Comparison of matrix-based linear discriminant analysis methods

摘要：

Up to now, a number of linear discriminant analysis (LDA) methods have been developed. Though all these methods have a goal of maximizing the between-class distance and of minimizing the within-class distance, they are formally different and seem to have different performance. This paper aims at revealing the similarity and difference between different matrix-based LDA methods, which is helpful for people to better understand these methods. We first use a formally united framework to present the LDA methodology including different LDA methods and then perform comparison experiments of two LDA methods.

关键词： LDA face recognition vectors matrix 2DLDA between-class distance within-class distance Heterogeneous Face Biometrics complex-matrix

作者: Ningbo Zhu Cong Li Kaikai Lv

作者单位: School of Computer and Communication Hunan University Changsha,China School of Computer and Communication Hunan University Changsha,China School of Computer and Communication Hunan University Changsha,China

会议类型: 国际会议

会议名称: 2010 International Conference on Future Information Technology(2010年未来信息技术国际会议 ICFIT 2010)

会议地点: 长沙

会议语种:英文

页码: 1-4

在线出版日期: 2010-12-14（万方平台首次上网日期，不代表论文的发表时间）