计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (26): 38-40.DOI: 10.3778/j.issn.1002-8331.2008.26.011

• 理论研究 • 上一篇    下一篇

一种广义的主成分分析特征提取方法

朱明旱1,2,罗大庸1,易励群3   

  1. 1.中南大学 信息科学与工程学院,长沙 410083
    2.湖南文理学院 电气工程系,湖南 常德 415000
    3.湖南文理学院 计算机基础教学部,湖南 常德 415000
  • 收稿日期:2007-11-02 修回日期:2008-01-21 出版日期:2008-09-11 发布日期:2008-09-11
  • 通讯作者: 朱明旱

Generalized PCA algorithm for feature extraction

ZHU Ming-han1,2,LUO Da-yong1,YI Li-qun3   

  1. 1.College of Information Science and Engineering,Central South University,Changsha 410083,China
    2.Department of Electric Engineering,Hunan University of Arts and Science,Changde,Hunan 415000,China
    3.Department of Basic Computer Sciences,Hunan University of Arts and Science,Changde,Hunan 415000,China
  • Received:2007-11-02 Revised:2008-01-21 Online:2008-09-11 Published:2008-09-11
  • Contact: ZHU Ming-han

摘要: 提出了一种广义的PCA特征提取方法。该方法先将图像矩阵进行重组,根据重组的图像矩阵构造出总体散布矩阵,然后求出最佳投影向量进行特征提取。它是2DPCA和模块2DPCA的进一步推广,可以建立任意维数的散布矩阵,得到任意维数的投影向量。实验表明,随着总体散布矩阵维数的减小,广义PCA的特征提取能力更强,特征提取的速度也更快。

关键词: 散布矩阵, 主成分分析, 本征向量, 特征提取

Abstract: In this paper,a generalized principle component analysis for feature extraction is proposed.First,image matrixes are reshaped in proposed method.Then,an image scatter matrix is constructed using the reshaped image matrixes and its eigenvectors are derived for image feature extraction.The generalized principle component analysis is an extension of two-dimensional PCA and modular two-dimensional PCA.Scatter matrix with random size can be constructed using the method.So,the dimension of projection vector can change according need.Experiment results show that the generalized principle component analysis can extract better feature value and has higher performance efficiency with the reduction of scatter matrix size.

Key words: scatter matrix, principal component analysis, eigenvector, feature extraction