Printer Friendly

Morphological based skin segmentation using computer graphics color model.

INTRODUCTION

Human face processing is currently an active research area in the computer vision fiels. Skin detection is a prerequisite for tracking the human body for analyzing and further recognition. The applications of skin detection is ranging from face detection, tracking body parts, hand gesture analysis, to retrieval and blocking objectionable content (Jie Yang, 2005).

Skin detection methods are generally divided into pixel-based and region-based techniques. The research on the pixel-based methods is on the hotspots for skin detection, which classifies each pixel either as skin or non-skin. Each pixel in the image is considered to be an individual unit (Jones M.J, et al., 2002). Pixel-based skin recognition is considered as one of the finest models and under normal conditions those methods produce high level of accuracy, at the detection phase of the process. Due to its high applicability and efficiency, some color models are used extensively in the arena of skin detection (Shah Mostafa Khaled, 2009). It could be classified as parametric (W. Tian, 2004), statistical (Hsu R.-L, 2002) and miscellaneous skin clustering methods (Kakumanu P, et al., 2007).

Several methods have already been proposed to solve the skin detection problem (Rehanullah Khan, 2012). However, most of these methods suffer from accuracy and reliability problems when they are applied to a variety of images obtained under different conditions. Furthermore, the most state-of-the-art methods incorporate one or more thresholds, and it is difficult to determine accurate threshold settings to obtain desirable performance. These problems arise mostly because the available training data for skin detection are imprecise and incomplete, which leads to uncertainty in classification (Mohammad Shoyaib, et al., 2012) (Wei Ren Tan, et al., 2012).

Related work:

Skin detection is a technique used in most of the face detectors to find faces in an images or videos. However, there is no common opinion about which color space is best choice to do the task. Many face detection algorithms based on skin color features have been proposed. (Lu J.W,Gu Q, et al., 2012). Prattpresented the color model YHS, which consists of the brightness as linear combination of the RGB values, the hue as actual angle in the color cube and saturation as relative distance from the body diagonal to the surface of the color cube. It satisfies all three demands and makes some color manipulations easier (Pratt W. K, et al., 1991). Suk Tomas et al. proposed New YHS color coordinate systems. They test with the multichannel remote sensing images are encouraging (Suk Tomas Simberova Stanislava, et al., 2000). M. M. Aznaveh et al proposed an improved skin detection method for images with different illumination conditions and complex background (M.M. Aznaveh, et al., 2009). Jose M Chaves - Gonzales Miguel et al proposed a new skin detection technique for face recognition system. They determined the best color space among ten color spaces with fifteen truth images (Jose M Chaves, et al., 2009). Jie Yang, et al. proposed Skincolor modeling and adaptation. The analysis of color histograms has been a key tool in applying physics-based models to computer vision(Jie Yang , et al., 2005)

Wei-Min Zheng implemented a novel skin clustering method using GLHS to overcome the illumination sensitivity problem, but they implemented the algorithm that can be used for images with simple background. However, it is found in experiments that false detections occur in the case of extremely high illumination when continuous reflecting areas appear on the skin surface (Wei-Min Zheng, et al., 2006). Mohammadreza Ramezanpour et al (Mohammadreza Ramezanpour, et al., 2010) proposed a new method for face and eye detection in color images. The face region is extracted from the image by skin color information using GLHS color space and the image is projected horizontally to estimate the region of eye. Zhi-Wei Zhang proposed a modified skin color model using GLHS with low computational complexity and high detection efficiency for images with different illumination conditions and complex background images(ZhiWei Zhang, et al., 2014).

MATERIALS AND METHODS

Human vision:

Vision is very natural for human and many animals. Visual perception is a function of our eyes and brain. We see images as a whole rather than in parts. However, images can be broken down into their visual elements: line, shape, texture, and color. Perception is the factor defines everything around us which we perceive. Reality is irrelevant in some of the cases and it is only perception that prevails. Only the gap between reality and perceptions needs to be bridged. Light is the first, fundamental element of vision. Some limitations of the human vision system (HVS) in terms of contrast perception and ambient conditions are described so that experimental design and optimum image viewing conditions can be appreciated.

RGB Color Space:

RGB color space is usually accepted as nonappropriate for color image processing applications due to the three major drawbacks are: a) it is device dependent and excludes some visible colors, b) Perceptually non-uniform and distance metric will not represent the real differences between colors, c) difficult to find thresholds for RGB cube(Ravi Subban,et al., 2014). The RGB pixel measurement, P, is described by the product of illumination, surface reectance, and camera sensitivities determined by

P ([lambda]) = E ([lambda]) S ([lambda]) (1)

Further, the observer perceives color in terms of three color signals based on the trichromacy theory can be modeled by:

R = [[integral].sub.[lambda]] E ([lambda]) S ([lambda]) [f.sub.R] ([lambda]) d [lambda] (2)

G = [[integral].sub.[lambda]] E ([lambda]) S ([lambda]) [f.sub.G] ([lambda]) d [lambda] (3)

B = [[integral].sub.[lambda]] E([lambda]) S([lambda]) [f.sub.B]([lambda]) d [lambda] (4)

where the tristimulus values are obtained by adding the product of the SPD of the light

source E ([lambda]) ,the reflectance(or transmittance)factor of the object S ([lambda]) and the color matching functions [f.sub.C] ([lambda]) for C [member of] {R, G, B} of the observer (eye or camera)at each wavelength of the visible spectrum.

Computer Graphics Color Space:

In computer graphics, there is a need to specify colors in a way that is compatible with the hardware and user oriented requirements respectively. There are many color spaces, some are designed for computer graphics, such as YIQ, HSV, HIS and GLHS (generalized LHS) (AO Linmi, et al., 2001).

YIQ System:

The National Television Systems Committee (NTSC) developed the three color attributes Y, I and Q for transmission efficiency. The tristimulus value Y corresponds to the luminance of a color. I and Q correspond closely the hue and saturation of a color. The conversion matrix to compute the YIQ values from the original RGB NTSC tristimulus values is given by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

HSV Color System:

The Hue Saturation Value (HSV) color space, the conversion from RGB color space is obtained as follows (Muthukumar Subramanyam, et al., 2013):

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)

V = Max (7)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (8)

HSI Color System:

Hue represents the fundamental or dominant color. Saturation (S) represents the amount of a color present. The intensity (I) represents the overall brightness or the amount of light. It is independent of color and is a linear value. The HSI definition is given in:

cos[theta] = (R - G) + (R - B)/2[[[(R - G).sup.2] + (R - B) (G - B)].sup.1/2] (9)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (10)

S = 1 - 3.min (R, G, B)/R + G + B (11)

I = 1/3 (R + G + B) (12)

The plane I=1 (for normalized R, G, B values) defines the hue triangle (with vertices at the extremities of the Red, Green and Blue axes) from which these equations have been derived.

GLHS Color Space:

The generalized LHS (GLHS) model proposed by Levkowitz and Herman (1993) calculates the normalized saturation for a lightness. GLHS model provides the closest approximation to a uniform color space(H. Levkowitz, et al., 1993). It is proved that the RGB color space is not suitable for constructing an effective skin color model due to the high correlation among the three components. To improve the detection performances of skin color models, this paper adopts the GLHS color space (AO Linmi, et al., 2001) that has a nonlinear relationship with the RGB space. The lightness normalized saturation for a lightness of the form. Wt1, Wt2 and Wt3 are weights when Wt1>0 and Wt1+Wt2+Wt3=1. The HSL and HSV models are obtained from the GLHS model which enables the realization of different models as special cases by specifying the values of the weights. In addition, the GLHS model provides the potential for dynamic color model changes via interaction with the weights, given sufficient computational power (W. Tian, et al, 2004).

Morphological Based Skin Detection:

The skin areas are segmented after applying threshold values, the non skin pixel is considered as block colour. Morphological operations are applied to the extracted skin regions to eliminate possible non-face skin regions. Theuse of morphological operations to refine the skin regions extracted from the segmentation step. It fragmented sub-regions can be easily grouped together by applying simple dilation on the large regions. Hole and gaps within each region can also be closed by a flood fill operation. The problem of occlusion often occurs in the detection of faces in large groups of people. Even faces of close proximity may result in the detection of one single region due to the nature of pixel-based methods.

[FIGURE 1 OMITTED]

[TABLE 1 OMITTED]

[TABLE 2 OMITTED]

[TABLE 3 OMITTED]

RESULT AND DISCUSSION

The proposed method supports various kinds of images such as images with spectacles, different hair style, different background, and illumination conditions. It produces better results for photographic images with single and multiple faces under different environments. Table I Shows upshot for single images.The Table II represents the photographic image with single face results obtained with different conditions. The multiple faces under various environment is represented in table III. It shows the upshot of under the blur image, black, white, dark backgrounds and how to support different illuminations conditions and skin like color in practical manner. The GLHS is very effective than all other computer graphics color spaces such as YIQ, HSV, HIS.

Conclusion:

This paper proposes a new skin color model to detect skin pixels. Through studying the distribution of the skin pixels in different color spaces and the features of hue and saturation. The GLHS color model is improved and then novel criteria is proposed to check skin pixels. Since this model can overcome the illumination sensitivity problem.GLHS model provides the potential for dynamic color model changes via interaction with the weights, given sufficient computational power. It focus on combining local information of images with skin color models to achieve better detection results for image with more complex background.

ARTICLE INFO

Article history:

Received 23 July 2015

Accepted 28 August 2015

Available online 25 September 2015

ACKNOWLEDGMENT

This work is supported by the Department of Computer Science, Pondicherry University, Puducherry, India and funded by the University Grant Commission (UGC) of India under Major Research Project (F.No. 41-648-2012 (SR) dated 1607-2012) to the department of Computer Science of Pondicherry University, Pondicherry, India.

REFERENCES

AO Linmi and XU Guangyo, 2001. "Color in machine vision and its application" Chinese Science Bulletin Vol. 46 No. 17

Aznaveh, M.M., H. Mirzaei, E. Roshan, M.H. Saraee, 2009. "A New and Improved Skin Detection Method Using Mixed Color Space", Springer Advances in Intelligent and Soft Computing, 60: 471-480.

Hsu, R.L., M. Abdel-Mottaleb, A.K. Jain, 2002. "Face detection in color images", IEEE. Trans. Pattern. Anal. Mach. Intel, 24(5): 696-706.

Jie Yang, Lu. Weier , Alex Waibel, 2005. "Skin-color modeling and adaptation", Lecture Notes in Computer Science, 1352: 687-694.

Jones M.J., J.M. Rehg, 2002. "Statistical color models with application to skin detection", Int. Journal of Computer Vision, 46(1): 81-96.

Jose M Chaves--Gonzales Miguel, A. Vega Rodriguez, Juan A. Gomez--puildo , 2009. "Detecting skin in face Recognition system : A color space study", Digital signal processing, Doi: 10.1016,/j.dsp.2009.10.008, Elsiever.

Kakumanu, P., S. Makrogiannis, N. Bourbakis, 2007. "A survey of skin-color modeling and detection methods", PR, 40: 1106-1122.

Levkowitz, H., G. Herman and T.,1993. "GLHS, A generalized lightness hue and saturation color model, Graphical Models and Image Processing", 55(4): 271-285.

Lu, J.W., Q. Gu, K.N. Plataniotis, J. Wang, 2012. "A comparative study of skin-color models", The International Conference of Image Analysis and recognition (ICIAR), pp: 729-736.

Mohammad Shoyaib, M. Abdullah-Al-Wadud, Oksam Chae, 2012. "A skin detection approach based on the Dempster-Shafer theory of evidence", International Journal of Approximate Reasonin, 53(4): 636-659.

Mohammadreza Ramezanpour, Mohamad Ali Azimi and Mohammad Rahmati, 2010. "A New Method for Eye Detection in Color Images," Journal of Advances in Computer Research, 2: 55-61.

