Face Recognition System Using Neural Network
A thesis
submitted towards the partial fulfillment of
the requirements of the degree of
Master of Engineering
In
Electronic Instrumentation and Control Engineering
Submitted By
Sourabh Gupta
Roll No-80651018
Under the esteemed guidance of
Sunil Kumar Singla
Sr.Lecturer,EIED
DEPARTMENT OF ELECTRICAL AND INSTRUMENTATION ENGINEERING
THAPAR UNIVERSITY
PATIALA –147004.
June - 2008
DEDICATED
TO
MY PARENTS
i
CERTIFICATE
This is to certify that my work presented in this thesis entitled “Face
Recognition System Using Neural Network” submitted in partial fulfillment
of the requirement for the award of the degree of Master of Engineering in
Electronic Instrumentation and Control Engineering at Thapar University,
Patiala, is an original record under supervision and guidance of Mr. Sunil
Kumar Singla (Sr. Lecturer).
The matter embodied in this report has not been submitted anywhere for the
award of any degree.
Date:
(Sourabh Gupta)
Roll No - 80651018
It is certified that the above statement made by the student is correct to the
best of our knowledge and belief.
( Sunil Kumar Singla)
Sr. Lecturer, EIED
(Supervisor)
Thapar University, Patiala
(Dr. Smarajit Ghosh)
Professor & Head, EIED
Thapar University, Patiala
(Dr. R. K. Sharma)
Dean of Academic Affairs
Thapar University, Patial
ii
ACKNOWLEDGEMENT
The real spirit of achieving a goal is through the way of excellence and austerous
discipline. I would have never succeeded in completing my task without the cooperation,
encouragement and help provided to me by various personalities.
First of all, I render my gratitude to the ALMIGHTY who bestowed selfconfidence, ability and strength in me to complete this work. Without his grace this would
never come to be today’s reality.
With deep sense of gratitude I express my sincere thanks to my esteemed and
worthy Supervisor Mr. Sunil Kumar Singla (Sr. Lecturer) in the Department of
Electrical and Instrumentation Engineering for his valuable guidance in carrying out this
work under his effective supervision, encouragement, enlightenment and cooperation.
Most of the novel ideas and solutions found in this thesis are the result of our numerous
stimulating discussions. His feedback and editorial comments were also invaluable for
writing of this thesis.
I shall be failing in my duties if I do not express my deep sense of gratitude towards
Dr. Smarajit Ghosh, Professor and Head of Electrical and Instrumentation Department
who has been a constant source of inspiration for me throughout this work.
I am greatful to Dr. R.K. Sharma, Dean of Academic Affair for his constant
encouragement that was of great importance in the completion of the thesis.
I extend my thanks to Dr. K.K. Raina, Deputy Director, Dr. Abhijit Mukherjee,
Director, Thapar University for their valuable support that made me consistent performer.
I am also thankful to all the staff members of the Department for their full
cooperation and help.
My greatest thanks are to all who wished me success especially my parents, my
sisters whose support and care makes me stay on earth.
Place: TU, Patiala
Date:
(SOURABH GUPTA)
iii
ABSTRACT
Biometric deals with identifying individual with the help of their biological data.
Human face recognition is a potential method of biometric authentication. Face is a
primary focus of attention in social intercourse, playing a major role in conveying identity
and emotion. The human ability to recognize faces is remarkable. People can recognize
thousands of faces learned throughout their lifetime and identify familiar faces at a glance
even after years of separation. This skill is quite robust, despite large changes in the visual
stimulus due to viewing conditions, expression, aging, and distractions such as glasses,
beards or changes in hair style etc. Face recognition is used in many applications such as
security systems, credit card verification and criminal identification. Due to numerous
potential applications face recognition has become a very active research area.
Face recognition is an interdisciplinary research area, involving researchers from
pattern recognition, computer vision, and graphics, image processing/ understanding,
statistical computing and machine learning. In the present work a neural network based
face recognition system has been developed. In the developed system the Gabor filter bank
is used to overcome the problem of rotation. The system is commenced on convolving a
face image after preprocessing the image at different scales and orientations. The neural
network is used as a classifier in which the weights of the neurons are updated by
supervised learning (target are set as .9 and .1) using Resilient backpropagation algorithm.
Yale database has been used for training the network and for testing the authenticity of the
person. If the person belongs to the trained database then the network will return the value
0.9 otherwise 0.1. With the developed system an accuracy of 90% has been achieved.
iv
TABLE OF CONTENTS
Certificate
ii
Acknowledgement
iii
Abstract
iv
Table of Contents
v-viii
List of Figure
ix-xi
xii
List of Table
Chapter 1 Introduction
1-11
1.1 Biometrics
1
1.2 Background
2
1.3 Biometric Systems
3
1.4 Biometric Modules
5
1.4.1 Sensor Module
6
1.4.2 Feature Extraction Module
6
1.4.3 Matcher Module
6
1.4.4 System Database Module
6
1.5 Application of Biometric Systems
7
1.6 Commonly used Biometrics
8
1.6.1 Facial recognition
8
1.6.2 Fingerprint
9
1.6.3 Hand/finger Geometry
9
1.6.4 Iris Scan
9
1.6.5 Keystroke Dynamics
10
1.6.6 Dynamic Signature Verification
10
1.6.7 Speaker/ Voice Recognition
10
1.7 Problem Formulation
11
v
Chapter 2 Literature Survey
13-28
2.1 Introduction
13
2.2 Human Face Recognition
14
2.3 Machine Recognition of Faces
15
2.3.1 Face Detection Methods
16
2.3.1.1 Knowledge Based Methods
16
2.3.1.2 Feature Invariant Approaches
17
2.3.1.3 Template Matching Methods
17
2.3.1.4 Appearance Based Methods
18
2.3.2 Face Recognition Method Classification
19
2.3.2.1 Holistic Approach
19
2.3.2.2 Feature Based Approach
19
2.3.2.3 Hybrid Approach
20
2.4 Face Recognition Methods
23
2.4.1 Eigenfaces Method
23
2.4.2 Nearest Feature Line Method
24
2.4.3 Hidden Markov Models Method
25
2.4.4 Hybrid Method of Component Based and Morphable Models
27
2.4.5 Gabor Filter
28
Chapter 3 Neural Network
30-48
3.1 Human Brain
30
3.2 Comparison Between Brain and Computer
32
3.3 Artificial Neuron
33
3.4 Activation Function
34
3.4.1 Identity Function
34
3.4.2 Step Function
35
3.4.3 Sigmoidal Function
35
3.4.3.1 Logistic Function (Sigmoid)
36
3.4.3.2 Symmetric Sigmoid
36
3.4.4 Radial Basis Functions
37
vi
3.5 Bias
38
3.6 Threshold
38
3.7 Learning
39
3.7.1 Fixed Networks
39
3.7.2 Adaptive Networks
39
3.7.2.1 Supervised Learning
40
3.7.2.2 Unsupervised Learning
40
3.7.2.3 Reinforcement Learning
40
3.8 Network Architecture
41
3.8.1 Single Layered Feedforward Networks
41
3.8.2 Multi Layered Feedforward Networks
41
3.8.3 Recurrent Networks
42
3.9 Learning Rules
43
3.9.1 Hebbian Learning Rule
43
3.9.2 Perceptron Learning Algorithm
43
3.9.3 Delta Learning Rule (Widrow-Hoff or Least Mean Square Rule)
44
3.9.4 Boltzamann Learning
46
3.9.5 Competitive Learning Rule
46
3.9.6 Back Propagation Algorithm
48
Chapter 4 System Implementation and Results
4.1 Introduction
51-74
51
4.1.1 Acquisition Module
52
4.1.2 Preprocessing Module
53
4.1.2.1 Histogram Equalization
53
4.1.2.2 Image Size Normailzation
57
4.1.3 Feature Extraction Module
59
4.1.4 Thresholding
67
4.1.5 Classification Module
68
4.1.5.1 Input Layer
68
4.1.5.2 Output Layer
69
vii
4.1.6 Training Set
69
4.1.6.1 Choice of Training Algorithm
69
4.1.6.2
71
Choice of Learning Rate
Conclusion
75
Future Scope
75
References
76-79
viii
LIST OF FIGURE
S.No.
Figure Number
Figure Name
Page No.
1.
Figure 1.2.1
Different types of Biometrics
3
2.
Figure 1.3.1
A Biometric System
5
3.
Figure 1.4.1
Biometric Modules
7
4.
Figure 2.3.1
Face Recognition System
16
5.
Figure 2.4.2.1
Generalizing two prototype feature points
25
x1 and x2
6.
Figure 2.4.3.1
HMM for face recognition
26
7.
Figure 2.4.3.2
Block extraction from image
26
8.
Figure 3.1.1
Biological Neuron
30
9.
Figure 3.1.2
Relation
between
different
parts
of
31
Biological Neuron
10.
Figure 3.3.1
Artificial Neural Net
33
11.
Figure 3.4.1.1
Identity Function
34
12.
Figure 3.4.2.1
Step Function
35
13.
Figure 3.4.3.1.1
Logistic Function (Sigmoid)
36
14.
Figure 3.4.3.2.1
Symmetric Sigmoid
37
15..
Figure 3.4.4.1
Radial Basis Function
37
16.
Figure 3.7.1
Neural Network
39
17.
Figure 3.8.2.1
A feed-forward network with a single
42
output layer of neurons (a) and a fully
connected feed-forward network with one
hidden layer and one output layer (b)
18.
Figure 3.8.3.1
A recurrent network with hidden neurons
42
19.
Figure 4.1.1
A face recognition system
51
20..
Figure 4.1.2
Yale Face Database
52
ix
21.
Figure 4.1.2.1.1
Face image of Yale database
55
22.
Figure 4.1.2.1.2
Histogram of image before equalization
55
23.
Figure 4.1.2.1.3
Image after histogram equalization
56
24..
Figure 4.1.2.1.4
Histogram of image after equalization
56
25.
Figure 4.1.2.1.5
Histogram equalized Yale Database
57
26.
Figure 4.1.2.2.1
(a) Original face image, (b) reduced face
58
image
27.
Figure 4.1.2.2.2
Resized image database
58
28.
Figure 4.1.3.1
Gabor filter bank at frequency .18 for
60
different orientation
29.
Figure 4.1.3.2
Gabor filter bank at frequency .19 for
60
different orientation
30.
Figure 4.1.3.3
Gabor filter bank at frequency .20 for
61
different orientation
31.
Figure 4.1.3.4
Gabor filter bank at frequency .22 for
61
different orientation
32.
Figure 4.1.3.5
Gabor filter bank at frequency .24 for
62
different orientation
33.
Figure 4.1.3.6
Gabor filter bank at frequency .28 for
62
different orientation
34.
Figure 4.1.3.7
Gabor filter bank at frequency .30 for
63
different orientation
35.
Figure 4.1.3.8
Convolved images with filter bank for
63
frequency .18
36.
Figure 4.1.3.9
Convolved images with filter bank for
64
frequency .19
37.
Figure 4.1.3.10
Convolved
image with filter bank for
64
Convolved image with filter bank for
65
frequency .20
38.
Figure 4.1.3.11
frequency .22
39.
Figure 4.1.3.12
Convolved image with filter bank for
x
65
frequency .24
40.
Figure 4.1.3.13
Convolved image with filter bank for
66
frequency .28
41.
Figure 4.1.3.14
Convolved image with filter bank for
66
frequency .30
42.
Figure 4.1.4.1
Thresholded images of typical face image
68
43.
Figure 4.1.6.1
Training curve for
70
(a) Gradient decent
(b) Gradient decent with momentum
(c) Adaptive rate
(d) Resilent Backpropagation algorithm
xi
LIST OF TABLES
S.No.
Table No.
Table Name
Page No.
1.
Table 3.2.1
Comparison of Brain and Computer
33
2.
Table 3.6.1
Relation
38
between
biological
neural
network and artificial neural network
3.
Table 6.1.4.2.1
Experimental result for various learning
74
rate
4.
Table 6.1.4.2.2
True test results
74
5.
Table 6.1.4.2.3
Imposter test results
74
xii