核协同表示在人脸识别的遮挡问题中的应用  

Application of Kernel Collaborative Representation in Occlusion Problem of Face Recognition

作　　者：董吉文[1] 赵磊[1] 张亮[1] 

机构地区：[1]济南大学信息科学与工程学院,山东济南250022

出　　处：《计算机技术与发展》2013年第7期141-143,170,共4页Computer Technology and Development

基　　金：山东省自然科学基金(ZR2010FL006)

摘　　要：人脸识别算法是一个特征提取和分类器设计的过程。针对人脸识别中的遮挡问题,提出一种基于核主成分分析(KPCA)和协同表示(CRC)相结合的人脸识别算法。提取特征时,利用KPCA提取人脸图像中利于判决的非线性结构特征,使得样本在保留了最有效判别信息的同时降低了特征维数。设计分类器时,考虑到样本之间的协同性,采用综合考虑样本之间信息的协同表示分类器进行分类识别。实验结果证明,该算法获得了很好的识别效果,效率也得到了提高。Face recognition algorithms are a process of feature extraction and classifier design. According to the problems of occlusion and illumination in face recognition, a new method of face recognition based on Kernel Principal Components Analysis （KPCA） and Collaborative Representation Classifier （CRC） is developed. When extracting features,the KPCA can extract faces＇ nonlinear structures features which were benefit to discriminate, by this way, the samples retained the effective discriminating information, at the same time reduced the feature dimensions. When designing classifier, considering the collaboration among the samples, the CRC which synthetically consider the relationship among samples was being used. Experimental results demonstrated that the algorithm could obtain good recognition effect and also improve the efficiency.

关 键 词：人脸识别 核主成分分析 协同表示分类器 遮挡问题 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]