+ All Categories
Home > Documents > Security System for Bank using Biometric · PDF file · 2014-03-28Security System...

Security System for Bank using Biometric · PDF file · 2014-03-28Security System...

Date post: 14-Mar-2018
Category:
Upload: truongtuong
View: 224 times
Download: 2 times
Share this document with a friend
5
Security System for Bank using Biometric Recognition Prof. Ms.N.K.Bhandari Department of electronic engineering. P.R.E.C, LONI. Miss. Thorat Rutuja R. Miss.Nikale Dipali B. Miss.Jawale Sayali A. Department of electronic engg. Department of electronic engg. Department of electronic engg. P.R.E.C,LONI. P.R.E.C,LONI. P.R.E.C,LONI. Abstract:- Biometric recognition is based on some specific physiological and behavioural characteristics of human for an automatic identification of a person. Currently biometrics is used in ATMs, cell phones, bank, laptops, credit cards and social services. We have used multibiometricsystem for better recognition of person. This paper includes parameters like ear, voice andsignature. PCA (Principal component analysis) algorithm for ear, MFCC (Melfrequency cepstral coefficient) for voice and Shape matching for signature verification. We are using PCA algorithm because it gives feature compression, MFCC is used to give high level algorithm. A characteristic of signature depends on its shape and curve’s. So shape matching is suitable for signature recognition. Indexterm:-Biometrics, identification, multi biometrics, recognition, verification. I. Introduction:- In today‟s life security is of prime importance in every field like Banking, ATM, medical field, military purpose, Networking and in forensic etc. Traditional method of security are ID card (Token based security), password (knowledge base security), But these methods are not reliable and they can be easily hacked.We are developing a security system using biometric parameters, which are considered to haveunique characteristic. Biometric Verification is based on physical and behavioural characteristics of human so this method is more secure and reliable. Biometric parameters like ear, voice, signature, face, fingerprint, retina, DNA, iris, gait, palm-print are used to establish a person‟s identification. Biometric recognition system operates by acquiring data from human, extracting features and comparing them with the sample data. Biometric recognition operates on two mode, Identification and verification. In Identification mode unknown parameter is determined and in verification mode identity of that parameter is either accepted or rejected. Biometric recognition offers much higher accuracy than the traditional ones. Characteristics of biometric parameter:- 1. Permanent:-It should not change frequently and remain constant for a long interval of a time. 2. Unique:-God has gifted us some traits which are unique like ear, retina, faceand fingerprint. They distinguish one person from other. 3. Measurement:-Biometric properties should be suitable for measurement in short duration of time. It takes the information without making any harm to person. 4. Performance:- Performance of biometric recognition is highlyaccurate. Speed and quality of biometric recognition is more.Hence performance increases [4]. Table1:-The following table shows comparison between varies biometric parameter. Where L=low M=medium H=high [4]. Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545 IJCTA | March-April 2014 Available [email protected] 541 ISSN:2229-6093
Transcript
Page 1: Security System for Bank using Biometric · PDF file · 2014-03-28Security System for Bank using Biometric Recognition. Prof. Ms.N.K.Bhandari . ... ATM, medical field, ... Radi.Z”

Security System for Bank using Biometric Recognition

Prof. Ms.N.K.Bhandari

Department of electronic engineering.

P.R.E.C, LONI.

Miss. Thorat Rutuja R. Miss.Nikale Dipali B. Miss.Jawale Sayali A.

Department of electronic engg. Department of electronic engg. Department of electronic engg.

P.R.E.C,LONI. P.R.E.C,LONI. P.R.E.C,LONI.

Abstract:-

Biometric recognition is based on some specific

physiological and behavioural characteristics of

human for an automatic identification of a person.

Currently biometrics is used in ATMs, cell phones,

bank, laptops, credit cards and social services. We

have used multibiometricsystem for better

recognition of person. This paper includes

parameters like ear, voice andsignature. PCA

(Principal component analysis) algorithm for ear,

MFCC (Melfrequency cepstral coefficient) for voice

and Shape matching for signature verification. We

are using PCA algorithm because it gives feature

compression, MFCC is used to give high level

algorithm. A characteristic of signature depends on

its shape and curve’s. So shape matching is suitable

for signature recognition.

Indexterm:-Biometrics, identification, multi

biometrics, recognition, verification.

I. Introduction:-

In today‟s life security is of prime importance in

every field like Banking, ATM, medical field,

military purpose, Networking and in forensic etc.

Traditional method of security are ID card (Token

based security), password (knowledge base

security), But these methods are not reliable and

they can be easily hacked.We are developing a

security system using biometric parameters, which

are considered to haveunique characteristic.

Biometric Verification is based on physical and

behavioural characteristics of human so this

method is more secure and reliable. Biometric

parameters like ear, voice, signature, face,

fingerprint, retina, DNA, iris, gait, palm-print are

used to establish a person‟s identification.

Biometric recognition system operates by acquiring

data from human, extracting features and

comparing them with the sample data. Biometric

recognition operates on two mode, Identification

and verification. In Identification mode unknown

parameter is determined and in verification mode

identity of that parameter is either accepted or

rejected. Biometric recognition offers much higher

accuracy than the traditional ones.

