VIEWS: 65 PAGES: 6 CATEGORY: Emerging Technologies POSTED ON: 5/15/2012
International Journal of Computer Science and Information Security (IJCSIS) provide a forum for publishing empirical results relevant to both researchers and practitioners, and also promotes the publication of industry-relevant research, to address the significant gap between research and practice. Being a fully open access scholarly journal, original research works and review articles are published in all areas of the computer science including emerging topics like cloud computing, software development etc. It continues promote insight and understanding of the state of the art and trends in technology. To a large extent, the credit for high quality, visibility and recognition of the journal goes to the editorial board and the technical review committee. Authors are solicited to contribute to the journal by submitting articles that illustrate research results, projects, surveying works and industrial experiences. The topics covered by this journal are diversed. (See monthly Call for Papers) For complete details about IJCSIS archives publications, abstracting/indexing, editorial board and other important information, please refer to IJCSIS homepage. IJCSIS appreciates all the insights and advice from authors/readers and reviewers. Indexed by the following International Agencies and institutions: EI, Scopus, DBLP, DOI, ProQuest, ISI Thomson Reuters. Average acceptance for the period January-March 2012 is 31%. We look forward to receive your valuable papers. If you have further questions please do not hesitate to contact us at firstname.lastname@example.org. Our team is committed to provide a quick and supportive service throughout the publication process. A complete list of journals can be found at: http://sites.google.com/site/ijcsis/ IJCSIS Vol. 10, No. 3, March 2012 Edition ISSN 1947-5500 � IJCSIS, USA & UK.
(IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012 Performance Comparison of Assorted Color Spaces for Multilevel Block Truncation Coding based Face Recognition Dr. H.B. Kekre Dr. Sudeep Thepade Karan Dhamejani, Senior Professor Associate Professor Sanchit Khandelwal, Computer Engineering Department Computer Engineering Department Adnan Azmi MPSTME, SVKM’s NMIMS MPSTME,SVKM’s NMIMS B.Tech Students (Deemed-to-be University) (Deemed-to-be University) Computer Engineering Department Mumbai, India Mumbai, India MPSTME, SVKM’s NMIMS (Deemed-to-be University) Mumbai, India Abstract— The paper presents a performance analysis of A large number of face detection algorithms are derived from Multilevel Block Truncation Coding based Face Recognition algorithmic approach [2, 3, 4, 5, 6, 7, 8, 9, 24] and some image among widely used color spaces. In , Multilevel Block morphological techniques . However most of the works Truncation Coding was applied on the RGB color space up to concentrate on single face detection, with some constrained four levels for face recognition. Better results were obtained environments. In this paper performance comparison of when the proposed technique was implemented using Kekre’s Multilevel Block Truncation Coding  using various color LUV (K’LUV) color space . This was the motivation to test the proposed technique using assorted color spaces. For spaces has been carried out on two face databases. Results experimental analysis, two face databases are used. First one is further revealed that the YIQ color space outperforms all the “Face Database”, developed by Dr.Libor Spacek which has 1000 other color spaces at each stage of Multilevel BTC. face images and the second one is “Our Own Database” which has 1600 face images. The experimental results showed that II. BLOCK TRUNCATION CODING AND MULTILEVEL BLOCK Block Truncation Level 4 (BTC-Level 4) gave the best result in TRUNCATION CODING every color space. It is observed that the proposed technique Block truncation coding (BTC) [11, 12, 13, 14] is a relatively functions better in the YIQ color space. simple image coding technique developed in the early years of Keywords- Face recognition, Block Truncation Coding, RGB, digital imaging more than 29 years ago. Block Truncation K’LUV, YIQ, YUV, YCbCr, YCrgCrb, Multilevel BTC. Coding (BTC) was first developed in 1979 for grayscale image coding . Although it is a simple technique, BTC has I. INTRODUCTION played an important role in the history of digital image coding Face recognition plays an imperative role in identification and in the sense that many advanced coding techniques have been for authentication purpose, in our everyday lives. In real time, developed based on BTC or inspired by the success of BTC. It this identification must be efficient, liable and faster. Face is a straightforward technique which demands very less recognition is preferred over other techniques like fingerprint computational complexity. recognition, iris recognition because it does not require In the proposed technique, Multilevel Block Truncation explicit cooperation from users. Also special equipments are Coding, BTC has been implemented using the RGB color not required to capture the image [21, 22, 23]. It is a computer space up till four levels [1, 13]. The feature vector size at application for automatically identifying or verifying a person BTC-Level 1, BTC-Level 2, BTC-Level 3 and BTC-Level 4 is from a digital image or a video frame from a video source. 6, 12, 24 and 48 respectively. In the same way BTC is implemented on the following color spaces: K’LUV, YUV, Face recognition can be achieved by comparing the input query face image with the existing face images stored in the YCbCr, YIQ and YCgCb. database. It is the fastest growing biometric technology. Some of the applications of face recognition include physical, security and computer access controls, law enforcement [12, 13], criminal list verification, surveillance at various places , forensic, authentication at airports , etc. 58 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012 III. CONSIDERED COLOR SPACES[12,26,27] The reverse conversion, that is from YUV color space to RGB color space is given in Equation (6). A. Kekre’s LUV  K’LUV color space is a special case of Kekre transform. Where L gives luminance and U and V gives chromaticity = . (6) values of color image. Positive value of U indicates prominence of red component in color image and negative value of V indicates prominence of green component. D. YIQ Equation (1) gives the RGB to LUV conversion matrix which indicates the corresponding L, U and V components for an The YIQ color space is derived from YUV colour space. I stands image from the R, G and B components. for in phase and Q for Quadrature. Equation (7) gives the RGB to YIQ conversion matrix which (1) indicates the corresponding Y, I and Q components for an image from the R, G and B components. The reverse conversion, that is from LUV color space to RGB color space is given in (2). = . (7) (2) The reverse conversion, that is from YIQ color space to RGB color space is given in (8). B. YCbCr = . (8) In YCbCr color Space, Y gives luminance and Cb and Cr gives chromaticity values of color image. E. YCgCb Equation (3) gives the RGB to YCbCr conversion matrix To get Y, Cg and Cb components we need the conversion of which indicates the corresponding Y, Cb and Cr components RGB to YCgCb. The RGB to YCgCb conversion matrix is for an image from the R, G and B components. given in (9) gives the Y, Cg, Cb components of color image for respective R, G and B components. = . (3) (9) The reverse conversion, that is from LUV color space to RGB The YCgCb to RGB conversion matrix given in (10) gives the color space is given in (4). R, G, B components of color image for respective Y, Cg and Cb components. = . (4) (10) C. YUV In YUV color space, Y component gives the luminance (brightness) of the color and while U and V components give the chrominance (color). Equation (5) gives the RGB to YUV conversion matrix which indicates the corresponding Y, U and V components for an image from the R, G and B components. = . (5) 59 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012 IV. PROPOSED METHOD To calculate the feature vector of each image in the database 2) Our Own Database [1, 20] set and the query image, Multilevel Block Truncation Coding This database consists of 1600 face images of 160 people (92 has been used for each of the assorted color space. males and 68 females).For each person 10 images are taken. The images in the database are captured under numerous At each level of BTC, the feature vector of the query image is illumination settings. The images are taken with a compared with the feature vector of each image in the training homogenous background with the subjects having different set. The comparison (Similarity measurement) is done by expressions. The images are of variable sizes, unlike the Face Mean Square Error (MSE) given by equation 11. database. The ten poses of Our Own Database are shown in Figure 2. (11) Where, I & I’ are two feature vectors of size M*N which are being compared. False Acceptance Ratio (FAR) and Genuine Acceptance Ratio (GAR) have been used as the performance evaluation parameters to assess the competence of each considered color space. V. IMPLEMENTATION A. Platform The effectuation of the Multilevel BTC is done in MATLAB Figure 2: Sample images from Our Own Database 2010. It is carried out on a computer using an Intel Core i5- 2410M CPU (2.4 GHz). VI. RESULTS AND DISCUSSIONS B. Database False Acceptance Rate (FAR) and Genuine Acceptance Rate (GAR) are standard performance evaluation parameters of The experiments were performed on two face databases. face recognition system. 1) Face Database  The False acceptance rate (FAR) is the measure of the This database is created by Dr Libor consisting of 1000 likelihood that the biometric security system will incorrectly images (each with 180 pixels by 200 pixels), corresponding to accept an access attempt by an unauthorized user. A system’s 100 persons in 10 poses each, including both males and FAR typically is stated as the ratio of the number of false females. All the images are captured against a dark or bright acceptances divided by the number of identification attempts. homogeneous background, little variation of illumination, different facial expressions and details. The subjects sit at FAR = (False Claims Accepted/Total Claims) X 100 fixed distance from the camera and are asked to speak, whilst (12) a sequence of images is taken. The speech is used to introduce facial expression variation. The images were taken in a single The Genuine Acceptance Rate (GAR) is evaluated by session. The ten poses of Face database are shown in Figure 1. subtracting the FAR values from 100. GAR=100-FAR (in percentage) (13) For each color space, 10000 queries (10 images for each of the 1000 people) are fired on face database and 16000 queries (10 images for each of the 1600 people) are fired on Our Own Database. At the end, average FAR and GAR of all queries in respective face databases are considered for performance ranking of BTC levels and of the color spaces. For optimal performance the FAR values must be less and accordingly the GAR values must be high for each successive levels of BTC. Figure 1: Sample images from Face database 60 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012 A. Face Database K'LUV YUV YCbCr RGB YIQ YCgCb To analyze the performance of proposed algorithm and for performance ranking of color spaces, 10000 queries are fired for each of the assorted color space. For every color space, every BTC level; feature vector of the query image is 98.5 calculated and compared with the feature vectors of every Genuine acceptance Ratio image in the database. The FAR and GAR values are calculated by employing equations 12 and 13. 98 K'LUV YUV YCbCr RGB YIQ YCgCb 97.5 0.032 97 0.027 False Acceptance Ratio 96.5 0.022 BTC Level 1 BTC Level 2 BTC Level 3 BTC Level 4 Considered BTC Levels 0.017 Figure 4. GAR values at different BTC levels of the assorted color spaces for 0.012 Face Database 0.007 Figure 4 gives the GAR values of the different BTC levels 0.002 based face recognition techniques tested on face database for BTC Level BTC Level BTC Level BTC Level the assorted color spaces. Here it is observed that with each 1 2 3 4 successive level of BTC the GAR values go on increasing in respective color spaces and hence a BTC-level 4 gives the best Considered BTC levels result with the highest value in all the color spaces. It is also observed that the YIQ color space shows the highest GAR Figure 3. FAR values at different BTC levels of the assorted color spaces for values at all levels of BTC followed by YCbCr, K’LUV, Face Database YUV, YCrgCrb and RGB respectively. Figure 3 gives the FAR values of the different BTC levels based face recognition techniques tested on face database for An anomaly is noticed in YCbCr color space for this database. the considered color spaces. Here it can be seen that the FAR Not conforming to the generally observed pattern, the FAR values go on decreasing for each succeeding level of BTC of values increase at the second level of the BTC based face respective color spaces. This shows that the accuracy of face recognition technique. recognition increases with increasing level of BTC and hence BTC-level 4 gives the best result with the least FAR value in B. Our Own Database all the color spaces. Also the FAR values of YIQ color space In all 16000 queries were tested on the database for analyzing are the least. Thus, it can be concluded that the the performance of the proposed BTC level based face implementation of BTC levels based face recognition recognition algorithm for the assorted color spaces. The techniques is better when applied in YIQ color space. experimental results of proposed face recognition techniques have shown that BTC level 4 gives the best performance in respective color spaces. The efficiency of the Multilevel BTC based face recognition increases with the increasing levels of BTC. 61 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012 Figure 6 gives the GAR values of the different BTC levels K'LUV YUV YCbCr RGB YIQ YCgCb based face recognition techniques tested on Our Own 0.48 Database. It can be seen from the above figure that BTC-Level 4 has the highest GAR values and hence it is better than other 0.43 BTC-Levels. Also the GAR values of YIQ color space are greater than the GAR values of all the other color spaces False Acceptance Ratio considered, at all the levels. Thus, it can be concluded that the 0.38 implementation of BTC levels based face recognition techniques is better when applied in YIQ color space. 0.33 VII. CONCLUSION 0.28 BTC based face recognition using assorted color spaces have been presented in the paper. Earlier the RGB and K’LUV 0.23 color spaces were considered and it was observed that better BTC Level 1 BTC Level 2 BTC Level 3 BTC Level 4 results were shown by the K’LUV color space. In this paper, Considered BTC levels six color spaces have been considered and the proposed technique has been implemented till four levels of BTC. In all 24 combinations have been tested on two databases; Our Own Figure 5. FAR values at different BTC levels of the assorted color spaces for Database (Not normalized, 1600 face images) and Face Our Own Database Database (Normalized, 1000 face images). It is concluded that Figure 5 gives the FAR values of the different BTC levels the YIQ color space at level four of BTC gives the best results based face recognition techniques tested on Our Own followed by the YCbCr color space at BTC level four. Database for all color spaces. The FAR values go on decreasing for each succeeding level of BTC of respective color spaces. Thus, when BTC based face recognition REFERENCES techniques is applied on Our Own Database, it gives a result similar to the Face Database; The BTC level 4 gives the best  H.B.Kekre, Sudeep D. Thepade, Sanchit Khandelwal, Karan Dhamejani, Adnan Azmi, “Face Recognition using Multilevel Block Truncation result for respective color spaces and YIQ color space is better Coding” International Journal of Computer Applications (IJCA) than other color spaces for implementing this proposed December 2011 Edition. algorithm.  Xiujuan Li, Jie Ma and Shutao Li 2007. A novel faces recognition method based on Principal Component Analysis and Kernel Partial Least. IEEE International Conference on Robotics and Biometrics, 2007. ROBIO 2007 K'LUV YUV YCbCr RGB YIQ YCgCb  Shermin J “Illumination invariant face recognition using Discrete Cosine Transform and Principal Component Analysis” 2011 International Conference on Emerging Trends in Electrical and Computer Technology (ICETECT). 70  Zhao Lihong , Guo Zikui “Face Recognition Method Based on Adaptively Weighted Block-Two Dimensional Principal Component Genuine Acceptance Ratio Analysis”; 2011 Third International Conference on Computational Intelligence, Communication Systems and Networks (CICSyN) 65  Gomathi, E, Baskaran, K. “Recognition of Faces Using Improved Principal Component Analysis”; 2010 Second International Conference on Machine Learning and Computing (ICMLC) 60  Haitao Zhao, Pong Chi Yuen” Incremental Linear Discriminant Analysis for Face Recognition”, IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics  Tae-Kyun Kim; Kittler, J. “Locally linear discriminant analysis for 55 multimodally distributed classes for face recognition with a single model image” IEEE Transactions on Pattern Analysis and Machine Intelligence, March 2005 50  James, E.A.K., Annadurai, S. “Implementation of incremental linear discriminant analysis using singular value decomposition for face BTC Level 1 BTC Level 2 BTC Level 3 BTC Level 4 recognition”. First International Conference on Advanced Computing, 2009. ICAC 2009 Considered BTC Levels  Zhao Lihong, Wang Ye, Teng Hongfeng; “Face recognition based on independent component analysis”, 2011 Chinese Control and Decision Figure 6. GAR values at different BTC levels of the assorted color spaces for Conference (CCDC) Our Own Database 62 http://sites.google.com/site/ijcsis/ ISSN 1947-5500 (IJCSIS) International Journal of Computer Science and Information Security, Vol. 10, No. 3, 2012  Yunxia Li, Changyuan Fan; “Face Recognition by Non negative AUTHORS PROFILE Independent Component Analysis” Fifth International Conference on Dr. H. B. Kekre has received B.E. (Hons.) in Telecomm. Natural Computation, 2009. ICNC'09’. Engineering. from Jabalpur University in 1958, M.Tech (Industrial  Yanchuan Huang, Mingchu Li, Chuang Lin and Linlin Tian. “Gabor- Electronics) from IIT Bombay in 1960, M.S.Engg. (Electrical Engg.) Based Kernel Independent Component Analysis on Intelligent Information Hiding and Multimedia Signal Processing (IIH-MSP). from University of Ottawa in 1965 and Ph.D. (System Identification) from IIT Bombay in 1970 He has worked as Faculty of Electrical  H.B.Kekre, Sudeep D. Thepade, Varun Lodha, Pooja Luthra, Ajoy Joseph, Chitrangada Nemani “Augmentation of Block Truncation Engg. and then HOD Computer Science and Engg. at IIT Bombay. Coding based Image Retrieval by using Even and Odd Images with For 13 years he was working as a professor and head in the Sundry Color Spaces” ,Int. Journal on Computer Science and Engg. Vol. Department of Computer Engg. at Thadomal Shahani Engineering. 02, No. 08, 2010, 2535-2544 College, Mumbai. Now he is Senior Professor at MPSTME,  H.B.Kekre, Sudeep D. Thepade, Shrikant P. Sanas, “Improved CBIR SVKM‟s NMIMS University. He has guided 17 Ph.Ds, more than using Multileveled Block Truncation Coding” ,International Journal on 100 M.E./M.Tech and several B.E./B.Tech projects. His areas of Computer Science and Engineering Vol. 02, No. 08, 2010, 2535-2544 interest are Digital Signal processing, Image Processing and  H.B.Kekre, Sudeep D. Thepade, “Boosting Block Truncation Coding Computer Networking. He has more than 350 papers in National / using Kekre’s LUV Color Space for Image Retrieval”, WASET International Conferences and Journals to his credit. He was Senior International Journal of Electrical, Computer and System Engineering Member of IEEE. Presently He is Fellow of IETE and Life Member (IJECSE), Volume 2, Number 3, pp. 172-180, Summer 2008. of ISTE Recently ten students working under his guidance have  H.B.Kekre, Sudeep D. Thepade, “Image Retrieval using Augmented received best paper awards and two have been conferred Ph.D. Block Truncation Coding Techniques”, ACM International Conference on Advances in Computing, Communication and Control (ICAC3- degree of SVKM‟sNMIMS University. Currently 10 research 2009), pp. 384-390, 23-24 Jan 2009, Fr. Conceicao Rodrigous College scholars are pursuing Ph.D. program under his guidance. of Engg., Mumbai  Developed by Dr. Libor Spacek. Available Online at: Dr. Sudeep D. Thepade has Received B.E.(Computer) degree from http://cswww.essex.ac.uk/mv/otherprojects.html North Maharashtra University with Distinction in 2003, M.E. in  Mark D. Fairchild, “Color Appearance Models”, 2nd Edition, Wiley- Computer Engineering from University of Mumbai in 2008 with IS&T, Chichester, UK, 2005. ISBN 0-470-01216-1 Distinction, Ph.D. from SVKM‟s NMIMS (Deemed to be University)  Rafael C. Gonzalez and Richard Eugene Woods “Digital Image in July 2011, Mumbai. He has more than 08 years of experience in Processing”, 3rd edition, Prentice Hall, Upper Saddle River, NJ, 2008. teaching and industry. He was Lecturer in Dept. of Information ISBN 0-13-168728-X. pp. 407–413.S Technology at Thadomal Shahani Engineering College, Bandra(W),  Dr.H.B.Kekre, Sudeep D. Thepade and Shrikant P. Sanas, “Improved Mumbai for nearly 04 years. Currently working as Associate CBIR using Multileveled Block Truncation Coding”, (IJCSE) Professor in Computer Engineering at Mukesh Patel School of International Journal on Computer Science and Engineering Vol. 02, No. Technology Management and Engineering, SVKM‟s NMIMS 07, 2010, 2471-2476 (Deemed to be University), Vile Parle (W), Mumbai, INDIA. He is  Dr. H.B.Kekre , Sudeep D. Thepade and Akshay Maloo, “Face member of International Advisory Committee for many International Recognition using Texture Feartures Extracted from Walshlet Pyramid Conferences, acting as reviewer for many referred international ”, Int. J. on Recent Trends in Engineering & Technology, Vol. 05, No. 01, Mar 2011. journals/transactionsincluding IEEE and IET. His areas of interest are Image Processing and Biometric Identification. He has guided five  Koji kotani, Chen Qiu and Tadahiro Ohmi, “Face Recognition Using Vector Quantization Histogram Method”. International Conference on M.Tech. projects and several B.Tech projects. He has more than 130 Image Processing,Volume 2, pp.105-108,2002. papers in National/International Conferences/Journals to his credit  Shang-Hung Lin, “An Introduction to Face Recognition Technology”, with a Best Paper Award at International Conference SSPCCIN- Informing Science Special Issue on Multimedia Informing 2008, Second Best Paper Award at ThinkQuest-2009, Second Best Technologies- Part 2, Volume 3 No.1, 2000. Research Project Award at Manshodhan 2010, Best Paper Award for  H. B. Kekre, Sudeep Thepade, Akshay Maloo, “Eigenvectors of paper published in June 2011 issue of International Journal IJCSIS Covariance Matrix using Row Mean and Column Mean Sequences for (USA), Editor‟s Choice Awards for papers published in International Face Recognition”, CSC-International Journal of Biometrics and Journal IJCA (USA) in 2010 and 2011. Bioinformatics (IJBB), Volume (4): Issue (2), pp. 42-50, May 2010.  H. C. Vijaya Lakshmi, D. Patil Kulakarni “Segmentation algorithm for Sanchit Khandelwal is currently pursuing B.Tech. (CE) from multiple face detection in color images with skin tone regions using MPSTME, NMIMS University, Mumbai. His areas of interest are color spaces and edge detection techniques,” International journal of Image Processing and Computer Networks and security. He has 2 computer theory and engineering 1793-8201,2010. paper in an international journal to his credit.  Dr. H. B. Kekre, Sudeep Thepade, Karan Dhamejani, Adnan Azmi, Sanchit Khandelwal, “Improved Face Recognition with Multilevel BTC using Kekre’s LUV Color Space”, IJACSA Karan Dhamejani is currently pursuing B.Tech. (CE) from MPSTME, NMIMS University, Mumbai. His areas of interest are  Dr. H. B. Kekre, Sudeep Thepade, Adib Parkar “A Comparison of Haar Wavelets and Kekre’s Wavelets for Storing Colour Information in a Image Processing, Computer Networks and UNIX programming. He Greyscale Image” International Journal of Computer Applications (0975 has 3 papers in an international journal to his credit. – 8887) Volume 11– No.11, December 2011.  Dr. H. B. Kekre, Sudeep Thepade, Nikita Bhandari “Colorization of Adnan Azmi is currently pursuing B.Tech. (CE) from MPSTME, Greyscale Images Using Kekre’s Biorthogonal Color Spaces and NMIMS University, Mumbai. His areas of interest are Image Kekre’s Fast Codebook Generation Advances in Multimedia” An Processing and Computer Networks. He has 2 paper in an International Journal (AMIJ), Volume (1): Issue (3) international journal to his credit. 63 http://sites.google.com/site/ijcsis/ ISSN 1947-5500
Pages to are hidden for
"Performance Comparison of Assorted Color Spaces for Multilevel Block Truncation Coding based Face Recognition"Please download to view full document