English

应用科学学报 ›› 2006, Vol. 24 ›› Issue (2): 140-144.

• 论文 • 上一篇    下一篇

一种基于KCCA的小样本脸像鉴别方法

贺云辉, 赵力, 邹采荣   

  1. 东南大学无线电工程系, 江苏南京 210096
  • 收稿日期:2004-11-24 修回日期:2005-05-25 出版日期:2006-03-31 发布日期:2006-03-31
  • 作者简介:贺云辉,博士生,研究方向:图像信号处理,E-mail:heyunhui@seu.deu.cn;赵力,教授,博导,研究方向:语言信号处理,E-mail:zhaoli@seu.edu.cn;邹采荣,教授,博导,研究方向:数字信号处理,E-mail:cairong@seu.edu.cn

Kernel Canonical Correlation Analysis and Application for Face Discrimination

HE Yun-hui, ZHAO Li, ZOU Cai-rong   

  1. Department of Radio Engineering, Southeast University, Nanjing 210096, China
  • Received:2004-11-24 Revised:2005-05-25 Online:2006-03-31 Published:2006-03-31

PDF

110

被引次数

36

摘要： 基于典型相关分析和Fisher线性鉴别分析的等价性,提出了利用核典型相关分析来抽取小样本人脸图像的非线性鉴别特征,并用其进行脸像鉴别.这样得到的非线性特征本质上等价于核Fisher非线性最佳鉴别特征.基于ORL库的实验表明,对小样本人脸图像,KCCA可以得到和广义鉴别分析近似的识别性能,其所得非线性特征明显优于FLDA的线性鉴别特征.

关键词: Fisher鉴别分析, 脸像鉴别, 典型相关分析, 核方法, 小样本问题

Abstract: Based on the equivalence between canonical correlation analysis (CCA) and Fisher linear discriminant analysis (FLDA), nonlinear discriminant features of face images are extracted with kernel CCA.These features are equivalent to those extracted with KFDA.Experimental results demonstrate that KCCA is similar to GDA and significantly better than FLDA.

Key words: canonical correlation analysis, Fisher discriminant analysis, kernel method, small sample size problem, face image discrimination

中图分类号: