摘要
基于稀疏表示的分类方法SRC与基于协同表示的分类方法 CRC分别通过L1范数和L2范数最小化获得具有稀疏性的线性表示系数,在人脸识别中取得了很好的效果。为了解决这两种方法没有考虑数据局部信息的问题,提出了基于局部表示的分类方法 LRC。LRC使用测试样本局部范围内的训练样本对其进行线性表示,这样获得的局部表示系数在保持稀疏性的同时包含有效的局部信息。另外,通过求解一简单的约束最优化问题,LRC可快速获取局部表示系数。在ORL、YALE以及FERET人脸数据库上的实验结果,表明了LRC的有效性和高效性。
SRC(Sparse Representation based Classification)and CRC(Collaborative Representation based Classification)achieve linear representation coefficients with sparsity by Lx norm and L2 norm minimization respectively and perform very well in face recognition.However,SRC and CRC do not consider the local information of the data.To solve this problem,this paper proposes LRC(Local Rep-resentation based Classification).LRC uses training samples in the local scope of the test sample to re-present it and obtains local representation coefficients.The local representation coefficients keep sparsity and contain effective local information.LRC can find the local representation coefficients quickly by sol-ving a simple constrained optimization problem.Experimental results on ORL,YALE and FERET face databases demonstrate the effectiveness and efficiency of LRC.
作者
殷俊
杨万扣
YIN Jun;YANG Wan-kou(College of Information Engineering,Shanghai Maritime University,Shanghai 201306;Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education(Nanjing University of Science and Technology),Nanjing 210094;School of Automation,Southeast University,Nanjing 210096,China)
出处
《计算机工程与科学》
CSCD
北大核心
2018年第3期500-506,共7页
Computer Engineering & Science
基金
国家自然科学基金(61473086
61603243)
上海市自然科学基金(13ZR1455600)
高维信息智能感知与系统教育部重点实验室创新基金(JYB201607)
关键词
稀疏表示
协同表示
局部表示
分类
sparse representation
collaborative representation
local representation
classification
