电子学报 ›› 2013, Vol. 41 ›› Issue (5): 987-991.DOI: 10.3969/j.issn.0372-2112.2013.05.024

• 科研通信 • 上一篇    下一篇

基于低秩子空间恢复的联合稀疏表示人脸识别算法

胡正平, 李静   

  1. 燕山大学信息科学与工程学院,河北秦皇岛 066004
  • 收稿日期:2012-07-13 修回日期:2013-01-24 出版日期:2013-05-25
    • 作者简介:
    • 胡正平 男,1970年8月生于四川仪陇县, 2007年于哈尔滨工业大学获得信息与通信工程专业博士学位,教授,博士导师,燕山大学通信电子工程系副主任,目前为中国电子学会高级会员,中国图像图形学会高级会员,目前研究方向为模式识别. E-mail:hzp@ysu.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61071199); 河北省自然科学基金 (No.F2010001297)

Face Recognition of Joint Sparse Representation Based on Low-Rank Subspace Recovery

HU Zheng-ping, LI Jing   

  1. School of Information Science and Engineering,Yanshan University,Qinhuangdao,Hebei 066004,China
  • Received:2012-07-13 Revised:2013-01-24 Online:2013-05-25 Published:2013-05-25
    • Supported by:
    • National Natural Science Foundation of China (No.61071199); Natural Science Foundation of Hebei Province,  China (No.F2010001297)

摘要: 针对阴影、反光及遮挡等原因破坏图像低秩结构这一问题,提出基于低秩子空间恢复的联合稀疏表示识别算法.首先将每个个体的所有训练样本图像看作矩阵 D ,将矩阵 D 分解为低秩矩阵 A 和稀疏误差矩阵 E ,其中 A 表示某类个体的'干净’人脸,严格遵循子空间结构, E 表示由阴影、反光、遮挡等引起的误差项,这些误差项破坏了人脸图像的低秩结构.然后用低秩矩阵 A 和误差矩阵 E 构造训练字典,将测试样本表示为低秩矩阵 A 和误差矩阵 E 的联合稀疏线性组合,利用这两部分的稀疏逼近计算残差,进行分类判别.实验证明该稀疏表示识别算法有效,识别精度得到了有效提高.

关键词: 人脸识别, 稀疏表示, 联合稀疏, 低秩子空间恢复

Abstract: In consideration of the cast shadows,specularities,occlusions and corruptions in the images that violate the low-rank structure,a novel recognition method of joint sparse representation based on low-rank subspace recovery is proposed.Firstly,using all training images of each class to form a data matrix D,we can decompose D as the sum of a low-rank matrix A and a sparse error matrix E,where A denotes the"clean"images which follow strictly the low-rank subspace structure and E accounts for cast shadows,specularities,occlusions and corruptions in the images that violate the low-rank structure.Then the test sample can be represented as the linear combination of dictionary which is composed of low rank matrix and error matrix,using the sparse approximation of this two parts calculates the residual which used for classification.Experiment results show that the algorithm is effective ,and effectively improve the recognition accuracy.

Key words: face recognition, sparse representation, joint sparse, low-rank subspace recovery

中图分类号: