Name : Research Field Research Proposal Title 1.

INTRODUCTION

: Computer Engineering : A comparison of face recognition algorithm

Face recognition is one of biometric methods identifying a person by the features of face. Face recognition has received a considerable attention in recent years both from the industry and research communities. The importance of face recognition rises from the fact that a face recognition system does not require the cooperation of individual while the other biometric system needs such cooperation. Facial recognition technology is one of the fastest growing fields in the biometric industry. Interest in facial recognition is being fueled by the availability and low cost of video hardware, the ever-increasing number of video cameras being placed in the workspace, and the noninvasive aspect of facial recognition systems. But automatic face recognition systems need to overcome various problems like pose invariance, illumination invariance, facial expression changes etc. Many methods of face recognition have been proposed during the past 30 years. As a result, the current status of face recognition technology is well advanced. Various novel techniques have been proposed ranging from the traditional template matching to the latest 3-dimensional techniques. Although over than 30 years of extensive research has been conducted in this area, there still exist open research issues, the performance of the current algorithms being still far from that of human perception. The goal of the face recognition is to identify or verify the persons present in the shots based on their facial features, despite of wide variations in pose, facial expressions and illumination changes [1].

2. PROBLEM STATEMENT
Currently there are many face recognition research report published. Most of them used different database and different criteria for their experiments, thus a comparison between every results reported is hard to be made. From all the face recognition report published, there are no comparisons made for resources used when doing experiments.

3. OBJECTIVES
The objective of the proposed title is to use a single publicly available face database, and compare with one published method of each technique. The paper reported to have the highest recognition rate will be use as a comparison. All the experiment conducted will use same criteria for every method so that results comparison can be made. Evaluation of tested methods suitability for real time face recognition will be made.

4. FACE RECOGNITION APPROACH

Face recognition can be done in both a still image and video which has its origin in still-image face recognition. Different approaches of face recognition for still images can be categorized into tree 3 groups such as holistic approach, feature-based approach, and hybrid approach [1]: Holistic (Global) approach. This approach uses the whole face region as the raw input to a recognition system. Holistic approach became very popular in the 1990s with the well known approach of Eigenfaces. Examples of holistic approach are Principal Component Analysis, Linear Discriminant Analysis, and Independent Component Analysis. Geometric (feature-based) approach. Geometric approaches dominated in the 1980s where simple measurements such as the distance between the eyes and shapes of lines connecting facial features were used to recognize faces. Examples of geometric approach are Hidden Markov Model, and Geometry Analysis. Hybrid approach. Just as the human perception system uses both local features and the whole face region to recognize a face, a machine recognition system should use both. One can argue that these methods could potentially offer the best of the two types of methods. Examples of hybrid approach are Hybrid Local Feature Analysis, and Component-based.

5. LITERATURE REVIEW 5.1 Holistic Approach

One of the most widely used representations of the face region is eigenfaces, which are based on principal component analysis (PCA). Fan Yang et al. [2] proposed a system that performs mosaicing of panoramic faces. This system is based on PCA and can be easily implemented in real-time. Panoramic face database was created and composed of 12 person 4 expressions 2

sessions = 96 panoramic faces. Two panoramic representations were tested: spatial and frequential. From the results, frequential representation gives the best performance with correct recognition rate of 97.46% versus 93.21% for spatial representation. Chunming Li et al. [3] proposed a face recognition method called Statistical PCA. The improved PCA algorithm is used to compute the eigen-vector and eigen-values of the face. Then, Bayesian rule is used to design the classification designer. The experimental result shows that the method introduced has the advantages of simple computation and high recognition rate up to 95.08%. Shaokang Chen et al. [4] evaluate how degradation affects the final recognition performance. Adaptive PCA was performed on still images captured with digital camera, and video sequences images from a CCTV camera. Both images set were simulated degraded. 81.6% recognition rate reported when using chosen still images as gallery, and 58.7% recognition reported when using chosen CCTV images as gallery. Recognition accuracy when using CCTV images decreases considerably because of the alignment issue. Xie Jianhong [5] combined Kernel PCA with LS-SVM (Least Square Support Vector Machine) applied to realize faces recognition based on ORL face database. The experimental results on the front 20 persons faces in ORL face database show that KPCA based on LS-SVM has a up to 95% correct recognition rate, and a faster computational speed. Jamikaliza Abdul Karim et al. [6] present a face recognition system based on Eigenfaces as feature extractor and Fuzzy ArtMap (FAM) neural network as classifier. The motivation of using FAM as a classifier is because of its unique solution to the stability-plasticity dilemma, where it has the ability to preserve previously learned knowledge and potential to adapt new patterns indefinitely. They reported recognition rate of 97.22% using local dataset and 98% using ORL dataset. Lang Liying et al. [7] presented the method of 2DPCA+SVD-U for face recognition. It overcomes the shortcomings of the eigenvectors which are obtained by singular vector projection, or the information of singular value which is regarded as identification characteristic. Recognition rate reported for ORL database is 87% and Yale database is 98.9%. Their experiment results show the superiority of 2DPCA+SVD-U algorithm over other algorithms, and show the proposed method solve recognition rate reduction resulting from illumination, pose etc variations to some extent. Jiaming Li et al. [8] proposed a robust face recognition algorithm using multiple face region features selected by the AdaBoost algorithm. They showed that dividing a face into a number of sub-regions can improve face recognition performance. They used conventional AdaBoost with a weak learner based on multiple region orthogonal principal component analysis (OCPCA) features. The regions are selected areas of the face (such as eye, mouth, nose etc.). The experiments had been done to evaluate the performance on the CMU Pose Illumination Expression (PIE) databases. The proposed method has shown that comparing with the general OCPCA method, multiple region OCPCA improves the recognition performance by 1% to 8% for 22.5 pose gallery image, and 1% to 10% for front pose gallery image. Comparing with Visionics FaceIt recognition system, the performance of multi-region OCPCA method was much better for both 22.5 pose and front pose gallery images. In all cases OCPCA has shown most improvement for wide-pose probe images. Xiaoxing and Hao [9] proposed an approach for 3D face recognition to achieve robustness against facial expression. They investigated the use of multiple intrinsic geometric attributes; including angles, geodesic distances, and curvatures, for 3D face recognition, where each face is represented by a triangle mesh, preprocessed to possess a uniform connectivity. The main contribution of this work towards expression invariant 3D face recognition is the idea that multiple geometric attributes derived from the face meshes can be utilized and even though they are all elementary, the key to improved performance lies in adapting these attributes to instabilities introduced by facial expressions and in effectively combining them; both can be realized through training. They reported 85% recognition rate on Notre Dame (UNC) FRGC database. Qian Zhi-ming et al [10] proposed a new face recognition method based on the analysis of local features. First, they get the images of magnitude by means of analyzing face images with the Gabor wavelets. Then the magnitude images are divided into blocks, then principle components analysis (PCA) could be directly used to all the blocks to construct the feature space. Lastly, all the blocks of images are projected to the feature space and get the face feature vectors. They get the recognition rate by counting and analyzing the feature vector. They reported 91% recognition rate on Yale database and 86% recognition rate on their own database.

