FACE RECOGNITION BIOMETRIC SYSTEM

connectionviewAI and Robotics

Nov 17, 2013 (3 years and 9 months ago)

134 views


FACE RECOGNITION BIOMETRIC SYSTEM

Ms.Poonam Mote
(M.E.student)

,

Prof.P.H.Zope

(
Asst.Professor)
,Prof.S.R.Suralker(Head of the Department)

College of Engineering And Technology,Bambori,
.

Department o
f E & Tc Engineering,North

Maharashtra University,
Jalgaon,Maharashtra

,

Email:
poonammote@rediffmail.com,
phzope@indiatimes.co
m


ABSTRACT

B
iometric authentication system is
widely used for human authentication

and
to increase the systems security.

In this

paper we
propose the
unimodal biometric system using the single


biometric
traits i.e. face
.
Gabor filter

and haar transformation
technique

is used
for

extracting the features from

face. In the proposed system
with good accuracy
the stored feature dataset is
update every tim
e hence the proposed system is more reliable than the others. As well as with an accurate authentication system
keep the record of login and logout time with total time spends of the user. This system is tested with the stand
ard data sets of
face
. The pro
posed system has lower higher accuracy

and less complexity of the system.

Keywords



Gabor filter,
haar like features,

Face

recognition.

1. INTRODUCTION




Automatic personal identification system

applications
are increased.
the biometrics
-
based authentication is widly
used in wide range of public application domains such as
National ID card, Electronic Commerce, ATMs etc. The
biometric systems such as face recognition,

fingerprint,
iris/retina and voice recognition

provide

a superior solution to
identifying

individuals because it uniquely

identify
individuals and also minimise the risk of

someone else using
another person’
s identity.
Face recognition
technique has a
more convenient

over rest of the techniques because it

is
flexible

in the sense that individuals are identified

actively by
standing in front of a face scanner, o
r

camera.

Biometrics
automatic systems are provides higher security the traditional
authentication systems. In biometric authentication persons
ide
ntification is based on their physiological and/or behavioral
characteristics. Biometric systems are more accurate and
provides more convenience. The biometric systems has the
advantages over traditional authentication system such as
eliminates fraud,

Enha
nces security, Cannot be easily
transferred, forgotten, lost or copied,

eliminates repudiation
claims.



Considerable amount of work has been published for
Gabor based image recognition.

In our proposed system we
used

t
he output obtained by using Gabor filter is good as
compared to the other methods.
[1
] presents a Gabor filter
coefficient based approach for face recognition. Also to reduce
the feature dimention this paper uses the 15 Gabor
filters.
Gabor filter have the
pro
perties of spatial localization
,
orientation selectivity and spatial
-
frequency selectivity

.Therefore, Gabor filter have been applied to many fields, such
as texture classification ,face recognition ,handwritten
character recognition, fingerprint classi
fication and fingerprint
recognition. It handles sensitively the different orientations in
the fingerprint image and it provide a robust representation is
with respect to minor local changes thus, individuals can be
recognized in spite of different facia
l expressions and poses.
[7]

This paper describes utilization of

Gabor filters for selecting
feature
vectors/coefficients. It

explains construction of Gabor
filters, selectio
n of peaks, feature storage and
classification
of
faces. Standard and improved
clas
sifier schemes have been
use
d to evaluate the
developed face

recognition system in
terms of
detection rate and speed by using 20 to 60 Gabor

filters.

[3]uses the three multiscale representation techniques
Gabor filter,

Log Gabor

filter and Discr
ete Wavelet Transform
are applied to reduce dimentions.

There are several
approaches
to face recognition[5]
incorporate P
CA and Gabor wavelet
approach.[10
] describes a fast identific
ation method with
Gabor filters i.e. uses optimal Gabor filters and Gabor
T
ransformation

.In order to reduce the processing time design
a Gabor filters with a filter arrangement theory

.
A system
which can detect and recognize faces regardless of pose
reliably and in real
-
time based on Haar
-
like features. Haar like
features are i
ntroduced by Viola and improved by
Lienharefines the shape of an object in terms of a subset of the
wavelet coefficient of the image. The detection technique is
based on the idea of the wavelet template that defines the
shape of the image.


The paper is o
rganized as foll
ows: in section II, we describe
the steps of

face detection .In section III, we describe the
procedure of featu
re extraction
face.
Then

in section

I
V we
shows the expe
rimental resul
ts.

In section V

we draw the
conclusion.

2
. FACE DETECTION


Face detection is defined as t
o determine whether or
not there are any faces in the image and if present, return the
image location and extent of each face.

This is the first step of
any fully automatic system that analyzes

the information
con
tained in faces (e.g., identity, gender, expression, age, race
and pose)
[8].
The challenges
of face detection
are
pose,frontal,45
0
,profile,upside down,

presence or absence of
structural

components
like beards
, mustache ,glasses, scarf,
facial
expressions

,occlusion,

image orientation

,image
condition i.e. lighting,

camera characteristics.

Therefore

Face
detection from cluttered images is very tough
.

The most
popular approaches to face recognition are based on i)the
location and shape of facial attributes s
uch as eyes, eyebrows,
nose ,lips and chin and there spatial relationships ,ii)the overall
analysis of face image represents a face as a weighted
combinations of number of conical faces.



In our proposed system we simply used the Gabor filter
with Haar Transformation


technique for feature extraction from
face which is used for face recognition.

Before extracting the
features from face we followed some preprocessing steps
which includes apply H
aar transformation algorithm [2
] for
detecting the face, cropping of image, centralization.
The figer1
shows the steps followed before feature extraction.




Figer1.Steps followed before feature extraction.


Haar like feature extraction using Adaboost algorit
hm [2]used
for face detection.
The haar like feature is specified by it’s
shape, position and the scale. In proposed system we use the
haar like feature algorithm for face detection from open CV
library and detect the face.

2.1
face detection using haar
like features

[2][
9
]
Following the Adaboost algorithm

a set of weak binary
classifier is learned from a training set. Each classifier is a
simple function made up of rectangular sums followed by a
threshold. In each round of boosting one feature is selected
with lowest error. An input window is passed from o
ne
classifier in cascade to the next as long as each classifier
classifies the window as a
face. The

image features are called
Rectangle Features .Each rectangle feature is binary threshold
function constructed from a
threshold, and a rectangle filter is
a

linear function of the image.
The value of a two
-
rectangle
filter is the difference between the sums of the pixels within
two rectangular regions. The regions have the same size and
shape and are horizontally or vertically adjacent.There is three
-
rectangle

filter,four

rectangle filter is also used. Computation
of rectangle filters can be increased using an intermediate
image because of this any rectangle filter at any scale or
location can be evaluated in constant time.
figer 3 shows the
rectangle filters.


Figer3.Rectangle filters



An input window evaluated on first classifier of the cascade
and if classifier returns false then computation on window end
and detector returns false and if it is true then window is
passed to the next classifier in the cascad
e.The more windows
look like a face then it get classify as face otherwise it quickly
discarded as non
-
face.



i(R1)=

Σ

i(x,y)

(1)




(x,y
ϵ
R1)

i(R2)= Σ i(x,y) (2)


(x,y
ϵ
R2)


If i(R1)
-
i(R2)>C

C

is

constant

threshold
.

i(x ,y)is pixel luminance value.

After face detection face image is get centralized and then is
cropped

of the size 175 ×175.

Then cropped the image for
removing the unwanted noise fro
m the image.

Figer2 shows
the cropped image.


Original image cropped image

Figer2.cropped image

3. FEATURE EXTRACTION


After cropping of face image a

circular region around
the core point is

located and tessellated into 64

sectors. The
pixel

intensities in eac
h sector are normalized to a constant

mean and va
riance. Gabor filter is

a widely used

to capture
useful information in specific band

pass channels
. The average
absolute deviation with in a

sector quantifies the underlying
ridge structure and is used

as a

feature. The feature vector (
1280
values in length)

is

the collection of all the features, computed
from all the

64 sectors, in every filtered image.
It is desirable to
obtain
representations for face

which are translation and
rotation invariant.


In the proposed scheme, translation is taken care of by
a
center point
during the feature extraction stage and the image
rotation is handled by a cyclic rotation of the feature values in
the feature vector. The features are cyclically rotated

(
clockwise and anticlockwise)

to generate feature vectors
corresponding to different orientations to perform the
matching.By tuning a Gabor filter to specific frequency and
direction, the local frequency and orientation information can
be obtained. Thus, th
ey are suited for extracting texture
information from images. An even symmetric Gabor filter has
the following general form in the spatial domain
[1]
:

(3
)


x’=xsin
θ+ycosθ


(4)


y’=xcos

θ
-

ysin

θ

(5)


where
f
is the frequency of the sinusoidal plane wave

along
the direction θ from the
x
-
axis, and
x
δ and
y
δ

are the space
constants of the Gaussian envelope alon
g
x

and
y
axes,
respectively.

The filtering is performed in the spatial domain
with a

mask size of
16x1
6
. Hence, the
face
can examined at
different orientations and this correspond to
θ
values
(0
0
,45
0
,90
0
,…,315
0
)
.

At the matching stage

the

gabor

features of train and test image are compared and distance has
been calculated, if the distance is within threshold limit the
image is said to be

matched.
After matching the previously
stored template is replaced by existing new template.because
of updati
ng the dataset at every authentication the system will
give the same accuracy over the years.


3. EXPERIMENTAL RESULTS


The reliab
ility of the proposed unimadal

system is
described with the help of experimental results. The system
has been tested on three st
andard datasets for face
(att,ifd)
,each
dataset has nine images of each individual person with
different orientation as well as with different facial expressi
ons

and also

the system is

tested

on

the face image

acquired
using 3
-
CCD camera.

We implemented this method in
MATLAB7.5.0(R2007b version ) and processed on Pentium
machine 20.2 GHz.


