基于协作表示和模糊渐进最大边界嵌入的特征抽取方法  

Feature extraction based on collaborative representation and fuzzy progressive maximal marginal embedding

作　　者：苏宝莉[1] 

机构地区：[1]江苏常州机电职业技术学院信息工程系,江苏常州213000

出　　处：《计算机应用》2013年第6期1677-1681,共5页journal of Computer Applications

基　　金：2012年度江苏省高校哲学社会科学研究基金资助项目(2012SJB880009);2012年度常州科教城(高职教育园区)院校科研基金资助项目(K2012212)

摘　　要：针对图嵌入方法在构造邻域关系图的过程中,简单地将样本数据划入某一类的做法并不妥当的问题,提出了模糊渐进的隶属度表示方法。该方法借助模糊数学的思想,通过模糊渐进的隶属度,将样本归属于不同类别。针对图嵌入方法中分类器效率偏低的问题,引入了协作表示分类方法,该分类方法大幅度提高了算法的计算效率。基于这两点,提出了基于协作表示和模糊渐进最大边界嵌入的特征抽取算法。在ORL、AR人脸数据库上,以及USPS数字手写体数据库上的实验表明,该算法优于主成分分析(PCA)、线性鉴别分析(LDA)、局部保留投影(LPP)和边界Fisher分析(MFA)。In the procedure of the construction of neighborhood graph, traditional graph-embedding algorithms adopt a simple two-value hard classifier criterion. Concerning this problem, with reference to the fuzzy mathematics, a new fuzzy progressive neighbor graph was proposed in this paper. Furthermore, collaborative representation classifies patterns by employing all the training images to represent the query image collaboratively. Therefore, in this paper, collaborative representation was introduced into classifier. Concerning the problems mentioned above, a feature extraction algorithm based on collaborative representation and fuzzy progressive maximal marginal embedding was proposed for face recognition. The experimental results on the ORL, AR face databases and USPS handwriting number database show that the proposed algorithm outperforms Principal Component Analysis （PCA）, Linear Discriminant Analysis （LDA）, Localities Preserving Projections （LPP） and Margin Fisher Analysis （MFA）.

关 键 词：模式识别 人脸识别 协作表示 模糊渐进构造 图嵌入 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]