一种基于全局和局部特征集成的人脸识别方法.pdfVIP

∗ 一种基于全局和局部特征集成的人脸识别方法 1+ 2 2 1,3 苏 煜 , 山世光 , 陈熙霖 , 高 文 1(哈尔滨工业大学 计算机学院, 黑龙江 哈尔滨 150001) 2( 中国科学院智能信息处理重点实验室, 中国科学院计算技术研究所, 北京 100190) 3(北京大学 数字媒体研究所, 北京 100871) * Integration of Global and Local Feature for Face Recognition 1+ 2 2 1,3 SU Yu , SHAN Shi-Guang , CHEN Xi-Lin , GAO Wen 1(School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001, China) 2(Key Lab of Intelligent Information Processing, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China) 3(Institute of Digital Media, Peking University, Beijing 100871, China) + Corresponding authors: Phn: +86-10 Fax: +86-10 E-mail: ysu@, Received 2008- 10-28; Accepted 2009-03-31 Abstract : Extracting facial features with high specificity and invariance is key to the construction of efficient and robust face recognition system. In the literature of psychology and physiology, many studies have shown that global and local features play different roles in face representation and recognition: global feature describes some holistic characteristic of the face, whereas local features only encode the detailed traits within certain local area. Considering that, this paper proposes to combine global and local features by both serial and parallel manner. Firstly, global features are used for coarse classification. Then, global and local features are integrated for