IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.1, January 2009 27 Manuscript received January 5, 2009 Manuscript revised January 20, 2009 3D Face Recognition Using Multiple Features for Local Depth Information Fatimah Khalid, Lili N. A. Department of Multimedia, Faculty of Computer Science and Information Technology University Putra Malaysia, 43400 UPM Serdang,Selangor Darul Ehsan Summary In this paper, we recognized multiple features from the local depth information of distance and angle calculation. These features are calculated from the twelve salient points by considering the distance and angle calculation. Then, fifty three non-independent features are extracted and the discriminating power is used for analyzing these features. The result shows an improvement compared to the previous work. Key words: 3D face recognition, feature localization, feature extraction 1 Introduction Face recognition is based on the computer identification of unknown face images by comparison with a database of known images. Several disciplines involved such as image processing, pattern recognition, computer vision, computer graphics, and machine learning. Since 1975, researchers in psychophysics, neural sciences and engineering, image processing, analysis, and computer vision have been investigating a number of issues related to face recognition by humans and machines. The vast majority of face recognition research has focused on the use of two-dimensional intensity (Bowyer et al. 2004). Over a decade ago, a new research paradigm for face recognition focused on three- dimensional (3D) images, either by itself or in combination with two-dimensional (2D) images. The use of multiple imaging modalities, such as 3D and 2D images of the face refers to multi-modal biometrics. Recently, the curiosity interest in model-based 3D automatic recognition systems has been increased (Lee & Ranganath 2003; Huang et al. 2003). The process of 3D face recognition is the same as 2D face recognition where it involves detection as well as representation and matching. Gupta et al. (2007) categorized 3D face recognition techniques into two main categories which is (a) based on the appearance of facial range images (appearance based) and (b) based on the geometric properties of ‘free form’ facial surfaces (‘free form’ based). They said that appearance based is also known as statistical learning based where the techniques are all in statistical approach (as shown in Figure 1). These appearances of facial range approaches are similar to 2D holistic appearance based techniques where the only difference is the employment of range images instead of intensity images. Meanwhile, ensemble approaches consisting of combinations of multiple ‘free form’ techniques. Figure 1. 3D face recognition categorization Source: Gupta et al. Studies considering in local geometric feature of 3D facial surfaces still few in numbers. This is due to the requirement of automatic segmentation of facial landmarks. The local shapes of facial landmarks/regions on fiducial points have been quantified. For instance, Gordon (1992) and Moreno et al. (2005) used Gaussian curvature values for segmenting the regions and lines of 3D Face Recognition Techniques ‘Free form’ based Facial Surface Matching Surface Normal Orientation Profile Matching 3D Local Geometric Features Appearance based PCA ICA LDA LFA HMM OCA