The performance of the system

for two standard datasets

is
shown by
comparison

of the accuracy with the other
methods such as PCA and ICA techniques of face recognition
which are

widely used tec
hniques in AFIS shown in figer.3

and
Table I
.
Proposed sy
s
tem gives 94.38% of accuracy for att
database.

The average CPU
time for one test is 1
.68sec for
face.

TABLE II RECOGNITION PERFORMANCE OF SYSTEM


Algorithm



Dataset

a
tt


ifd


Gabor filetr+Haar transform


94.38%

73.75%

PCA


91.88%

70.42%

ICA

90.63%

67.50%






Figer 3.Performance of the proposed system with different
datasets

TableI

is drawn to show the performance of the proposed
system based on seven different factors They are
universality,permanence,collectability,performance,acceptabili
ty,distinctiveness,circumvention .Jain et al.[3]

have identified
these factors to determine the

feasibility of any biometric
system for various applications.
The factors are explained as,



Universality
: each person should have the characteristic.


Distinctiveness
: any two persons should be sufficiently
different in terms of the characteristic.


P
ermanence
: the characteristic should be sufficiently
invariant (with respect to the matching criterion) over a period
of time.


Collectability
: the characteristic can be measured
quantitatively.


performance
:which refers to the achievable recognition
acc
uracy and speed, the resources required to achieve the
desired recognition accuracy and speed, as well as the
operational and environmental factors that affect the accuracy
and speed;


acceptability
, which indicates the extent to which people are
willing
to accept the use of a particular biometric identifier
(characteristic) in their daily lives;


circumvention
, which reflects how easily the system can be
fooled using fraudulent methods.



TABLE

I
I
.PERFORMANCE OF PROPOSED SYSTEM


(H
-
HIGH,L
-
LOW,M
-
MEDIUM)

Biometric Trait

universality

permanence

collectability

performance

acceptability

distinctiveness

circumvention

Face

H

H

H

H

H

M

M



The proposed
system can be used for the various organization

where the login and logout time is recorded. Figer
4 and 5
shows the examples of tested images with login and logout
time.




Figer.4Image stored in database

0%
20%
40%
60%
80%
100%
gabor
pca
ica
att
ifd







Figer.5Results
during Authentication.


5.COCLUSION


Now a days the biometric systems are widely used for
authentication to overcome the problems of traditional
authentication.

Image
-
based face recognition is still a very
challenging topic. A number algorithms

are used for face
recognition system but still there is challenging problems for
sensitivity to variations in pose and different lighting
conditions. This paper presents a Gabor filter based approach
for face recognition The performance of the system has

been
improved

gives 94% accuracy.

Our next step will be to reduce
response time of system .


6.

REFERENCES

[1]
Md.Tajmilur Rahman,Md.Alamin Bhuiyan”Face Recognition
using Gabor Filters”,11
th

International Conference on Computer and
Information Technology 2
008IEEE.

[2
]Zhaomin zhu,Takashi Morimoto,Hidekazu Adachi,Osamu
Kiriyama,”Multi
-
view Face Detection and Recognition using
haar
-
like Features”

[3]DMurugan,Dr.SArumugam,KRajalakshmi,Manish,”Perfor
mance Evaluation of Face Recognition Using Gabor Filter,Log
Gab
or filter and Discrete Wavelet Transform”,International
journal of computer science and information
technology(IJCSIT),Vol.21,February 2010.

[4]

Anil K. Jain, Arun Ross and Salil Prabhakar” An
Introduction to BiometricRecognition”
,
IEEE

Transactions on
circuits and systems for Video Technology, vol. 14,

no. 1,
January
2004

[5]R.Prema,Prof.K.ThirunadanaSikamani,Prof.R.Suguna,”A
Novel Feature Extraction Scheme for face Recognition”,978
-
1
-
4244
-
85
94
-
9/10/26.00
©
2010IEEE

[6]

A.S.M Shihavuddin1, Mir Mohammad Naz
mul Arefin2, Mir
Nahidul Ambia3, Shah Ahsanul Haque4 andTanvir Ahammad2,”

Development of real time Face detection system using

Haar like
features and Adaboost algorithm
”,

IJCSNS International Journal of
Computer Science and Network Security, VOL.10 No.1, J
anuary
2010

[7]

Syed Maajid Mohsin1 Muhammad Younus Javed 2 Almas
Anjum3,”

Face Recognition using Bank of Gabor Filters
”, IEEE
--
ICET 2006
,
2nd International Conference on Emerging Technologies

Peshawar,Pakistan,13
-
14November
2006

[8]Ming
-
Hsuan

Yang
,
“Face
Detection”, Un
iversity of California,
Merced,CA
95344


[9]

Rainer Lienhart and Jochen Maydt,
“ An Extended Set of Haar
-
like Features for Rapid Object Detection”,

Intel Labs, Intel
Corporation,Santa
Clara, CA 95052, USA


[10

Haiyuan WU

, Yukio YOSHIDA and Tadayoshi SHIOYAMA
,”

Optimal Gabor Filters for High Speed Face Identification”