) QP (I frame)
PSNR in dB
MSESSIM
5590.56 0 79.159 0.00079 1
5232.48 4 68.825 0.00852 1
3891.84 8 57.204 0.12378 0.9995
2168.16 16 48.86 0.84553 0.9965
1088.88 24 41.803 4.29364 0.9846
745.68 28 38.892 8.39157 0.9735
331.92 36 33.173 31.32035 0.9342
152.64 44 27.981 103.5018 0.8388
72 51 23.335 301.693 0.6703
Image metrics for Foreman QCIF sequence in integer DCT implementation in H.264
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 51
Bit Rate(kbps)PSNR in
dBMSE SSIM
5486.9690.357 0.00006 1
5201.9682.436 0.00068 1
3882.5369.689 0.0014 1
2264.6360.147 0.00842 0.9996
1153.8452.976 0.55385 0.9972
686.5441.876 2.43788 0.9925
302.5338.653 10.2537 0.9801
142.5334.642 50.4376 0.9208
6730.764 110.268 0.8674
Image metrics for Foreman QCIF sequence in DDCT implementation in H.264
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 52
QP (I frame) Encoding Time of Int-DCT (sec)Encoding Time of DDCT (sec)
0 10.87618.96
4 10.03218.096
8 9.18317.264
16 7.29215.367
24 5.66612.5437
28 4.96811.0642
36 4.06710.853
44 3.4849.638
51 3.0738.428
Encoding Time of I frame for Foreman QCIF sequence in DDCT and Int-DCT implementation in H.264
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 53
PSNR v/s bit rate for DDCT and integer DCT for foreman QCIF sequence
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 54
MSE v/s bit rate for DDCT and integer DCT for foreman QCIF sequence
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 55
SSIM v/s bit rate for DDCT and integer DCT for foreman QCIF sequence
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 56
Encoding time v/s quantization parameter for DDCT and integer DCT for Foreman QCIF sequence
Test sequence used for simulationTest sequence used for simulation
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 57
Bit Rate:72kbits/frame Bit Rate:67kbits /frame
Properties of DDCTProperties of DDCT
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 58
• Adaptivity: Unlike AVC in which the same DCT-like transform is applied to the intra prediction errors for all intra prediction modes of the same block size (4x4, 8x8, or 16x16), DDCT assigns a different transform and scanning pattern to each intra prediction mode. These transforms and scanning patterns are designed taking into account the intra prediction direction.
• Directionality: Since the intra prediction mode is known, the DDCT is designed with the knowledge of the intra prediction direction. By first applying the transform along the prediction direction, DDCT has the potential to minimize the artifacts around the object boundaries.
• Symmetry: Although there are 22 DDCTs for 22 intra prediction modes (9 modes for 4x4, 9 modes for 8x8, and 4 modes for 16x16), these transforms can be derived, using simple operators such as rotation and/ or reflection, from only 7 different core modes.
ConclusionsConclusions
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 59
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 60
Directional DCT has a better coding gain when compared to integer DCT
• PSNR value for DDCT is more when compared to Integer DCT.
• MSE value of DDCT is less compared to integer DCT for the same bit rates.
• SSIM graph shows that the value obtained for different bit rates is almost 1 for DDCT when compared to Integer DCT.
• Foreman frame of QCIF format gives a better quality image obtained from DDCT with respect to the output obtained from Integer DCT.
Drawback of DDCT
• Encoding time for DDCT is more when compared to integer DCT.
Future WorkFuture Work
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 61
Future workFuture work
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 62
• DDCT can be extended to other video standards that use integer DCT in the transform domain.
• It can be extended for the entire video – inter frame prediction.
•It can be extended to other profiles in H.264 like main and extended profiles.
•Only 8 modes are described in this research. These can be extended to other directional modes. Payoff between increasing complexity and improved visual quality can be investigated.
ReferencesReferences
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 63
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 64
1. I. E.G. Richardson, “H.264 and MPEG-4 video compression: video coding for next-generation multimedia”, Wiley, 2003.
2. N. Ahmed, T. Natarajan, and K. R. Rao, “Discrete cosine transform,” IEEE Trans. Comput., vol. C-23, pp. 90-93, Jan. 1974.
3. K. R. Rao and P. Yip, “Discrete cosine transform: Algorithms, advantages, applications,” Boca Raton FL: Academic Press, 1990.
4. B. Zeng and J. Fu, “Directional discrete cosine transforms - A new framework for image coding”, IEEE Trans. on Circuits and Systems for Video Technology, vol. 18, no. 3, pp. 305-313, Mar. 2008.
5. B. Zeng and J. Fu, “A compensation techniques in directional DCT’s”, IEEE International Symposium on Circuits and Systems, pp. 521- 524, June, 2007.
6. E. Lallana and M. Uy, “The Information Age,” UNDP-APDIP, 2003.7. Open source article, “Digital Revolution,” Wikipedia Foundation,
http://en.wikipedia.org/wiki/Digital_Revolution8. K. Sayood, "Introduction to data compression,” 3rd Edition, Morgan Kaufmann Publisher Inc., 2006. 9. R. Schafer and T. Sikora, "Digital video coding standards and their role in video communications,"
Proceedings of the IEEE, Vol. 83, pp. 907-923, Jan. 1995.10. Information technology-generic coding of moving pictures and associated audio information: ISO/IEC
13818-2 (MPEG-2) Std.11. Advanced video coding for generic audiovisual services, ITU-T Rec. H.264 / ISO / IEC 14496-10, Nov.
