Two-dimensional color uncorrelated discriminant analysis for face recognition

详细信息    查看全文

• 作者：Cairong Zhao^a ; ^b ; ^{cairong.zhao@yahoo.com}Author Vitae ; Duoqian Miao^a ; ^b ; ^{dqmiao@tongji.edu.cn}Author Vitae ; Zhihui Lai^c ; ^{lai_zhi_hui@163.com}Author Vitae ; Can Gao^a ; ^b ; ^{2005gaocan@163.com}Author Vitae ; Chuancai Liu^dAuthor Vitae ; Jingyu Yang^dAuthor Vitae
• 关键词：Face recognition ; Color image ; Two-dimension ; Uncorrelated discriminant analysis ; Feature extraction ; Pattern recognition
• 刊名：Neurocomputing
• 出版年：2013
• 出版时间：3 August, 2013
• 年：2013
• 卷：113
• 期：Complete
• 页码：251-261
• 全文大小：1687 K

文摘

This paper presents a novel color face recognition method called two-dimensional color uncorrelated discriminant analysis (2DCUDA), which can extract two-dimensional color uncorrelated features and simultaneously retain the face spatial structure information. The 2DCUDA method seeks to explore color uncorrelated discriminant properties of the color face images and eliminate the correlations between color-based features. The novelties of this paper are twofold. First, this paper develops a new color-based feature for face recognition, which can provide substantial mutual complementation information and improve the recognition performance. Second, theoretical analysis guarantees the uncorrelated property of the obtained color-based features. Comparative experiments on AR and FRGC-2 color face databases have been conducted to investigate the effectiveness of the proposed algorithm. Experimental results show that the proposed algorithm performs better than other color face recognition methods and the two-dimensional color uncorrelated discriminant features are more effective for low-resolution image compared with conventional gray-based features. Finally, we explain why the proposed algorithm can improve the recognition performance compared with other color face recognition methods.