作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2009, Vol. 35 ›› Issue (3): 214-216. doi: 10.3969/j.issn.1000-3428.2009.03.072

• 人工智能及识别技术 • 上一篇    下一篇

基于CSVD-NMF的人脸识别算法

姚同庆,房 斌,尚赵伟   

  1. (重庆大学计算机学院,重庆 400044)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2009-02-05 发布日期:2009-02-05

Face Recognition Algorithms Based on CSVD-NMF

YAO Tong-qing, FANG Bin, SHANG Zhao-wei   

  1. (School of Computer Science, Chongqing University, Chongqing 400044)
  • Received:1900-01-01 Revised:1900-01-01 Online:2009-02-05 Published:2009-02-05

摘要: 基于SVD的人脸识别算法具有共同的缺点,即不同人脸图像对应的奇异值向量所在的基空间不一致,从而造成识别率低下。该文分析2种改进的类估计基空间奇异值分解算法(CSVD),通过对比实验选择出其中一种具有优势的CSVD算法。并在特征提取环节,提出CSVD算法与非负矩阵因子算法特征数据相融合的人脸识别算法。在ORL数据库上的实验结果表明,该结合方法有效地提高了识别率和训练速度。

关键词: 类估计基空间奇异值分解, 非负矩阵因子, 特征提取

Abstract: The face recognition algorithms based on SVD have low recognition accuracy due to the common essential defect which singular value vector of arbitrary two face images have the different basis spaces in general. According to this, two improved class estimated basis space singular value de-composition methods are analyzed. A superior CSVD method is selected after comparing with each other. And in the feature extraction process, a new face recognition method based on CSVD and non Negative Matrix Factorization(NMF) is presented. By combining both methods, for ORL database, the better recognition performance is obtained.

Key words: Class estimated basis Space singular Value De-composition(CSVD), non Negative Matrix Factorization(NMF), feature extraction

中图分类号: