摘要
多数稀疏表示方法需要原子数目远远大于原子维数的大规模冗余字典,并采用l1-范数最小化方法来计算稀疏系数。为了降低算法复杂度,提出一种基于稀疏表示的快速l2-范数人脸识别方法。通过提取融合特征和缩小字典规模来改善字典结构,增强l2-范数的稀疏性,从而在保证识别性能的前提下大幅提高算法运行速度。实验表明,与其他稀疏表示方法相比,该方法可以显著降低算法复杂度,同时可以保持良好的人脸识别率和排除干扰人脸的能力。
In recent years,sparse representation has been widely used for face recognition and achieved good results. But most sparse representation methods require a redundant dictionary that the number of atoms in dictionary is much larger than the dimension of it,and they ensure sparsity by solving l1-norm minimization. Both of the procedures will increase the complexity of the algorithm. In order to improve computation speed,this paper proposed a fast face recognition method with regularized least square via sparse representation,namely fast sparse representation classification with regularized least square( FSRC_RLS). It improved the structure of the dictionary and enhanced the sparsity of l2-norm through extracting fusion features and reducing the scale of the dictionary. The extensive experiments demonstrate that FSRC_RLS can improve the computation speed significantly compared with other sparse representation methods,while ensuring the recognition performance and the ability of rejecting distractor faces.
作者
汤镇宇
孟凡荣
王志晓
Tang Zhenyu;Meng Fanrong;Wang Zhixiao(School of Computer Science & Technology, China University of Mining & Technology, Xuzhou Jiangsu 221116 , China)
出处
《计算机应用研究》
CSCD
北大核心
2016年第9期2831-2836,共6页
Application Research of Computers
基金
国家"863"计划资助项目(2012AA0622022
2012AA011004)
国家教育部博士点基金资助项目(20100095110003
20110095110010)
中央高校基本科研业务费资金资助项目(2013XK10)
国家自然科学基金资助项目(61402482)
国家自然科学基金煤炭联合基金重点项目(U1261201)
关键词
人脸识别
稀疏表示
特征融合
字典缩减
正则化最小二乘法
face recognition
sparse representation
feature fusion
reduced dictionary
regularized least square