Characteristics of biometric parameter:-

1. Permanent:-It should not change frequently

and remain constant for a long interval of a

time.

2. Unique:-God has gifted us some traits which

are unique like ear, retina, faceand fingerprint.

They distinguish one person from other.

3. Measurement:-Biometric properties should be

suitable for measurement in short duration of

time. It takes the information without making

any harm to person.

4. Performance:- Performance of biometric

recognition is highlyaccurate. Speed and

quality of biometric recognition is more.Hence

performance increases [4].

Table1:-The following table shows comparison

between varies biometric parameter. Where L=low

M=medium H=high [4].

Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545

IJCTA | March-April 2014 Available [email protected]

541

ISSN:2229-6093

Page 2: Security System for Bank using Biometric · PDF file · 2014-03-28Security System for Bank using Biometric Recognition. Prof. Ms.N.K.Bhandari . ... ATM, medical field, ... Radi.Z”

Biometric identifier

un

iver

sali

ty

Dis

tin

ctiv

enes

s

Per

man

ence

Co

llec

tab

ilit

y

Per

form

ance

Acc

epta

bil

ity

Cir

cum

ven

tio

n

DNA H H H L H L L

Ear M M H M M H M

Face H L M H L H H

Fingerprint M H H M H M M

Gait M L L H L H M

Iris H H H M H L L

Palmprint M H H M H M M

Signature L L L H L H H

Voice M L L M L H H

II. How to system work:-

Fig1:- Block dig of a system

III. Ear:-

Medical literature states that ear growth is

proportional for first four months of birth and

minor changes occur during the age group from 8

to 70 years .So it is more suitable for long period

verification. Feature of ear are unique. They are

not affected by the factors such as mood, health

and environment. The pinna of ear is unaffected by

ageing. Ear recognition is good for security access

control, surveillance and crime investigation.

Ear should be detected from a person‟s side face in

order to extract an image contain only the ear.

There are 3 steps of ear detection. Skin tone

detection is used to detect a person‟s side face

containing the ear, short and isolated edge are

removed in Extraction steps and in segmentation

ear is isolated from other skin region.

PCA Algorithm for Ear Recognition:-

In PCA (principal component analysis) algorithm

raw image is taken by camera. In thepre-processing

the ear images are cropped in required size

(400x500). Geometric normalization and masking

is done in normalization step. Allnon-ear part, for

example background, hairs are masked in masking

process.

Eigenvalues and Eigenvectors are extracted during

training phase. The eigenvector are chosen based

on the top eigenvalues. After that there is a training

set which is a set of clean images without any

duplicates. In testing the algorithm gives a set of

known ears (sample data) and set of unknown ears

set and matches each set to its possible identity in

the gallery. Ultimately the result tells us whether

the image has matched with data set or not.

IV. Voice:-

Property of the voice is dependent on nasal tone,

cadence, inflection, lips, mouth etc. Also human

voice depends on mood, expression and weakness.

There are two types of human voice, unvoice and

voice. Unvoice means when we pronounce the

words like„s‟ and „f‟ then the vocal cord get reared.

Similarly when we pronounce like „a‟ and „e‟ then

vocal cord vibrates and frequency get generate [5].

MFCC for Voice Recognition:-

Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545

IJCTA | March-April 2014 Available [email protected]

542

ISSN:2229-6093

Page 3: Security System for Bank using Biometric · PDF file · 2014-03-28Security System for Bank using Biometric Recognition. Prof. Ms.N.K.Bhandari . ... ATM, medical field, ... Radi.Z”

Voice signal recognition consists of the process of

converting a speech signal into features that are

important for verification process. There are so

many techniques and algorithms for voice

recognition. TLP, LPC and RASTA are various

algorithms for voice recognition.

Step 1:- In this process signal passing through a

filter. Filter emphasizes signal in to higher

frequency and the energy of signal is increased in

this process.

Y[n]=x[n]-ax[n-1]

Y[n]=output signal

X[n]=Input signal

a=95% presumed of any one sample

Step 2:- In Framing step voice sample obtained

from ADC (Analog to digital converter) is

compared into a form of small frame. This has

period 20 to 40 msec. This signal is divided into N

samples M (M<N) is separated by adjacent frames.

We using eight bit data hence N=256 and M=100

Step 3:- Hamming window gives shape and it is

considered in next block of feature extraction. It

integrates all the closest frequency lines.

Step 4:- In fast Fourier transform (FFT) the Fourier

transform is used to convert time domain into

frequency domain from every frame of N samples.

Y[w]=FFT[h(t)*x(t)]

Where h(t)=vocal tract impulse response

Step 5:- In filter bank processing range of

frequency is varied in FFT spectrum voice signal.

The liner scale does not follow by voice signal.

Fig:-2 Mel scale filter bank.

Above figure shows set of triangular filter. The

magnitude of each filter is in triangular shape and

equal to unity at centre frequency.It decreases

linearly to zero at centre of two adjacent filters [8].

F(mel)=[2595*log 10 (1+f)700]

Step 6:- Discrete cosine transform (DCT) converts

the log mel spectrum into timed domain. This result

is givesMel frequency centrum coefficient.

