人脸识别中PCA方法的推广  

The Generalization of PCA for Human Face Recognition

作　　者：陈伏兵[1] 陈秀宏[1] 王文胜[2] 杨静宇[2] 

机构地区：[1]淮阴师范学院数学系,江苏淮安223001 [2]南京理工大学计算机科学系,南京210094

出　　处：《计算机工程与应用》2005年第34期34-38,共5页Computer Engineering and Applications

基　　金：国家自然科学基金项目(编号:60472060);江苏省自然科学基金项目(编号:05KJD520036)

摘　　要：主成分分析(PrincipalComponentAnalysis,PCA)是公认的特征抽取的最为重要的工具之一,目前仍然被广泛地应用在人脸等图像识别领域。基于PCA,该文提出了分块PCA的人脸识别方法。分块PCA方法先对图像矩阵进行分块,对分块得到的子图像矩阵利用PCA进行鉴别分析。其特点是能有效地抽取图像的局部特征,对人脸表情和光照条件变化较大的图像表现尤为突出。与PCA方法相比,由于使用子图像矩阵,分块PCA可以避免使用奇异值分解理论,过程简便。此外,PCA是分块PCA的特殊情况。在Yale和NUST603人脸库上的试验结果表明,所提出的方法在识别性能上明显优于经典的PCA方法,识别率可以分别提高6.7和4个百分点。Principal component analysis （PCA）is one of accepted and important technique for feature extraction widely used in the areas of images recognition such as human face recognition.Modular PCA,a human face recognition technique based on PCA,is presented in this paper.Firstly,in proposed approach,the original images are divided into modular images,which are also called sub-lmages.Then,the well-known PCA method can be directly employed to the sub-images obtained from the previous step.There are two advantages for this way： （1）local feature of the images can be extracted effficiently,and it is really true of the images that have large variations in facial expression and lighting; （2）singnlar value decomposition of matrix may be avoided in the process of feature extraction,which is simple than that of other technologies such as PCA.Moreover,PCA is a special case of modular PCA.To test modular PCA and to evaluate its performance,a series of experiments will be performed on two human face image databases：Yale and N JUST603 human face databases.The experimental results indicate that the performance of modular PCA is obviously superior to that of PCA.The recognition rates,which may be obtalned,are 100 percent and 97.1 percent with respect to the two databases,respectively.

关 键 词：主成分分析 特征抽取 分块PCA 特征矩阵 人脸识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]