5.2 Geometric approach

5.3 Hybrid approach

Weyrauch et al [11] proposed an approach to pose and illumination invariant face recognition that combines componentbased recognition and 3D morphable models. The 3D models are rendered under varying pose and illumination conditions to build a large set of synthetic images. These images are then used to train a component-based face recognition. Their system achieved 90% recognition rate on a database of 1200 real images of 6 people and significantly outperformed a comparable global face recognition system which only achieved 40% accuracy on their experiment.

6. WORK FLOW DIAGRAM

7. FRAMEWORK SCHEDULE 2010 TASK\MONTH

Literature review Research Proposal Data Collection Data analysis Research design System design Preparation of journal paper Thesis writing/editing Thesis review by supervisors Research viva Thesis correction Submission

2011 1 0 1 1 1 2 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2

1 2 3 4 5 6 7 8 9

6. REFERENCES
[1] [2] Zhao, W., Chellappa, R., Phillips, P. J., Rosenfeld, A., 2003, Face recognition: A literature survey, ACM Computing Surverys (CSUR), V. 35, Issue 4, pp. 399-458. Yang, F., Paindavoine, M., Abdi, & H., Arnoult, D. (2006). Fast panoramic face mosaicing and recognition. Journal of Multimedia, 1, 14-20. Li Chunming, Diao Yanhua, Ma Hongtao, and Li Yushan; A Statistical PCA Method for Face Recognition; Intelligent Information Technology Application, 2008. IITA 2008. Second International Symposium on Volume 3, 20-22 Dec. 2008 Page(s):376 380. Shaokang, Erik, Abbas, Conrad, and Brian C.; Experimental Analysis of Face Recognition on Still and CCTV Images, 2008; Advanced Video and Signal Based Surveillance, 2008. AVSS '08. IEEE Fifth International Conference on 1-3 Sept. 2008 Page(s):317 324. Jianhong Xie; KPCA Based on LS-SVM for Face Recognition; Intelligent Information Technology Application, 2008. IITA '08. Second International Symposium on Publication Date: 20-22 Dec. 2008 Volume: 2, on page(s): 638641. Karim Jamikaliza Abdul, Yusof Rubiyah, and Khalid Marzuki; Investigate the Performance of Fuzzy Artmap Classifier for Face Recognition System; Signal Image Technology and Internet Based Systems, 2008. SITIS '08. IEEE International Conference on Nov. 30 2008-Dec. 3 2008 Page(s):254 259. Lang Liying and Hong Yue; Study on the Core Algorithm of Access Control System Based on Face Recognition, 2008, Computational Intelligence and Security, 2008. CIS '08. International Conference on Volume 1, 13-17 Dec. 2008 Page(s):76 79. Jiaming Li, Geoff Poulton, Ying Guo, Rong-Yu Qiao, Face Recognition Based on Multiple Region Features, 2003, Proc. VIIth Digital Image Computing: Techniques and Applications, Sun C., Talbot H., Ourselin S. and Adriaansen T. (Eds.), 10-12 Dec. 2003, Sydney
Xiaoxing Li; Hao Zhang, Adapting Geometric Attributes for Expression-Invariant 3D Face Recognition, Shape Modeling and Applications, 2007. SMI apos;07. IEEE International Conference on 13-15 June 2007 Page(s):21 32 Qian Zhi-ming, Su

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Peng-yu, and Xu Dan; Face Recognition Based on Local Feature Analysis; Computer Science and Computational Technology, 2008. ISCSCT '08. International Symposium on Volume 2, 20-22 Dec. 2008 Page(s):264 267 [10] Weyrauch, B. Heisele, B. Huang, J. Blanz, V. "Component-Based Face Recognition with 3D Morphable Models", 2004, Computational Learning, M.I.T., Cambridge, MA; Computer Vision and Pattern Recognition Workshop, 2004 Conference on 27-02 June 2004, On page(s): 85- 85