Step 7:- The frames and voice signal changes.

That‟s why it is a need to add features related to

change in cepstral feature over time [8].

V. Signature:-

The signatureof a human being is an important

biometric parameter. This can be used for

verification purpose. Signature verification divided

in to two types, online signature verification and

off-line signature verification. Online

signatureverification is real time. This is based on

movement of pen-tip, pressure, velocity, and pen

up and pen down. Offline signature verification is

based on image processing. Firstly signature is

captured by a camera then feature of this signature

is match with the samplewhich is already stored in

sample data [1].

Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545

IJCTA | March-April 2014 Available [email protected]

543

ISSN:2229-6093

Page 4: Security System for Bank using Biometric · PDF file · 2014-03-28Security System for Bank using Biometric Recognition. Prof. Ms.N.K.Bhandari . ... ATM, medical field, ... Radi.Z”

Step in signature verification:-

Pre-processing:- In this stage cropping and noise

removing is done.In noise removal unwanted

information such as small dots are removed.In

croppingfirst boundaries of signature is determined.

Then it eliminates unnecessary area around it.

Registration:-Second stage in signature verification

is registration. In this stage scaling, shifting and

rotation taking place. In scaling operation the

signature is re-scale into proper form to more

accurate result. In shifting operation centric of

signature isdetermined. Rotation operation rotates

signature in correct direction.

Feature Extraction:-In feature extraction method

number of loops, curve, junction and width to

length ratio in signature is extracted. Using this

parameter two Signatures are combined which is

sample signature and other one of is the

signaturewe want to verify. The methodfinds

common pixels between these two signatures.

Verification:-This is last stage in signature

verification. A tested Signature is verified against

the sample Signature which already stored in

sample data. The differences between these two

signatures decide variation percentage. If signature

verification is above 85% then it is verified

otherwise not verified [1].

VI. Experimental Result:-

As the person enters the bank to access his locker

he has passed through these security checks ear,

voice and signature. Firstly sample of his ears are

taken by the camera and image is being compared

with the existing data. If the sample matches with

the existing database then it further checked with

voice and signature. Sample of voiceand signature

are taken with the help of mice and camera

respectively. If the result does not match the system

willstop working. If all these three parameters are

matched then message will display on LCD

“match” and door will be open. Similarly all these

three parameters are not matched then display

message on LCD “not match” and door will be not

open.

Fig 3. Input image of Ear

Fig 4:- Output Image of Ear

Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545

IJCTA | March-April 2014 Available [email protected]

544

ISSN:2229-6093

Page 5: Security System for Bank using Biometric · PDF file · 2014-03-28Security System for Bank using Biometric Recognition. Prof. Ms.N.K.Bhandari . ... ATM, medical field, ... Radi.Z”

Fig 5:- Original Voice Signal and its Spectrogram

Fig 6:- Power spectral Density

Fig 7:-Output Signature

VI.Conclusion:-

This paper presents method for identification and

verification by using biometric parameter such as

ear, voice and signature. Biometric is automatic

verification of a person which totally based on

physiological and behavioural characteristics of

human. Various parameters are extracted and

measured through PCA, MFCC and shape

matching algorithms. Biometric as a reliable means

of authentication is gaining momentum. Unimodal

biometric systems have a variety of problems and

presently application employing unimodal

biometricsystem is limited. The future of biometric

can thus be imagined to belong to multimodal

biometric system.

VII. Reference:-

1) Hazem Hiary, Raja Alomari, Thaeer Kobbaey,

Radi.Z” Offline Signature Verification System

Based On DWT and Common Feature”Journal

of Theoretical and Applied

InformationTechnology 20th May 2013.

2) Meera V. Kanawade & KatariyaS.S ”Signature

Verification And Recognition”International

Journal of Electronics, communication and

instrumentationEngineering Research and

Development (IJECIERD)\ ISSN 2249-684X

Vol. 3, Issue 1Mar 2013

3) Kamaldeep“Various Authentication Technique

For Security Enhancement” International

Journal of computer science & communication

networks OCT-NOV-2011

4) A.K.Jain & Arun Ross “An Introduction to

Biometrics” IEEE Transactions on Circuits and

system for video technology, vol. 14, no. 1,

January 2004.

5) AldebaroKlautan “The mfcc”AldebaroKlautau

- 11/22/05. Page 1.

6) A.K.Jain & Arun Ross “Learninguser Specific

parameters in bio biometric ” International

conference on Image processing 2002

7) Syed Khaled Ahmed “Worikng with

Matlab”IEEE(Malaysla section) February

21,2013

8) Lindasalwa Mudu,Mumtaj Begam and I.

Elamvazuthi “Voice Recognition Algorithm

Using Mel Frequency Cepstral Cofficient

(MFCC) and Dynamic Time Warping (DTW)

Techniques” Journal of computing, Volume 2 ,

Issue 3 March 2010.

Nikale Dipali B et al, Int.J.Computer Technology & Applications,Vol 5 (2),541-545

IJCTA | March-April 2014 Available [email protected]

545

ISSN:2229-6093


Recommended