计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (5): 151-153.DOI: 10.3778/j.issn.1002-8331.2010.05.046

• 图形、图像、模式识别 • 上一篇    下一篇

基于WPCA和修正的最大间距准则的人脸识别

王进军1,王汇源1,2   

  1. 1.山东大学 信息科学与工程学院,济南 250100
  • 收稿日期:2008-08-15 修回日期:2008-11-10 出版日期:2010-02-11 发布日期:2010-02-11
  • 通讯作者: 王进军

Face recognition based on WPCA and modified maximum margin criterion

WANG Jin-jun1,WANG Hui-yuan1,2   

  1. 1.School of Information Science and Engineering,Shandong University,Jinan 250100,China
    2.IEETA,Campus Universitário de Santiago,3810-193 Aveiro,Portugal
  • Received:2008-08-15 Revised:2008-11-10 Online:2010-02-11 Published:2010-02-11
  • Contact: WANG Jin-jun

摘要: 针对Fisher准则遇到的高维小样本问题和最大间距准则遇到的“次优化问题”,提出一种基于加权PCA(WPCA)和修正的最大间距准则(MMMC)的线性判别分析方法。首先对PCA空间进行加权,对最大间距准则的散布矩阵进行修正,然后结合WPCA和MMMC进行特征提取。该方法为有效地解决上述两个问题提供了途径。在ORL和FERET人脸库上的实验结果验证了该方法的有效性。

关键词: 人脸识别, 加权主成分分析(PCA), 最大间距准则

Abstract: Considering the high dimensions and the‘small sample size’ problem in Fisher linear discriminant analysis and the ‘inferior’ problem in the maximum margin criterion,a face recognition approach is proposed based on weighted PCA and modified maximum margin criterion.The approach gives an effective way to resolve the two problems above.Experimental results on ORL and FERET database verify the effectiveness of the proposed method.

Key words: face recognition, weighted Principal Component Analysis(PCA), maximum margin criterion

中图分类号: