基于局部奇异值分解和D-S理论的人脸识别方法  

Face Recognition Method Based on Local Singular Value Decomposition(SVD) and D-S Theory

作　　者：李晓东[1] 费树岷[1] 张涛[1] 

机构地区：[1]东南大学自动化学院,南京210096

出　　处：《数据采集与处理》2009年第5期621-625,共5页Journal of Data Acquisition and Processing

基　　金：国家自然科学基金(60574006)资助项目

摘　　要：提出了基于局部奇异值分解和D-S证据理论的人脸识别方法。首先分割出人脸图像的5个特殊区域并分别进行奇异值分解,提取一些较大的奇异值构成每一区域的特征向量;在识别阶段,计算待识别人脸图像每一区域对所有训练样本人脸图像相应区域的隶属度,最后由D-S理论的组合规则做出判断。基于ORL人脸数据库的实验结果证明了本文方法的有效性和可行性。An effective method for the face recognition is proposed based on the local singular value decomposition（SVD） and D-S theory. Firstly, the facial image is divided into five special areas, and SVD is performed on these areas. Some maximal singular values are picked up to form the feature vector of each area, subsequently. During recognition period, the membership degrees of each area on testing facial image to each corresponding area of all training facial samples are computed. Finally, the recognition result can be obtained using the combination rules of D-S theory. Experimental results based on ORL face database show that the method is efficient and feasible.

关 键 词：人脸识别 奇异值分解 D—S理论 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]