Journal of ISOSS 2016 Vol. 2(1), 85-95 PERFORMANCE ANALYSIS OF SKIN DETECTION USING GIE AND WITHOUT GIE TECHNIQUES Arfa Hassan 1, Unsa Tariq 2, Asim Iqbal 3 and Muhammad Adnan Khan 4 Department of Computer Science, Lahore Garrison University, Lahore, Pakistan Email: 1 arfach711@gmail.com 2 unsatariq23@gmail.com 3 malikasimiqbalawan @gmail.com, 4 madnankhan@lgu.edu.pk ABSTRACT The Skin detection is a computerize method of locating the Skin in the digital image. It is an important challenge to locate skin from uncontrolled and indistinguishable background of the digital image. This paper presents comparison between GIE and without GIE based human skin detection from the colored images. Skin color segmentation is used for localizations of skin colored components in the digital image. After seeing the result of skin detection with and without GIE based methods, we can conclude that both techniques give us the approximate same results in skin detection but in HSV domain without GIE based method gives better results as compared to GIE based method. But it s not true for all cases; it depends upon the quality of acquired images. 1. INTRODUCTION Skin detection is considered very important skill in the field of digital image processing. Skin detection provides the foundation of many applications in different fields of science [1]. Skin detection is used in electronics and robotic based applications. In medical field, skin detection also plays an important role [1][2]. Through skin detection we can diagnose the different types of skin diseases [2]. The researchers know the importance of this topic very well, so they do a lot of work on this topic in past few years[3][4]. We are unable to discuss all work which is done in the past few years in this article [5]. 2. LITERATURE REVIEW In others work article their main work is to find out skin regions in an image. They use HSV & YCbCr techniques & is based on threshold value of every component of corresponding color space. Converting of color images from RGB to HSV is time consuming process. Cartesian coordinate system is transformed into Polar coordinate system. In YCbCr color space, separation & color transform is easy as compared to HSI & HSV plus this is more effective. So they concluded that YCbCr color space can be applied for complex color images with uneven illumination [6]. Some researchers concern was on RGB, YCbCr & HSI. They compared the algorithms on color spaces & 2016 Journal of ISOSS 85
86 Performance Analysis of Skin Detection using GIE and without GIE Techniques then combined them to get high accuracy results. YCbCr & HIS are efficient than RGB to get skin region. They concluded that both didn t gave good results. It was efficient in classifying skin color region & face region. Accuracy of proposed algorithm was 95.18% [7]. Other people also worked on this and used techniques using skin color information in visual spectrum for 2D images. In their paper they reviewed critical issues for skin detection. Choice of appropriate color space, skin-color modelling & classification techniques, color constancy & dynamic adaptation approaches. For improving accuracy of classifiers other than features such as shape, spatial & motion information can be used. Building accurate classifier which can detect all skin types is still an unsolved problem [8]. Others use CBIR (Color Based Image Retrieval) technique. By this technique, Image tiling and the relation between pixel & neighbors, feature vector will be defined then in training step, skin would be detected. They concluded that their method can be used in many visual multi class classification cases. The approach to image background can be decreased for few types of skin, also this issue can be solved by training step improvement. Their propose benefits were: Low time complexity, Low computational complexity, high accuracy of all kinds of skin, capabilities of skin type s classification [9]. Other people showed their concern on this and uses TSL color space, initial filter. They concluded that better results are gained when calibration is done. High observation level can manage the problem [10]. 3. ALGORITHM In this article we use Hue- Saturation Value (HSV) algorithms with two different techniques for skin detection. We also compare the result of both techniques to find out that which one is going to be the best. Skin detection by using without GIE based method. Skin detection by using GIE based method. 3.1 Skin Detection by using without GIE based Method On the very first step, we take a RGB image and plot the histogram. We also find the probability and PDF of gray scale image. There are four basic steps of skin detection by using without GIE based method. These are followings: 1. Acquire the image 2. Find HSV. 3. Do morphological operations on it. 4. Skin detection. 3.1.1 Acquire the Image On a very first step we take a RGB (Red, Green, and Blue) image [8]. Now we plot the histogram, PDF and Probability of acquire image and then convert into HSV.
Arfa Hassan et al. 87 Fig. 1: Original Image Fig. 2: Histogram, Probability and PDF of Gray Levels 3.1.2 Find HSV Component Hue Saturation value (HSV) in the color space is represented in cylindrical form. Because of the reddish shaded quality of the human face and skin detection, HSV is more reliable [9]. Now we set the HSV value to find out the Hue component of this image. HSV values that are used in this article are (HSV(:,:,1) > = 1.50 HSV(:,:,1) < = 0.10) & (HSV(:,:,2) > =.08 & HSV(:,:,2) < = 0.60) & (HSV (:,:,3) > =.08 HSV (:,:,3) < = 0.28)
88 Performance Analysis of Skin Detection using GIE and without GIE Techniques Fig. 3: HSV of Original Image Fig. 4: HSV Components of Original Image 3.1.3. Morphological Operations Now we perform Morphological Operations on this image by using disk shape. The morphological operations are: Morphological opening Morphological closing Fig. 5: Morphological Operations
Arfa Hassan et al. 89 3.1.4 Skin Detection Now we can detect a skin. We detect a skin of Red, green and blue image separately. Fig. 6: Skin Detection in RGB Components Then at a final step we add the Red, Green and Blue images to find the final image of skin detection. Fig. 7: Skin Detection using without GIE Method 3.2 Skin Detection by using GIE based Method There are five basic steps of skin detection by using GIE based method [8]. These are followings: 1. Acquire the image 2. Perform equalization 3. Find HSV of equalized image 4. Use morphological operations 5. Skin Detection
90 Performance Analysis of Skin Detection using GIE and without GIE Techniques 3.2.1 Acquire the Image In first step we read the RGB (Red, Green, and Blue) image [8]. After acquiring the image we calculate the histogram, PDF and Probability of gray levels. 3.2.2 Equalization Now we equalized the original acquire image. Then find the histogram, probabilities of gray levels and PDF of equalized image as shown in figure 10. Fig. 7: Original Image Fig. 8: Histogram, Probability and PDF of Original Image
Arfa Hassan et al. 91 Fig. 9: Equalized Histogram Image Fig. 10: Histogram, Probability and PDF of GIE Image 3.2.3 Find the HSV Components Now we find the hue component of equalized image. The vales of hue component can very image to image. In this article we consider the following values: (HSV(:,:,1) > = 1.50 HSV(:,:,1) < = 0.10) & (HSV(:,:,2) > =.08 & HSV (:,:,2) < = 0.60) & (HSV(:,:,3) > =.08 HSV(:,:,3) < = 0.28) Fig. 11: HSV of EH Image
92 Performance Analysis of Skin Detection using GIE and without GIE Techniques 3.2.4 Morphological Operations We can perform morphological opening and closing operation on images by using different types of shapes [13] [14]. In opening and closing operation we combine the same type of pixel by using different types of shapes. The shapes which we can be used to combine pixel are: disk, diamond, arbitrary, ball, line, square etc [15]. But in this article we use disk shape for image with radius 3. 3.2.5 Skin Detection Fig. 12: Morphological Operations After morphological operations we are able to detect the skin as shown in figure 13 and 14. The figure 13 shows that the skin detection in red, green and blue components separately. And the figure 14 shows the resultant image of skin detection using GIE based method Fig. 13: Skin Detection in RBG using GIE
Arfa Hassan et al. 93 Fig. 14: Skin Detection in GIE based method 4. COMMENTS AND CONCLUSION In this article figure 16 and 17 shows the difference between both without GIE based and GIE based method. After seeing the result of skin detection with and without GIE based methods, we can conclude that both techniques give us the approximate same results in skin detection but in HSV domain without GIE based method gives better results as compared to GIE based method. But it s not true for all cases; it depends upon the acquired image quality. Fig. 16: Comparison between without GIE and with GIE based Methods
94 Performance Analysis of Skin Detection using GIE and without GIE Techniques Fig. 17: Skin Detection using GIE and without GIE based methods 5. ACKNOWLEDGEMENT We are thankful to Allah (SWT) for leading us to the path for reaching this stage. This research paper is made possible through the help and support from everyone, including: parents, teachers, family, friends, and in essence, all sentient beings. Especially, please allow us to dedicate our acknowledgment of gratitude toward the following significant advisors and contributors: First and foremost, we are thankful to our worthy supervisor Mr. Muhammad Adnan Khan for his support and guidance during our semester. He always shows us light in the darkness whenever, it is needed. We are thankful to him for keeping his doors open at every hurdle during our research. Second, we would like to thank Mrs. Shazia Saqib for her most support and encouragement. She kindly read our paper and offered invaluable detailed advices on grammar, organization, and the theme of the paper. We are thankful for moral support by Mr. Ifraseab Afzal, Mr. Saqib Aslam Awan, Mr. Umer Farooq, Mr. Muhammad Nadeem Ali, Mr. Haider Sultan, Mr. Habib and Mr. Tahir Allyas, as well as all the other faculty members who have taught us over the past three years of our pursuit of the bachelor degree. Finally, we sincerely thank to our parents, family, and friends, who provide the advice and financial support. The product of this research paper would not be possible without all of them.
Arfa Hassan et al. 95 REFERENCES 1. Li, Z. and Latva-aho, M. (2005). Nonblind and semiblind space-time-frequency multiuser detection for space-time block-coded MC-CDMA. IEEE Transactions on Wireless Communications, 4(4), 1311-1318. 2. Zaqout, I., Zainuddin, R. and Baba, S. (2005). Pixel-based skin color detection technique. Machine Graphics and Vision, 14(1), 61-70. 3. Joshi, C.J. and Mishra, S.K. (2015). Investigating the Possibility of Recognizing the Forgery by Using Spatial & Transform Domain. International Journal of Advance Research in Computer Science and Management Studies. 3(5), 495-503. 4. Kawulok, M., Kawulok, J., Nalepa, J. and Smolka, B. (2014). Self-adaptive algorithm for segmenting skin regions. EURASIP Journal on Advances in Signal Processing, 2014(1), 1-22. 5. Singh, S.K., Chauhan, D.S., Vatsa, M. and Singh, R. (2003). A robust skin color based face detection algorithm. Tamkang Journal of Science and Engineering, 6(4), 227-234. 6. Shaik, K.B., Ganesan, P., Kalist, V., Sathish, B.S. and Jenitha, J.M.M. (2015). Comparative Study of Skin Color Detection and Segmentation in HSV and YCbCr Color Space. Procedia Computer Science, 57, 41-48. 7. Kakumanu, P., Makrogiannis, S. and Bourbakis, N. (2007). A survey of skin-color modeling and detection methods. Pattern Recognition, 40(3), 1106-1122. 8. Fekri-Ershad, S., Saberi, M. and Tajeripour, F. (2012). An Innovative Skin Detection Approach Using Color Based Image Retrieval Technique. The International Journal of Multimedia & Its Applications (IJMA) 4(3), 57-65. 9. Tayal, Y., Lamba, R. and Padhee, S. (2012). Automatic face detection using color based segmentation. International Journal of Scientific and Research Publications, 2(6), 1-7. 10. Kruppa, H., Bauer, M.A. and Schiele, B. (2002). Skin patch detection in real-world images. In Annual Symposium for Pattern Recognition of the DAGM (pp. 109-116). Springer Berlin Heidelberg. 11. Yang, M.H. and Ahuja, N. (1998). Detecting human faces in color images. Proceedings of International Conference on Image Processing (ICIP-98), 1, 127-130. 12. Kawulok, M., Kawulok, J. and Nalepa, J. (2014). Spatial-based skin detection using discriminative skin-presence features. Pattern Recognition Letters, 41, 3-13.