COMPARATIVE ANALYSIS OF SKIN COLOR BASED MODELS FOR FACE DETECTION

Signal & Image Processing : An International Journal (SIPIJ) Vol.4, No.2, April 2013
DOI : 10.5121/sipij.2013.4206 69
COMPARATIVE ANALYSIS OF SKIN COLOR
BASED MODELS FOR FACE DETECTION
Ekta Rewar1
and Saroj Kumar Lenka2
1
Electronics and communication Engineering Department, MITS University,
Lakshmangarh, Sikar
1
ektarewar@yahoo.com
2
Information Technology Department, MITS University, Lakshmangarh, Sikar
2
lenka.sarojkumar@gmail.com
ABSTRACT
Human face detection plays an important role in many applications such as face recognition , human
computer interface, biometrics, area of energy conservation, video surveillance and face image database
management. The selection of accurate color model is the first need of the face detection. In this paper, a
study on the various color models for face detection i.e. RGB, YCbCr, HSV and CIELAB are included. This
paper compares different color models based on the detection rate of skin regions. The results shows that
YCbCr as compared to other color models yields the best output even in varying lightening conditions.
KEYWORDS
Skin color model & face detection.
1. INTRODUCTION
Detecting human faces automatically is becoming a very important task in many applications,
such as content-based indexing video retrieval systems or security access control systems like the
distributed audio visual archives network system [1]. Face detection is the first step in any
automated system that solves problems such as facial expression recognition, face tracking and
face recognition [2]. Up to now much work has been done on detecting and locating faces in color
images and the methods like neural network based [3], level set methodology [4], chrominance
based [5], ada boost based [6], skin color based [7] studied by many researchers. Among many
face detection algorithms skin color model has been widely used because of simple performance,
convenient use and high detection rate.
1.1 Face Detection
Face detection is a visual task which can be done by humans without any effort but in computer
vision this task is very difficult. Given a single image, detect and localize the number of faces
regardless of pose, illumination and expression. Face detection is used for self-serviced
immigration clearance, for person verification, security, image search and research is being done
in the area of energy conservation etc. The goal of face detection is to identify and locate all of
the human faces regardless of their positions, scales, orientations, poses and light conditions and

70
this is a challenging problem because human faces are highly non-rigid with a high degree of
variability of in size, shape, color and texture. A simple face detection block diagram is shown in
fig.1:-
Face Non-face
Figure 1. Basic block diagram of face detection
1.2 Challenges in Face Detection
Detection and recognition of faces are challenging because face has a wide variability in poses,
shapes, sizes and texture. The problems or challenges in face detection are listed as follows [8]:-
• Illumination
• Facial Pose
• Facial Expression
• Age Span
• Motion
• Occlusion
• Image orientation
• Image condition
• Background Complexity
• Environment Changes
• Noise
• Textural differences among Faces
Camera
Data
acquisition
Image pre-
processing
Feature
extraction
Classification

71
These are the most general problems that occur during face detection. We can remove most of the
drawbacks by increasing the efficiency and by applying the loops to get better results.
2. SKIN COLOR BASED ALGORITHM
Color is a useful piece of information for skin detection. It is the first most common approach for
detecting meaningful skin color [9], as it avoids many exhaustive searches for faces in an entire
image. Human skin colours are distinguished from different ethnic through the intensity of the
skin colour not the chromatic features [10].According to the Hyun-Chul, Ju-Yeon You and Sung-
II Chien, skin color algorithm can be used under the various illumination conditions [11] and
according to Yogesh Tayal , Ruchika Lamba and Subhransu Padhee, color image is segmented
into skin and non-skin region for different color spaces and the skin color algorithm can be
applied on a wide variety of the images under different lighting conditions and with different
backgrounds [12]. Many researches have been done using the color models. In this paper four
basic color models i.e. RGB, HSV, YCbCr and CIELAB are reviewed.
2.1. RGB Color Model
RGB Colors are specified in terms of three primary colors i.e. Red (R), Green (G), and Blue (B).
In RGB color space, a normalized color histogram is used to detect the pixels of skin color of an
image and can be further normalized for changes in intensity on dividing by luminance. This
localizes and detects the face [10].
Figure 2. RGB color model
It is the basic color model and all other color models are derived from it. RGB color model is
light sensitive. In comparison with other color models such as YCbCr or HSI it has a major
drawback that it cannot separate clearly the mere color (Chroma) and the intensity of a pixel, so it
is difficult to distinguish skin colored regions [2]. These factors contribute to the less favorable of
RGB. It is widely used color space for processing and storing digital images because of the fact
that chrominance and luminance components are mixed in RGB; it is not widely used in Skin
detection algorithms [13].

72
Input image RGB image
Figure 3. RGB image
2.2. HSV Color Model
HSV Colors are specified in terms of Hue (H), Saturation (S) and Intensity value (V) which are
the three attributes. Those are perceived about color. Hue refers to the color of red, blue and
yellow having the range of 0 to 360. Saturation means purity of the color and takes the value from
0 to 100% whereas Value refers to the brightness of color and provides the achromatic idea of the
color [14]. From this color space, H and S will provide the necessary information about the skin
color. The skin color pixel should satisfy the following condition:-
0 <= H <= 0.25
0.15 <= S <= 0.9
The transformation between HSV and RGB is non-linear [12]. In HSV color model Hue (H) is
not reliable for the discrimination task when the saturation is low. But where color description is
consider, HSV color model is preferred over RGB model according to Yogesh Tayal [13], colors
similarly is described that how human eye tends to perceive color and also describe color using
more familiar comparisons such as vibrancy, color and brightness.
Input image HSV image
Figure 4. HSV image
2.3. YCbCr Color Model
YCbCr Color model is specified in terms of luminance(Y channel) and chrominance (Cb and Cr
channels). It segments the image into a luminous component and chrominance components [12].
In YCbCr color model, the distribution of the skin areas is consistent across different races in the
Cb and Cr color spaces [15]. As RGB color model is light sensitive so to improve the
performance of skin color clustering, YCbCr color model is used. Its chrominance components
are almost independent of luminance and there is non-linear relationship between chrominance

73
(Cb, Cr) and luminance(Y) of the skin color in the high and low luminance region [2]. Range of
Y lies between 16-235 where 16 for black and 235 for white whereas Cb & Cr are scaled in the
range of 16-240 [14]. The main advantage of YCCr color model is that influence of luminosity
can be removed during processing of an image. Different plots for Y, Cb and Cr values for face
and non-face pixels were plotted using the reference images and studied to find the range of Y,
Cb and Cr values for the face pixels.
Input image YCbCr image
Figure 5. YCbCr image
2.4. CIELAB Color Model
In 1976, the CIE (international commission on illumination) recommended the CIEL*a*b* or
CIELAB, color scale for use. It provides a standard, approximately uniform color scale which
could be used by everyone so that the color values can be easily compared. This color model is
designed to approximate perceptually uniform Color spaces (UCSs). It is related to the RGB color
space through a highly nonlinear transformation. Examples of similar color spaces are CIE-Luv
and Farnsworth UCS [12]. It has three axes in it two are color axes and the third is lightness.
L=100
-a +b
-b +a
L=0
Figure 6. CIELAB Color Model
Where L indicates lightness, +a and –a indicates amount of green and red color respectively, +b
and –b indicates amount of yellow and blue color respectively.
Here maximum value of L is 100 which represent a perfect reflecting diffuser (white color) and
the minimum value for L is 0 which represents black color. Axes a and b do not have any specific
numerical value.

74
Input image CIEL*a*b image
Figure 7. CIELAB image
3. RESULTS AND COMPARISON
The results after survey on these four color models are shown below in tabular format:-
Table 1. Comparison of four Color Models
Type of color model Detection rate % False detection %
RGB 65 35
HSV 70 30
CIELAB 89 11
YCbCr 90 10
The table compares face detection rate of the four color models i.e. RGB, HSV, CIELAB and
YCbCr. Detection rate of RGB color model is only 65% that means it can only detect 65% of the
faces correctly. Detection rate of HSV color model is 70%, CIELAB color model is 89% and
detection rate of YCbCr color model is 90%. Out of these four color model YCbCr has highest
detection rate and lowest false detection rate.
To increase the speed, accuracy of the system and to overcome on false face detection rate these
four color models can be combined to get better result, after combining all these models we can
get correct detection rate which can be up to 92%.
4. CONCLUSION
Four different color models in face detection have been studied and compared namely, RGB,
YCbCr, HSV and CIELAB. The paper concludes that out of these four color models YCbCr
yields a better results and performance under varying lightening conditions and changes in
illumination.
REFERENCES
[1] Rein-lien Hsu, Mohamed Abdel-Mottaleb & Anil k. Jain, (2008) “Face Detection in Color Images”,
IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.24, No.5, pp696-706.
[2] Chandrappa D N, M Ravishankar & D R Ramesh Babu, (2011) “Face Detection in Color Images
using Skin Color Model Algorithm Based on Skin Color Information”, IEEE Transactions, pp254-
258.

