ABSTRACT

Images containing faces are essential to intelligent vision-based human computer

interaction, and research efforts in face processing include face recognition, face
tracking, pose estimation, and expression recognition. Given a single image, the goal of
face detection is to identify all image regions which contain a face regardless of its threedimensional position, orientation, and lighting conditions. Such a problem is challenging
because faces are nonrigid and have a high degree of variability in size, shape, colour,
and texture. The faces in a given image are detected by using a neural network and
Gabor filter features. Considering the desirable characteristics of spatial locality and
orientation selectivitys of the Gabor filter, the filters had been designed for extracting
facial features from the local image. The goal of facial feature extraction is to detect the
presence and location of features, such as eyes, nose, nostrils, eyebrow, mouth, lips, ears,
etc. The feature vector based on Gabor filters is used as the input of the classifier, which
is a Feed Forward neural network.
The feed forward neural network consists of 3 layers, the input layer is
constituted by n2 units of neurons, hidden layer with one hundred neurons and one
neuron in the output layer which is active to 1 if the face is presented and to otherwise. A
feed forward learning algorithm was chosen for the system because of its simplicity and
its capability in supervised pattern matching.
The image will be convolved with Gabor filters by multiplying the image by
Gabor filters in frequency domain. Face detection and recognition has many applications
in a variety of fields such as security system, videoconferencing and identification. Face
classification is currently implemented in software. The objective of this work is to
implement a classifier based on neural networks and Gabor feature extraction for face
detection. The ANN is used to classify face and non-face patterns.

CONTENTS
Abstract

Contents

ii

List of Figures

iv

Abbreviations

vi

CHAPTER I

INTRODUCTION

1-2

CHAPTER II

FACE DETECTION IN IMAGE

3-19

2.1

2.2

Digital Image Processing

2.1.1 Picture upgrade

2.1.2. Point operations

2.1.3. Veil operations

Gabor filter

2.2.1 Explanation of parameters

2.2.2 2D Gabor Wavelet Representations Of Faces

13

CHAPTER III

NEURAL NETWORK

20-28

3.1

Introduction

20

3.2

Advantages of Artificial Neural Networks

21

3.3

Artificial Neural Networks for Image Processing

22

3.4

Feed-Forward Neural Network in Image Processing

23

3.5

Transfer Function

24

3.6

Network Layers

26

3.7

Optimization Methods

27

3.7.1. Steepest Descent

27

3.7.2 Newtons Method

28

CHAPTER IV

DESIGN

29-40

4.1

Requirements

29

4.2

Platform

29

4.3

Modelling the Neural Network: Issues to consider

30

ii

4.4

Optimization: Design Principles

30

4.5

Neural Network Architecture

34

4.6

Artificial Neural Network Model

38

4.7

Training Methodology

40

CHAPTER-V

IMPLEMENTATION

41-44

5.1

Program Components

41

5.2

Selection of Face Database

43

5.3

Image Resizing

43

5.4

Training Parameters and Weight Initialization

43

5.5

Training

44

CHAPTER- VI

RESULTS

46-51

REFERENCES

52-53

iii

LIST OF FIGURES
FIG NO

TITLE

PAGE NO

2.1

The pictures (of size 100 x 100) on the above shows Gabor
channel parts with estimations of the wavelength parameter of 5,
10 and 15

2.2

The pictures (of size 100 x 100) on the above shows Gabor
channel parts with estimations of the introduction parameter of 0,
45 and 90

2.3

The images (of size 100 x 100) on the above shows Gabor filter
kernels with values of the phase offset parameter of 0, 180, -90
and 90 degrees, from left to right,

2.4

The pictures (of size 100 x 100) on the above shows Gabor
channel parts with estimations of the angle proportion parameter
of 0.5 and 1,

2.5

The pictures (of size 100 x 100) on the above shows Gabor
channel parts with estimations of the transfer speed parameter of
0.5, 1, and 2,

2.6

Gabor filters correspond to 5 spatial frequencies and 8

13

2.7
(a&b)

Example of a facial image response to above Gabor filters, 6(a)

original face image (from internet database), and b) filter
responses.

14

2.8

Facial feature points found as the high-energized points of Gabor

wavelet responses

15

2.9

Flowchart of the feature extraction stage of the facial images.

16

2.10

First Section

17

2.11

Cell .net

17

2.12

Filtering above pattern for values above threshold (xy_)

17

2.13

Dilating pattern with a disk structure (xy_)

18

2.14

Finding the center of each region

18

2.15

Draw a rectangle for each point

18

2.16

Final Result will be like this-

18

2.17

Face Discovery in a database

19

3.1

Single Neuron without Bias

23
iv

3.2

Single Neuron with Bias

24

3.3

Threshold Function

24

3.4

Linear Transfer Function

25

3.5

Sigmoid Transfer Function

25

3.6

Single Layer Network

26

3.7

Multi Layer Network

27

4.1

The neuron of supervised training

34

4.2.

Architecture of proposed system

35

4.3.

Architecture of proposed system

36

4.4

Multi-Layer Artificial Neural Network

38

5.1

Main GUI executed by main.m

41

5.2

Raining process

44

6.1

Training data set with faces

47

6.2

Training data set without faces

48

6.3

Output obtained from the system which is taken from the digital
camera.

49

6.4

Output obtained from the system which is collected from the

internet.

49

6.5

Output obtained from the system. Totally there are eight faces are
there but only seven faces are detected due to blurriness.

50

6.6

Output obtained from the system

50

6.7

Output obtained from the system in which false detections are

present.

51

ABBREVIATIONS
GSM

Global system for mobile communication

v

GPS

Global Positioning System

LED

Light Emitting Diode

LCD

Liquid Crystal Display

IR

Infrared rays

MEMS

Micro Electro Mechanical System

RAM

Random Access Memory

LPC

Low Power Controller

KB

Kilo byte

vi