E115UUM31EEUGIEVUUElERIRlllIIIIIUIPIIIIIWA(PUUIGIIIKM3UIU5-7, September, 2005, Kuala Lumpur, MALAYSIA A Comparative Analysis of Feature Extraction Methods for Face Recognition System Nor'aini A. J.', P. Raveendran1, and N. Selvanathan2 Department of Electrical Engineering, Faculty of Engineering University Malaya 50603, Kuala Lumpur ZDepartment of Artificial Intelligence, Faculty of Computer Science and Information Technology University Malaya 50603, Kuala Lumpur Abstract--Recognizing face images due to changes in illumination condition, pose, facial expression and others are challenging task. Solving these problems requires a feature extraction method that can generate distinct features for each class of image. Hence, this paper describes the comparative analysis of feature extraction methods namely Geometric moments, Zernike moments, Krawtchouk moments and Principle component analysis (PCA) in terms of their capability to recognize face images. The classification technique employed in the recognition stage is Back propagation neural network (BPNN). The experiments utilized database face images from Olivetti research laboratory (ORL) consisting of 40 subjects of 10 samples each where none of them are identical [1]. They vary in position, rotation, scale and expression. From the comparative study, the most suitable feature extraction method is considered for face recognition system. Keywords:Geometric moments, Zernike moments, Krawtchouk moments, PCA, BPNN. 1. Introduction Face recognition has been actively research over a decade and now it is still being research although many of the latest work are either the improvement of existing techniques or hybrid techniques. The face recognition systems have wide range of application such as access control systems, content-based video browsing, building or office security, criminal identification and authentication in secure systems like computers or bank teller machines [2]. Building an automated system that can accomplish the above mentioned objectives is not an easy task. Constraints such as poses, illumination conditions, facial expressions, aging and many others are still the main problem to achieve high classification accuracy. A successful face recognition system however depends heavily on the particular choice of feature extraction methods. Regardless of the method used, extracted features must minimize the within-class face variability and maximize between-class face variability in order to provide sufficient discrimination among different faces [3]. Moment based feature extraction methods such as geometric invariant central moments, Lengendre moments, Zernike moments, Pseudo Zemike moments, Krawtchouk moments and others have gain attention lately. They have proven to be suitable for handling images with binary patterns such as pattern recognition, palm print verification and etceteras [4]. These moments acquire the characteristic of translation, scaling and rotation invariance and thus can be chosen for image analysis and pattem recognition application [4]. Other feature extractors like PCA, Fourier descriptors and others have also gain attention and use as a hybrid feature to represent faces. For instance Fourier descriptors were used along with Zernike moments by A. Saradha et al and Ahmadi et al combine PCA with Pseudo Zernike moments [4]. In this paper the classification results from the experiments utilizing the feature extraction methods namely Geometric moments, Zernike moments, Krawtchouk moments and PCA are presented and analyzed. The data input consists of original data, edge detected data and localized data. The rest of the paper is organized as follows. Section 2 describes the theory of the feature extraction methods. Section 3 detailed out the classifier used for face recognition. Experimental results are presented in section 4. Section 5 analyses the perfonnance of the feature extractors and finally section 6 concludes the paper. 2. Feature Extraction methods 2.1 Geometric moments For two dimensional function f(x,y), the geometric moment of order (p+q) is defined as Go 00 mp= I JxPYf(x,y)dxdy (1) -, -00 p, q=0,1,2,3,...oo. For a digital image then equation I becomes M-IN-1 X= 'E fl(xY), X=Oy=O p, q=0, 1,2,3,.-.. oo. (2) M and N are the horizontal and vertical dimensions respectively andf(x,y) is a digital image at 0-7803-9370-8/05/$20.00 ©2005 IEEE. 176