75
[3] Henry A. Rowley, Shumeet Baluja, & Takeo Kanade, (1988) “Neural Neural Network-Based Face
Detection”, IEEE, pp1-28.
[4] M.Kumaravel, S.Karthik, P.Sivraj & K.P.Soman, (2012) “Human Face Image Segmentation using
Level Set Methodology”, International Journal of Computer Applications, Vol. 44, No.12.
[5] Hongxun Yao & Wen Gao, (2000) “Face detection and Location based on Skin Chrominance and Lip
Chrominance Transformation from Color Images”, Journal of Pattern Recognition Society, pp1555-
1564.
[6] Jing-Ming Guo, Chen-Chi Lin, Min-Feng Wu, Che-Hao Chang & Hua Lee, (2011) “Complexity
Reduced Face Detection Using Probability-Based Face Mask Prefiltering and Pixel-Based
Hierarchical-Feature Adaboosting”, IEEE Signal Processing Letters, Vol.18,No.8, pp447-450.
[7] Jiang Oiang-rong & Li Hua-Ian, (2010) “Robust Human Face Detection in Complicated Color
Images”, IEEE Transactions.
[8] Yang, M.H., Kriegman, D.J. & Ahuja, “Detecting Faces in Images: A Survey”, IEEE Transactions on
Pattern Analysis and Machine Intelligence, Vol.24, No.1, pp34-58.
[9] Devendra Singh Raghuvansi & Dheeraj Agrawal, (2012) “Human Face Detection by Using Skin
Color Segmentation, Face Features and Region Properties”, International Journal of Computer
Application, Vol. 38, No.9, pp14-17.
[10] Nurulhuda Ismail, Mas Idayu & Sabri, (2009) “Review of Existing Algorithms for Face Detection and
Recognition”, Proceedings of the 8th WSEAS International, pp.30-39.
[11] Hyun-Chul Do, Ju-Yeon You & Sung-II Chien , (2007) “Skin Color Detection through Estimation
and Conversion of Illuminant Color under Various Illumination”, IEEE Transactions on Consumer
Electronics, Vol.53, No.3, pp1103-1108.
[12] SonLam Phung, Abdesselam Bouzerdoum & Douglas Chai, (2005) “Skin Segmentation using Color
Pixel Classification: Analysis and Comparison”, IEEE Transaction on Pattern Analysis and Machine
Intelligence, Vol.27, No.1.
[13] Vivek Desai, Pranav Vankar, Jugal Mehta & Ghanshyam Prajapati, (2012) “Face Detection Using
Skin Color”, International Conference on Computing and Control Engineering, ISBN No. 978-1-
4675-2248-9.
[14] S. Chitra & G. Balakrishnan, (2012) “Comparative Study for Two Color Spaces HSCbCr and YCbCr
in Skin Color Detection”, Applied Mathematical Sciences, Vol. 6, No. 85, pp4229 – 4238.
[15] Yogesh Tayal, Ruchika Lamba & Subhransu Padhee, (2012) “Automatic Face Detection using Color
Based Segmentation”, International Journal of Scientific and Research Publications, Vol.2, No.6, pp1-
7.
About Author (s):
Ekta Rewar received her B.Tech degree in Electronics and Communication Engineering
from Jagannath Gupta Institute of Engineering and Technology, Rajasthan Technical
University, India in 2011 and currently is a M. Tech (Signal Processing) student in Mody
Institute of Technology and Science (Deemed University), Rajasthan, India. Her research
interest lies in Signal Processing & Image Processing.
S.K.Lenka received his B.E. from Utkal University in 1994, M.Tech. from Kalinga
University, PhD degree from Behrampur University, India. He is currently Professor and
Head, Department of Information Technology, MITS University, Lakshmangarh,
Rajasthan. His research interest lies in Digital Image Processing, Computer Architecture &
Information Security

COMPARATIVE ANALYSIS OF SKIN COLOR BASED MODELS FOR FACE DETECTION

More Related Content

What's hot (20)

Viewers also liked (19)

Similar to COMPARATIVE ANALYSIS OF SKIN COLOR BASED MODELS FOR FACE DETECTION (20)

Recently uploaded (20)

COMPARATIVE ANALYSIS OF SKIN COLOR BASED MODELS FOR FACE DETECTION