Abstract  This  paper  investigates  a  face  recognition  system  b ased  on  Scale  Invariant  Feature  Transform  (SIFT)  fe ature  and  its distribution on feature space. The system takes  advantage of SIFT which possess strong robustness  to expression, accessory  pose  and  illumination  variations.  Since  we  use  each   of  SIFT  keypoint  as  the  feature  of  face  and  SIFT  k eypoints  are  very  complicated  in  feature  space,  we  apply  the  feature  partition  on  Self  Organizing  Map  (SOM)  and  adopt  lo cal  Multilayer  Perceptron (MLP)  for each node on map to improve t he classification performance. Moreover the distinctive features from all  SIFT  keypoints  in  each  face  class  are  defined  and  e xtracted  based  on  feature  distribution  on  SOM.  Fina lly  the  face  can  be  recognized  through  the  proposed  scoring  method  depe nding  on  the  classification  result  of  these  distinctive  features.  In  the  experiments,  the  proposed  method  gave  a  higher  face   recognition  rate  than  other  methods  including  matching  and  holistic  feature based methods in three famous databases.  

References

[1] Aly M., Face Recognition using SIFT Features, available at: http://www.vision.caletech.edu/ malaa/research.php, last visited 2006.

[2] Belhumeur P., Hespanha J., and Kriegman D., Eigenfaces vs. FisherFaces: Recognition using Class Specific Linear Projection, IEEE Transactions Pattern Analysis and Machine Intelligence , vol. 19, no. 7, pp. 711-720,1997.

[3] Bicego M., Lagorio A., Grosso E., and Tistarelli M., On the Use of SIFT Features for Face Authentication, in Proceedings of IEEE International Workshop on Biometrics, in association with CVPR , USA, pp. 35-41, 2006. Selection of Distinctive SIFT Feature Based on its Distribution on Feature Space... 101

[4] Bishop C., Neural Networks for Pattern Recogntion , New York: Oxford University Clarendon Press, 1995.

[5] Face Database, available at: http://www.face- rec.org/databases, last visited 2011.

[6] Han Y. and Yin J., Human Face Feature Extraction and Recognition Base on SIFT, in Proceedings of IEEE International Symposium on Computer Science and Computational Technology , Shanghai, vol. 1, pp. 719-722, 2008.

[7] Kaski S., Kangas J., and Kohonen T., Bibliography of Self-Organizing Map (SOM) Papers: 1981-1997, Neural Computing Surveys , 1:102-350, 1998.

[8] Kohonen T., Self-Organizing Maps , 3 rd Extended Edition, Springer, 2001.

[9] Looney C., Pattern Recognition using Neural Networks: Theory and Algorithms for Engineers and Scientists , Oxford University Press, 1997.

[10] Lowe D., Distinctive Image Features from Scale-Invariant Keypoints, International Journal of Computer Vision , vol. 60, no. 2, pp. 91-110, 2004.

[11] Luo J., Ma Y., Takikawa E., Lao S., Kawade M., and Bao-Liang L., Person-Specific SIFT Features for Face Recognition, in Proceedings of International Conference on Acoustic, Speech and Signal Processing , Honolulu, pp. 563-566, 2007.

[12] Majumdar A. and Ward R., Discriminative SIFT Features for Face Recognition, in Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing , Canada, pp. 27- 30, 2009.

[13] Turk M. and Pentland A., Face Recognition using Eigenfaces, in Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recogniton , USA, pp. 586- 591, 1991.

[14] Wiskott L., Fellous J., Kruger N., and Malsburg C., Face Recognition by Elastic Bunch Graph Matching, IEEE Transactions Pattern Analysis and Machine Intelligence , vol. 19, no. 7, pp. 775- 779, 1997.

[15] Yang J., Zhang D., and Yang J., Two- Dimensional PCA: A New Approach to Apprerance-Based Face Representation and Recogntion, IEEE Transaction on Pattern Analysis and Machine Intelligence , vol. 26, no. 1, pp. 131-137, 2004.

[16] Yang M., Kernel Eigenfaces vs. Kernel Fisherfaces: Face Recognition Using Kernel Methods, in Proceedings of 5 th IEEE International Conference Automatic Face and Gesture Recognition , USA , pp. 215-220, 2002.

[17] Yuen P. and Lai J., Face Representation using Independent Component Analysis, Pattern Recognition , vol. 35, no. 6, pp. 1247-1257, 2002.

[18] Zhao W., Chellappa R., Philips P., and Rosenfeld A., Face Recognition: A Literature Survey, ACM Computing Survey , vol. 35, no. 4, pp. 399- 458, 2003.

[19] Zhang L., Chen J., Lu Y., and Wang P., Face Recogntion using Scale Invariant Feature Transform and Support Vector Machine, in Proceedings of 9 th International Conference for Young Computer Scientists , Hunan, pp. 1766- 1770, 2008. Tong Liu received his BSc degree in information security engineering from Daebul University in 2008 and MS degree in computer engineering from Kyung Hee University in 2010. His research interests include face recognition, image processing, computer vision and machine learning. Sung-Hoon Kim received his BSc degree in computer engineering from Daebul University, Korea in 2002. He received MS degree in Electronic Computing Engineering from Kyung Hee University in 2004. Now he is pursuing his PhD in the Department of Computer Engineering, Kyung Hee University, Korea. His research interests include neural networ k, pattern recognition and face detection. Sung-Kil Lim received his BSc degree in Mathematics, MSc degree in computing engineering and phd degree in computer engineering from Kyung Hee University, Korea, in 1997, 1999 and 2009, respectively. His research interests include auditory scene analysis, pattern recognition and ne ural network. He is now a senior engineer working in Brisys Technologies Company, Korea. Hyon-Soo Lee received his BSc degree in electronics engineering from Kyung Hee University, Korea, and the MS, PhD degree in electrical engineering from Keio University, Yokohama, Japan in 1979, 1982 and 1985, respectively. He is now a professor in the Department of Computer Engineering , Kyung Hee University. His research interests includ e neural network, pattern recognition and parallel architecture.