鉴别稀疏保持投影的人脸识别算法  

Face Recognition Based on Discriminant Sparsity Preserving Projection

作　　者：李昆仑[1] 耿雪菲[1] 曹静媛[1] 

机构地区：[1]河北大学电子信息工程学院,河北保定071000

出　　处：《小型微型计算机系统》2017年第2期376-380,共5页Journal of Chinese Computer Systems

基　　金：国家科技支撑计划项目(2013BAK07B04)资助;河北省自然科学基金项目(F2013201170)资助;河北省高等学校科学技术研究重点项目(ZD2014008)资助

摘　　要：在人脸识别领域中遇到的数据往往是高维的,一般会导致维数灾难问题.近年来稀疏表示(Sparse representation,SR)在处理人脸识别等问题时显示出一定的有效性,而后出现的稀疏保持投影(Sparse preserving projections,SPP)算法又以保持数据的稀疏表示结构为目的成功应用于人脸识别领域,但仍存在一些问题.本文针对SPP算法在人脸识别中存在的问题进行了改进,提出了一种叫做鉴别稀疏保持投影(Discriminant sparsity preserving projection,DSPP)的算法.该算法有以下两方面的改进:(1)针对SPP算法未能有效地利用类标签信息的问题,本文利用最大散度差准则(Maximum scatter difference criterion,MSDC)重建SPP算法的目标函数;(2)针对SPP算法计算复杂度高的问题,本文利用带有相同类标签的训练样本用于稀疏重构.在ORL库、CAS-PEAL库、IMM库上的大量实验结果验证了算法的有效性.In the field of face recognition, data is often encountered in the high-dimensional, generally leads to dimension disaster problem. In recent years, sparse representation { sparse representation, SR ） in dealing with the face recognition shows certain effectiveness. Then sparse preserving projections （ sparse preserving projections, SPP ） algorithm which is keeping the data structure of sparse representation successfully applied to face recognition ~but there are still some problems. An improved version of sparsity preserving projection （ SPP} named discriminant sparsity preserving projection （DSPP） is proposed in this paper. DSPP has the following two aspects to improve： （ 1 ） To overcome the SPP algorithm failed to effectively use the class label information problem,this paper reconstructs objective function of the SPP algorithm by using the maximum scatter difference criterion （ Maximum scatter difference criterion, MS- DC ） ; （ 2 ） In order to solve the problem of high computational complexity of the SPP algorithm, we use the training samples with the same label as the current samples for sparse reconstruction. Extensive experiments on three face image datasets （ORL,CAS-PEAL and IMM） demonstrate the effectiveness of the proposed DSPP method.

关 键 词：人脸识别 稀疏表示 稀疏保持投影 鉴别稀疏保持投影 最大散度差准则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]