Muthukumar Subramanyam, Krishnan Nallaperumal, Ravi Subban, Pasupathi Perumalsamy, Shashikala Durairaj, S. Gayathri Devi, S.Selva Kumar, "An Analysis of Suitable Color Space for Visually Plausible Shadow-free Scene Reconstruction from Single Image", IEEEICCIC, 2013.

Pratt, W.K., 1991. "Digital image processing", Wiley, New York.

Ravi Subban, Pasupathi Perumalsamy, Muthukumar Subramnian, Shasikala Durairaj, 2014, "A Review on Face detection Approaches", CNC-Elsevier.

Rehanullah Khan, Allan Hanbury Abdul Bais, Julian Stottinger, Abdul Bais, 2012. "Color based skin classification", Volume 33, Issue 2, 15 Pages 157-163, Pattern Recognition Letters, Elsiever.

Shah Mostafa Khaled, Md. Saiful Islam, Md. Golam Rabbani, Mirza Rehenuma Tabassum, Alim UlGias, Md. Mostafa Kamal, Hossain Muhammad Muctadir, Asif Khan Shakir, Asif Imran, and Saiful Islam, 2009. "Combinatorial Color Space Models for Skin Detection in Sub-continental Human Images", IVIC 2009, LNCS 5857, pp: 532-542, 2009. [C] Springer-Verlag Berlin Heidelberg.

Suk Tomas Simberova Stanislava, 2000. " New YHS color coordinate system and its spplication in remote sensing", Proceedings of the 1st Pattern Recognition for Remote Sensing Workshop, pp: 6-9, Eds: Petrou, M., BMVA Press, (Surrey 2000), PRRS 2000 /1./, (Andorra La Vella, AD, 01.09.2000

Tian, W. and Z.Q. Zhuang, 2004. "Self-adaptive skin color detection based on HSV color space", Computer Engineering and Application, 40(7): 8185.

Tian, W. and Z.Q. Zhuang, 2004. "Self-adaptive skin color detection based on HSV color space", Computer Engineering and Application, 40(7): 8185.

Wei Ren Tan, Chee Seng Chan, Pratheepan Yogarajah, and Joan Condell, 2012. "A Fusion Approach for Efficient Human Skin Detection", IEEE Transactions On Industrial Informatics, Vol. 8, No. 1.

Wei-Min Zheng, Lu. Zhe-Ming, Xu. Xiu-Na, 2006. "A Novel Skin Clustering Methods for Face Detection", ISSN 0-7695-2616/06,ICCIC 06, IEEE.

Zhi-Wei Zhang, Ming-Hui Wang, Zhe Ming L and Yong Zhang, 2014. "A Skin Color Model Based on Modified GLHS Space for Face Detection", International Journal of Information Hiding and Multimedia Signal Processing, ISSN 2073-4212, Vol 5.

(1) Pasupathi Perumalsamy, (2) Ravi Subban and (3) Muthukumar Subramanyam

(1) Centre for Information Technology and Engineering, MS University, Tiruneiveii, Tamil Nadu, India

(2) Department of Computer Science, School of Engineering and Technology, Pondicherry University, Pondicherry, India

(3) Indian Institute of Information Technology, Tamil Nadu, India

Corresponding Author: Pasupathi Perumalsamy, pp.cite.msu@gmail.com, Dr. Ravi Subban, sravicite@gmail.com, Dr. Muthukumar Subramanyam, sm.cite.msu@gmail.com
COPYRIGHT 2015 American-Eurasian Network for Scientific Information
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2015 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Perumalsamy, Pasupathi; Subban, Ravi; Subramanyam, Muthukumar
Publication:Advances in Natural and Applied Sciences
Article Type:Report
Date:Aug 1, 2015
Words:2514
Previous Article:Development of Super CAPTCHA for secured web transaction.
Next Article:Low energy adaptive clustering hierarchy--redundancy aware protocol (LEACH-RA).
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |