Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (10): 185-187.

• 图形、图像处理 • Previous Articles     Next Articles

Independent local feature analysis with applications to face recognition

XU Gao-feng1,HUANG Lei2,LIU Chang-ping2,DING Shi-qi1   

  1. 1.Harbin Engineering University,Harbin 150001,China
    2.Institute of Automation,Chinese Academy of Sciences,Beijing 100080,China
  • Received:2007-07-23 Revised:2007-10-22 Online:2008-04-01 Published:2008-04-01
  • Contact: XU Gao-feng

面部局部特征的独立分量分析及其应用

许高凤1,黄 磊2,刘昌平2,丁士圻1   

  1. 1.哈尔滨工程大学,哈尔滨 150001
    2.中科院 自动化研究所,北京 100080
  • 通讯作者: 许高凤

Abstract: Feature extraction is always an important domain in face recognition.Local Feature Analysis(LFA) algorithm not only can extract global information features but also can extract local information features from face image,but the LFA feature is a redundant representation of the original image,it remains residue correlations.Since Independent Component Analysis(ICA) has the good performance in decorrelation both in low order and high order.This paper integrates the above two algorithms,and the followed experiments shows its good performance.

Key words: LFA, ICA, feature extraction, face recognition

摘要: 特征提取方法一直是人脸识别研究中的热点,局部特征分析(Local Feature Analysis)算法不仅能得到面部的全局特征,而且能提取出其局部特征信息,但该算法得到的结果具有过多冗余相关信息不利于识别。由于独立成分分析(Independent Component Analysis)算法能够有效地提取信号的高阶统计特性,很好地去除了各分量之间的相关性。给出了融合这两种方法的特征提取方法,经实验测试表明该算法能有效地提取面部特征。

关键词: LFA, ICA, 特征提取, 人脸识别