HUMAN FACE DETECTION USING SKIN COLOR TONE
BY
HEWA MAJEED MOHIALDEEN
A dissertation submitted in fulfilment of the requirement for the degree of Master of Information Technology
Kulliyyah of Information and Communication Technology International Islamic University Malaysia
JULY 2013
ii
ABSTRACT
Human face recognition systems have gained a considerable attention during last decade due to its vast applications in the field of computer and advantages over previous biometric methods. There are many applications with respect to security, sensitivity and secrecy. Human face detection is the most important and first step of recognition system. This dissertation introduces a new approach to face detection systems using the skin color of a subject. This system can detect a face regardless of the background of the picture, which is an important phase for face identification. The images used in this system are color images which give additional information about the image than the gray images provide. In human face detection, the Two respective classes are the "face area" and the "non-face area". This new approach to face detection is based on color tone values specially defined for skin area detection within the image frame. The proposed system first resizes the image, and then separates it into its component R, G, and B bands. These bands are transformed into another color space which is YCbCr space and then into YC’bC’r space (the skin color tone). The morphological process is implemented on the presented image to make it more accurate. At last, the projection face area is taken by this system to determine the face area. Experimental results show that the proposed algorithm is good enough to localize a human face in an image with an accuracy of 92.69%.
Key Words: Face detection, Skin Color, Color Space.
iii
ثحبلا صخلم
اتهاقيبطتل آرظن يرخلأا دقعلا للاخ آيربك آمامتها ناسنلأا هجو نع فشكلا ةمظنأ تبستكأ انه .بوسالحا لامج في ةعساولا ل
ةيساسلحاو ،نملأاب قلعتي اميف تاقيبطتلا نم ديدعلا ك
.صخشلا ىلع فرعتلا ماظن في ةوطخ لوأو مهأ وه ناسنلأا هجو نع فشكلا .ةيرسلاو هذه نلأ آديدج آجنه مدقت ةحورطلأا ظ
.ةرشبلا نول مادختسأب كلذو هجولا نع فشكلا ةم
ةماه ةلحرم دعت تيلاو ،ةروصلا ةيفلخ نع رظنلا ضغب هجولا نع فشكلا هنكيم ماظنلا اذه تامولعم يطعت تيلاو ةنولم روص يه ماظنلا اذه في ةمدختسلما روصلا .هجولا ديدحتل نم رثكأ ةروصلا لوح ةيفاضإ اهم ناتقطنم كانه ،ناسنلإا هجو فشك في .ةيدامرلا روصلا
ديدلجا جهنلا اذه دنتسيو ."ناسنلأل آهجو تسيل تيلا ةقطنلما" و ،"ناسنلأا هجو ةقطنم"
ةقطنم نع فشكلل آصيصخ ةددلمحا نوللا ةجرد ميق ىلع هجولا فشك لىإ ةرشبلا
لخاد
مجح يريغتب لآوأ موقي حترقلما ماظنلا .ةروصلا راطإ تيلا اتهانوكم لىإ اهلصفي ثم نمو ،ةروصلا
ناوللأا نم رخآ عون لىإ حئارشلا هذه ليوتح متي ثم .قرزلأاو رضخلأاو رحملأا حئارش يه ىمسي يذلا YCbCr
لىإ ثم نمو YC'bC'r
ةيلمع ذيفنت متي ثم نمو .)ةرشبلا نول(
ةياهنلا في .ةقد رثكأ اهلعلج ةضورعلما ةروصلا ىلع يجولوفروم ىمست ةقطنم ديدتح متي ،
ابم ةديج ةحترقلما ةيمزراولخا نأ ةيبيرجتلا جئاتنلا رهظت امك .طاقسلأا ةيلمع قيرط نع هجولا يواست ةقد عم ةروصلا في ناسنلأا هجو فاشتكأو ديدحتل يفكي 96.29
.٪
iv
APPROVAL PAGE
I certify that I have supervised and read this study and in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for degree of Master of Information Technology.
….………
Imad Fakhri Al-Shaikhli Supervisor
I certify that I have read this study and in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for degree of Master of Information Technology.
….………
Akram M. Zeki Internal Examiner
This dissertation was submitted to the Department of Information Systems and is accepted as fulfilment of the requirement for degree of Master of Information Technology.
……….………
Muhd Rosydi Muhammad
Head, Department of Information Systems
This dissertation was submitted to the Kulliyyah of Information and Communication Technology and is accepted as fulfilment of the requirement for degree of Master of Information Technology.
…...………...
Tengku Mohd Tengku Sembok Dean, Kulliyyah of Information and Communication Technology
v
DECLARATION
I hereby declare that this dissertation is the result of my own investigation, except where otherwise stated. I also declare that it has not been previously or currently submitted as a whole for any other degrees at IIUM or other institutions.
Hewa Majeed Mohialdeen
Signature ……… Date: ……….
vi
INTERNATIONAL ISLAMIC UNIVERSITY MALAYSIA
DECLARATION OF COPYRIGHT AND AFFIRMATION OF FAIR USE OF UNPUBLISHED RESEARCH
Copyright © 2013 by Hewa Majeed Mohialdeen. All rights reserved.
HUMAN FACE DETECTION USING SKIN COLOR TONE
No part of this unpublished research may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without prior written permission of the copyright holder except as provided below.
1. Any material contained in or derived from this unpublished research may only be used by others writing with due acknowledgement.
2. IIUM or the library will have the right to make and transmit copies (print or electronic) for institutional and academic purposes.
3. The IIUM library will have the right to make, store in a retrieval system and supply copies of the unpublished research if requested by other universities and research libraries.
Affirmed by Hewa Majeed Mohialdeen
……… ………
Signature Date
vii
ACKNOWLEDGEMENTS
Foremost, I am grateful to the Almighty God for establishing me to complete this dissertation. I would like to express my sincere gratitude to my advisor Assoc. Prof.
Dr. Imad Fakhri Al-Shaikhli for the continuous support of my research, for his patience, motivation, enthusiasm, and immense knowledge. His guidance helped me in all the time of research and writing of this dissertation. I could not have imagined having a better advisor and mentor for my dissertation. Besides, my sincere thanks also go to fallow my dear mother, for all the encouragement and support that she has given me throughout my life and during my pursuance of higher education in particular. My gratitude also goes to my father Majeed Mohialdeen Zangana “Allah’s mercy be upon him” and my dear brothers and sister for their unceasing encouragement and support. I also place on record, my sense of gratitude to one and all who, directly or indirectly, have lent their helping hand in this dissertation.
viii
TABLE OF CONTENTS
Abstract ………. ii
Abstract in Arabic ………... iii
Approval Page ………... iv
Declaration ……….... v
Declaration of Copyright ……….. vi
Acknowledgement ……….... vii
List of Figures ………... xi
List of Tables ……….... xii
CHAPTER 1: INTRODUCTION ……….……….. 1
1.1 Overview ……….…..………... 1
1.2 Problem Statement ………..………... 3
1.3 Significance of The Study ………..…………..……….. 3
1.4 Aims and Objectives ………....………..…………... 4
1.5 Expected Outcome ………..……….………..…. 4
1.6 What is The Face Detection? ……….……..………... 4
1.6.1 Challenges ………..……… 5
1.7 Possible Applications for Human Face Detection ………...…. 6
1.8 Research Structure …………..…….…………..…………... 6
CHAPTER 2: LITERATURE REVIEW ………... 7
2.1 Background ………….…….…..……...……….. 7
2.2 Face Detection in Image ………….……….. 12
2.2.1 Knowledge Based Method ……..………...….……... 12
2.2.2 Features Based Method .……….…...………...………….. 13
2.2.3 Template Based Method ……...………..………... 14
2.2.4 Appearance Based Method …..…....………..……… 15
2.3 Color Specifications ………...………... 16
2.4 Color Spaces ………..……….……….... 16
2.5 Human Skin Color ...…………..…...………..………... 17
2.6 The Skin Color as a Feature ………..…...……..……... 19
2.7 Skin Tone Detection and Localization of Facial Region …….... 19
2.8 Morphology ………..………... 21
CHAPTER 3: INTRODUCTION TO IMAGE PROCESSING ………. 23
3.1 What is The Image? ………...……….…..……….. 23
3.2 What is The Pixel? ………..….………... 23
3.3 Image Processing ……….………... 23
3.4 Image Enhancement ……...………….………... 24
3.5 Image Restoration ……….………... 25
3.6 Color Image Processing ………..………….……... 26
ix
3.6.1 Radiance in Image Processing ………...…….... 28
3.6.2 Luminance in Image Processing ………...…... 28
3.6.3 Brightness in Image Processing ………...………….. 28
3.6.4 Hue in Image Processing ………...………... 28
3.6.5 Saturation in Image Processing ………...…………... 28
3.7 Image Segmentation ………..………. 29
3.8 Edge Detection ……….………... 30
3.9 Histogram Processing ……….……… 31
3.10 Thresholding ……….………... 32
3.11 Face Detection Approaches ……….………... 34
3.11.1The Top Down Method-Based Approach ...………... 34
3.11.2 The Bottom Up Feature-Based Approach ……...…... 34
3.11.3 In Texture Based Approach ………...……….. 34
3.11.4 The Neural Network Approach ...…...………... 34
3.11.5 The Color Based Approach ……...………... 35
3.11.6 Motion Based Approaches ……...……… 35
3.11.7 At Depth Based Approach …………...……… 35
3.12 Color Models for Skin Color Classification .………... 36
3.12.1 RGB Color Space ……...………... 36
3.12.1.1 Algorithm of RGB Color Space for Face Detection Using Skin Color……….…. 37
3.12.2 YCbCr Color Space ……….. 37
3.12.2.1 Algorithm of YCbCr Color Space for Face Detection Using Skin Color ………...……. 38
3.12.3 HSI Color Space ………...……….... 40
3.12.3.1 Algorithm of HSI Color Space for Face Detection Using Skin Color ……...…………... 41
3.13 Summary ……….……… 43
CHAPTER 4: PROPOSED FACE DETECTION SYSTEM ………….. 44
4.1 Introduction ……….………... 44
4.2 Face Image ……….………... 45
4.3 Color Transform ……….……… 47
4.4 Proposed Color-space Transformation ……….……... 49
4.5 Skin Color Tone Detection ………….………....……… 51
4.6 Summary ……….……….... 53
CHAPTER 5: EXPERIMENTAL RESULTS AND COMPARISON … 54 5.1 Introduction ……….………..……….. 54
5.2 Experimental Results ………….………..………... 54
5.2.1 Results of RGB Color Space for Face Detection Using Skin Color ………..………... 54
5.2.2 Results of YCbCr Color Space for Face Detection Using Skin Color .………...……….…... 54
5.2.3 Results of HSI Color Space for Face Detection Using Skin Color ………...………... 55
5.2.4 Results of Proposed Algorithm ....………... 55
5.3 Comparison of Algorithms ………... 56
x
5.4 Summary ……….……….... 57
CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS ……… 58
6.1 Conclusions ……….……… 58
6.2 Suggestions for Future Works ……… 59
REFERENCES ………. 60
APPENDIX A: MATLAB CODE ... 65
APPENDIX B: IMAGE LIBRARY ..……… 69
xi
LIST OF FIGURES
Figure No. Page No.
2.1 Human skin chromatic distribution 18
2.2 New nonlinear transformation of YCbCr color space 20
3.1 Eyes 25
3.2 X-ray 25
3.3 Image restoration 26
3.4 Refraction of light 27
3.5 Gray-scale 27
3.6 RGB image components 29
3.7 Edge detection 31
3.8 Histogram Parts 32
3.9 Thresholding 33
3.10 RGB cube 36
3.11 HSV-cone 42
4.1 Block diagram explaining proposed face detection system 45 4.2 Face image and original three color bands 46 4.3 An example of YCbCr Transform space color system 48 4.4 An example of YĆbĆr New Transform space color system 50
4.5 The skin color tone detection 51
4.6 Projection of face areas 51
4.7 The Proposed face detection system stages 52 5.1 Sample results of proposed face detection algorithm 56
xii
LIST OF TABLES
Table No. Page No.
2.1 Major face detection approaches 8 - 9
5.1 Skin color classification results for proposed algorithm 56
5.2 Comparison of Algorithms 57
1
CHAPTER ONE INTRODUCTION
1.1 OVERVIEW
When people are talking to each other, people usually look at other people’s faces; the expression of a human’s face plays a very important role when people are communicating. Because of its uniqueness, the human face is also the common influential and significant characteristics to recognize a human.
Comparing with retinas or fingerprints, picking an image of face of a human is very simple. Thus human face detection has become one of the most common implementations in the area of computer vision. Under the current attacks of terrorist in the civil world, there have been growingly essential benefits in the evolution of smart monitoring cameras which can automatically locate and recognize familiar malefactors’ likewise dubious characters. Because of such unreliable times, humans are starting to request support from computer systems to assist in the process of matching and position of faces in daily sights (Leung, 2001).
Images of faces vary considerably depending on lighting, occlusion, pose, facial expression, and identity. Color transforms must be implemented to deal with all remaining variation in distinguishing face skin color.
For human face detection there are many algorithms including algorithms using skin color tone. Skin color is a substantial feature of human faces. Using skin color tone as a feature for pursuing a face has many characteristics (Tabatabaie et al., 2009).
Color image processing is faster than processing other facial features. Under fixed terms of lighting, color is directing invariant. This attribute makes motion assessment
2
simpler because just an interpretation model is necessary for motion assessment.
Anyway, color is not a tangible phenomenon; color is a perceptual phenomenon which is linked to the spectral advantages of electromagnetic radiation in the visual wavelengths striking the retina. Pursuing human faces using skin color as a feature has many troubles such as the color exemplification of a face acquired by a camera is affected by several factors (object movement, ambient light, etc.) (Tabatabaie et al., 2009), various cameras show significantly various values of color even for the same human under the same lighting terms and the skin color varies from human to another.
For using skin color as a feature for human face detection, it is necessary to fix these troubles. It is also tough towards modifications in scaling and orientation and can afford occlusion well. The color cue has a disadvantage which is its allergy to illumination color modifications and, particularly in the state of RGB (Red, Green, and Blue), allergy to illumination density (Roy & Bandyopadhyay, 2013). There is only one way to raise allowance towards density modifications in images which is to convert the RGB (Red, Green, and Blue) image into another color space whose chromaticity and density are divided and just use the part of chromaticity for detection.
This dissertation has introduced a study of three known skin color human face detection algorithms and has come-up with a new algorithm using skin color tone in YCbCr color model. The results of proposed system have show that it can detect/localize the face more accurately and effectively by using the proposed system.
3 1.2 PROBLEM STATEMENT
The majority of already existing algorithms for human face detection mostly have at least one of the following problems:
- High computational either in time or space complexity.
- Low accuracy/effectiveness.
In the first problem talking about time because some of the existing algorithms take long time to be run, while some other algorithms take a lot of space of memory because of its high complexity and computational. In the second problem mentioned about the effectiveness which means the accuracy is low in the algorithms that have less computational.
It is important to confirm the fact which is automatic human face detection and also most other automatic object detection techniques is a very ambitious task, especially due to significant sample variations, which cannot be easily analytically described with parameter.
1.3 SIGNIFICANCE OF THE STUDY
Human Face Detection could show additional significance for future Robot/Human Interaction, where a robot requires first to locate and detect the human being in the normal images it views. More steps could be the analyzing regions within the hand/face to localize the fingers or eyes (Peer & Soline, 2003). The current research investigated different methods of human face detection which are using human skin color model matching.
Face detection of human skin color in Two dimensional plain and complicated image is a hard and interesting task in computer vision region that aims to localize and detect human face region. Moreover, this task is considered as the first step in a great
4
number of troubles like hand and human face detection. Face detection is also very significant step in implementations such as hand signals and sex/race detection, automatic identification, and understanding of facial expressions. These implementations are depending on the supposition that the regions of human skin are formerly located and detected.
1.4 AIMS AND OBJECTIVES
The target of the dissertation is to estimate the human face detection operation depending on skin color tone, so that advantages and applications of skin color can club together into a single proposed algorithm. The objectives of the dissertation are:
1. To understand the different methods of face detection available under which various techniques are used.
2. To devise a new suitable algorithm which will be more accurate and have high detection ratio?
3. To compare the proposed algorithm of face detection with the other available algorithms.
1.5 EXPECTED OUTCOMES
The expected output of this new human face detection algorithm using skin color tone is the same image which is given as input to the system but the output image will only contain the face of that person who was in the input image.
1.6 WHAT IS THE FACE DETECTION?
It is known as discovering the human faces existing in a picture and detecting size and position of each of them. Generally the operation of human face recognition is divided
5
into two phases; where in the first phase of face detection the complete picture is inspected in order to discover the region which can be specified as face, skin color is the common among all. And the next phase is called localization which provides additional accuracy assessment of position and size of the human faces. Human Face recognition considered as the first phase of human recognition and identification method.
1.6.1 Challenges: Face detection methods encompass various challenges such as pose and imaging condition.
Pose: The image of a face varies because of comparative camera face pose and some facial feature becomes partially or completely occluded.
Image orientation: The face images straightly differ for various rotations with the camera’s optical axis. Image orientation immediately affects the angle of the face.
Illumination: This problem is mainly due to lighting, makes a larger difference with the same face as compared to difference within different faces while comparing (Javidi.Bahram, 2002).
Occlusion: Sometime, faces in the image are occluded with other objects such as moustaches, beard, optical lenses & other types of object which make it very difficult to find the accurate image.
Facial expression: Some persons may have different expression at different times; this also contributes to challenges in face detection.
Face size: Size of the face also make it difficult to automate a system for face detection and recognition (Mahmood, 2006).
6
Image condition: This problem includes factors such as intensity, resolution, camera lighting, background, characteristics of image capturing device and distance between camera and person, plays an important role in the process of face detection.
1.7 POSSIBLE APPLICATIONS FOR HUMAN FACE DETECTION
Several applications can be considered for automatic human face detection as follows:
* Video conference implementation.
* Security Implementation.
* Supervision Application.
* Remote camera control implementation.
* Animation of facial expressions.
1.8 RESEARCH STRUCTURE
The organization of the research is done in this section as follows:
Chapter 1: Introduction to face detection covers the aim and objectives of face detection and also the problems faced in achieving effective face detection.
Chapter 2: Briefly describe the methods of face detection and explain the different techniques for face detections.
Chapter 3: Introduction to Image Processing.
Chapter 4: The Proposed Face Detection System.
Chapter 5: Experimental Results and Comparison.
Chapter 6: Conclusion and Future works.
References
Appendix A – MATLAB Code Appendix B – Image Library
7
CHAPTER TWO LITERATURE REVIEW
2.1 BACKGROUND
From the advent of cell phones in the electronics history till today there is continuous development in the technology used in the cell phones. Modern cell phones have so many features such as integrated cameras, high resolution display, multiple communications interfaces, and processing power and many more equal to at least, to a mid-1990’s PC. Cell phones are most widely distributed in terms of range of users, geographical distribution than any other class of computers. It makes them a catchy evolution platform for several image processing and computer vision algorithms.
Among these algorithms face detection is early step in many computer vision systems, including Video communication, Super video compression, Face recognition, Video surveillance, Responsive user interface, Video augmented speech understanding, Intelligent autofocus etc. (Jon A.Web, 2010).
There are many algorithms and representative works for human face detection, Table 2.1, epitomizes these works and algorithms.
8
Table 2.1: Major face detection approaches
Authors Year Approach Features Used
Hlaing Htake Khaung Tin 2012 Multiple features facial features and Eigen faces
Wagner, Andrew et al 2012 alignment and illumination sparse representation
Smita T., Varsha S., Sanjeev S.
2011 Multiple features Skin Color and Template Matching
Toole A.J. et al 2011 Dissecting identity fusion
Youngeun An, Muhammad Riaz and Jongan Park
2010 HSV color space CBIR based on adaptive segmentation Sania Arjomand Inalou,
Shohreh Kasaei
2010 Generic linear features AdaBoost
Nidal F. Shilbayeh, Gaith A.
Al-Qudah
2010 Neural Network MLP
Khalid M.Alajel, Wei Xiang, John Lies
2010 Multiple features Skin Color And Facial Features Wang X.et al 2009 Statistics-based
features
HoG and LBP
Yan S. et al 2008 Binarized features Locally assembled binary feature
Hotta K. 2007 SVM – multi view face detection
Local and global kernels
J. Meynet, V. Popovici, and J.-P. Thiran
2007 Generic linear features Anisotropic Gaussian filters
P. Sabzmeydani and G.
Mori
2007 Shape features Shape let
Zhang L. et al 2007 Binarized features LBP features
Heisele B. et al 2007 Part-based approaches SVM component detectors adaptively trained
Opelt A.et al 2006 Shape features Boundary/contour fragments
Huang C. et al 2006 Composite features Sparse feature set
Tuzel O. et al 2006 Statistics-based features
Region covariance
Zhang H. et al 2006 Statistics-based features
Spatial histogram (LBP- based)
9
B. Wu and R. Nevatia 2005 Shape features Edge let
Shotton J. et al 2005 Shape features Boundary/contour fragments
Mita T. et al 2005 Composite features Joint Haar-like features
C. A. Waring and X. Liu 2005 Statistics-based features
Spectral histogram
N. Dalal and B. Triggs 2005 Statistics-based features
Edge orientation histograms etc.
P.Wang and Q. Ji 2005 Generic linear features RNDA
Y. Abramson and B. Steux 2005 Pixel-based features Control point set
Mikolajczyk K. et al 2004 Part-based approaches Overlapping part detectors
H. Schneiderman and T.
Kanade
2004 Part-based approaches Wavelet localized parts
C. Garcia and M. Delakis 2004 Neural networks Convolutional neural network Wang P. and Ji Q. 2004 SVM – multi view face
detection
Cascade and bagging
Ratsch M. et al 2004 SVM – speed up Reduced set vectors and approximation K. Levi and Y. Weiss 2004 Statistics-based
features
Edge orientation histograms etc.
10
Yang and other researchers (Yang et al., 2002) categorized human face detection techniques in four categories. (i) Template matching (ii) Appearance based (iii) Knowledge based (iv) Feature invariant. Lu and other researchers (Lu J. et al., 2007) used equal neural network for human face detection. Zhao and other researchers (Zhao et al., 2008) proposed Linear Discriminant Analysis (LDA) for face detection.
Vijay Lakshmi et al., proposed a segmentation algorithm for various human face detection in color images with skin color tone areas using edge detection and color models methods, in which several color space models, especially, YCbCr and HSI along with Canny and Prewitt edge detection techniques are used for better face detection (Vijay Lakshmi et al., 2010). Iyad Aldasouqi et.al, offered a Smart Human Face Detection System, where they clarified digital image processing and fast detecting algorithms based on HSV Color model without sacrificing the speed of face detection (Iyad Aldasouqi & Mahmoud Hassan, 2011). Ming-Hsuan Yang et.al, conducted a survey of Detecting faces in images and also compared different techniques such as object recognition, machine learning, face recognition, face detection, statistical pattern recognition, and view based recognition etc. (Yang et al., 2002).
Diedrick and other researchers, worked on Human Face Detection Using Eigen image Template Matching, and Thresholding Color, compared and used the YCbCr, HSV color model for face detection and segment an image based on Skin color (Diedrick Marius, Sumita Pennathur, & Klint Rose). Arti Khaparde et.al, proposed an algorithm based on color segmentation and morphological operation like closing, opening and connecting etc and applied segmentation on HIS and YCbCr color model (Arti Khaparde, Sowmya Reddy, & Y, Swetha Ravipudi, 2010). Amol Dabholkar et.al, worked on Human Face Detection and Tracking, in which the features using
11
motion based and model-based algorithms were extracted and also used Kalman filter for estimation of the feature motion. Aamer et.al, presented a Face Detection using Skin Color Tone in Image by Neural Networks, in which the method relies on using skin color features extracted from two dimensional Discrete Cosine Transfer (DCT) and neural networks, that can be used to detect human faces by using skin color from DCT coefficient of Cb and Cr feature vectors (Aamer et al., 2007). Smita Tripathi and other researchers, proposed a Human Face Detection by using Combination between Skin Color Tone and Template Matching Technique in which, they discussed Skin Detection and the help of YCbCr color model and edge detection techniques to face detection (Smita Tripathi, Varsha Sharma, & Sanjeev Sharma, 2011). Jean Paul Niyoyita et.al, described the Face Detection method with Skin Color Information and also implemented Adaptive boost algorithms and six segmented rectangular filters for better performance (Jean Paul Niyoyita, Tang Zhao Hui, & Liu Jin Ping, 2009).
Sanjay Kr. Singh et.al compared the different color spaces such as HSV, YCbCr, and HSI, merged these color spaces and developed a new Algorithm for Face Detection using Skin Tone which gives improved accuracy compared to other techniques (Sanjay et al., 2003). Michael Padilla et.al, presented Human Face Detection Using Color Segmentation and Energy/Template Thresholding, also they compared the RGB, HSV and YCbCr Color model techniques for face detection and also applied some morphological operation for better accuracy (Michael Padilla & Zihong Fan Group 16, 2002). Sandeep et.al focused on Face Detection in Color Images that Cluttered Using Edge Information and Skin Color, in which they developed HSV color model based face detection using edge detection technique (K. Sandeep & A.N.
Rajagopalan, 2002). Chiunhsiun Lin et.al, proposed a an algorithm in which they could detect the multiple faces in color images with various illumination and
12
discussed triangle based segmentation and multilayer neural network using HSV Color space (Chiunhsiun Lin & Ching-Hung Su, 2007). Venu Shah and other researchers compared and combined the HSV Histogram Equalization with Adaptive HSV segmentation and Kekre Transform of Content using Image Retrieval which is called the Traditional Method (Venu et al., 2011).
Hamid and other researchers presented a Color Image Segmentation using Bayesian Theorem with Kernel Density Estimation in which they used eventuality density function using Bayes theorem to color segmentation and also used HSL color model for face detection (Hamid et al., 2009). Vladimir Vezhnevets et.al conducted a Study on Pixel using Skin Color Detection Methods, in which they accomplished that skin color is very helpful and cue for human face detection, tracking and locating and also compared the different methods and techniques of face detection (Vladimir et al., 2003).
2.2 FACE DETECTION IN IMAGES
The current methods for human face recognition in images are broadly categorized in the following Four methods.
2.2.1 Knowledge Based Method
It depends entirely on the simple’s basics that characterize the countenance of the human face and the connection (for example distance) among them like there exist a regular density in the middle area of the human face and also eyes are symmetric to each other. The search algorithms guided by the rules are applied to find the target face (Yang et al., 2001). The other features may be nose, eyebrows and chin. An approach can be used that examines the human face at different decisions. At the