2009.12. I. Ahmad et al, “Video transcoding: An overview of various techniques and research issues”, IEEE
Trans. on Multimedia, vol. 7, pp. 793-804, Oct. 2005.13. Open source article, “H.264/MPEG-4 AVC,” Wikipedia Foundation,
http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 65
14. S. Kwon, A. Tamhankar and K.R. Rao, ”Overview of H.264 / MPEG-4 Part 10”, J. Visual Communication and Image Representation, vol. 17, pp.186-216, April 2006.
15. T. Wiegand and G. J. Sullivan, “The H.264 video coding standard”, IEEE Signal Processing Magazine, vol. 24, pp. 148-153, March 2007.
16. A. Puri et al, “Video coding using the H.264/ MPEG-4 AVC compression standard”, Signal Processing: Image Communication, vol. 19, pp: 793 – 849, Oct. 2004.
17. G. Sullivan, P. Topiwala and A. Luthra, “The H.264/AVC advanced video coding standard: Overview and introduction to the fidelity range extensions”, SPIE conference on Applications of Digital Image Processing XXVII, vol. 5558, pp. 53-74, Aug. 2004.
18. K. R. Rao and P. C. Yip, “The transform and data compression handbook”, Boca Raton,FL: CRC press, 2001.
19. T. Wiegand et al, “Overview of the H.264/AVC video coding standard”, IEEE Trans. on Circuits and Systems for Video Technology, vol. 13, pp. 560-576, Jul. 2003.
20. T. Wiegand and G. J. Sullivan “The picturephone is here: Really” IEEE spectrum, vol.48, pp.50-54, Sept.2011.
21. I. Richardson, “The H.264 advanced video compression standard”, Wiley, 2nd edition, 2010.22. Intra prediction modes in H.264. Website: http://www.vcodex.com/files/h264_intrapred.pdf23. F. Kamisli and J. S. Lim, “Video compression with 1-d directional transforms in H.264/AVC”, IEEE
ICASSP, pp. 738-741, Mar. 2010.24. H.264/AVC reference software. Website: http://iphome.hhi.de/suehring/tml/download 25. Intra coding with directional DCT and directional DWT, Document: JCTVC-B107_r126. Directional Discrete Transform JCTV Website:
http://wftp3.itu.int/av-arch/jctvc-site/2010_07_B_Geneva/JCTVC-B107.zip27. B.Chen, H.Wang and L.Cheng, “Fast directional discrete cosine transform for image
compression”, Opt. Eng. vol. 49, issue 2, article 020101, Feb. 2010.
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 66
28. C. Deng et al, “Performance analysis, parameter selection and extensions to H.264/AVC FRExt for high resolution video coding”, J. Vis. Commun. Image R., vol. 22 (In Press), Available on line, Feb. 2011.
29. Z.Wang et al, “Image quality assessment: From error visibility to structural similarity”, IEEE Trans. on Image Processing, vol. 13, no. 4, pp. 600-612, Apr. 2004.
30. W.Zhao, J.Fan and A.Davari, “H.264-based wireless surveillance sensors in application to target identification and tracking”, i-manager’s Journal on Software Engineering, vol.4, no. 2, Oct. 2009.
31. Website: http://web.eng.fiu.edu/fanj/pdf/J5_i-manager09h264_camera.pdf32. W.Zhao et al, “H.264-based architecture of digital surveillance network in application to computer
visualization”, i-manager’s Journal on Software Engineering, vol.4, no. 4, Apr. 2010.33. Directional Discrete Cosine Transform theory Website: http://web.eng.fiu.edu/fanj/pdf/J8_i-
mgr10architecture_camera.pdf34. D. Marpe, T. Wiegand and G. J. Sullivan, “The H.264/MPEG-4 AVC standard and its applications”, IEEE
Communications Magazine, vol. 44, pp. 134-143, Aug. 2006.35. Website: http://iphome.hhi.de/wiegand/assets/pdfs/h264-AVC-Standard.pdf36. F. Kamisli and J. S. Lim, “Transforms for motion compensation residual”, IEEE ICASSP, pp.789-792,
Apr. 2009.37. Z.Wang, E.P.Simoncelli and A.C.Bovik, “Multi-scale structural similarity for image quality assessment”,
Thirty-Seventh Asilomar Conference on Signals, Systems and Computers, vol. 2, Nov. 2003.38. C.L.Chang and B.Girod, “Direction-adaptive partitioned block transform for image coding”, 15 th IEEE
International Conference on Image Processing, pp. 145-148, Oct. 2008.39. H.Xu, J.Xu and F.Xu, “Lifting-based directional DCT-like transform for image coding”, IEEE Trans. on
Circuits and Systems for Video Technology, vol. 17, issue 10, pp. 1325-1335, Oct. 2007.40. J.Xu, B.Zeng and F.Wu, “An overview of directional transforms in image coding”, Proceedings of 2010
IEEE International Symposium on Circuits and Systems, pp. 3036-3039, Aug. 2010.
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 67
41. MPEG-1 basics Website: http://en.wikipedia.org/wiki/Mpeg-142. MPEG-2 basics Website: http://en.wikipedia.org/wiki/Mpeg-243. MPEG-4 basics Website: http://en.wikipedia.org/wiki/Mpeg-444. H.261 basics Website: http://en.wikipedia.org/wiki/H.26145. H.262 basics Website: http://en.wikipedia.org/wiki/H.26246. H.263 basics Website: http://en.wikipedia.org/wiki/H.26347. DFT basics Website: http://en.wikipedia.org/wiki/Discrete_Fourier_transform48. K. R. Rao and J. J. Hwang, “Techniques and standards for image/video/audio coding”,
Prentice Hall, 1996.
QUESTIONS ?QUESTIONS ?
April 16, 2012Implementation and Analysis of Directional
Discrete Cosine Transform in H.264 68