基于RPCA与低秩投影的有遮挡人脸识别  

Robust Principal Component Analysis and Low-Rank Projection for Face Recognition with Occlusion

作　　者：曾璐[1] 鲁海荣[1] 罗璐[1] 杨国亮[1] 

机构地区：[1]江西理工大学电气工程与自动化学院,江西赣州341000

出　　处：《计算机仿真》2015年第10期420-425,共6页Computer Simulation

基　　金：国家自然科学基金项目(51365017;61305019);江西省科技厅青年科学基金(20132bab211032)

摘　　要：由于光照的变化、遮挡、阴影、及图像噪声等因素的影响,使得传统方法人脸识别效果不理想,针对人脸识别中遮挡问题,提出了一种基于RPCA和低秩投影的人脸识别算法。先利用RPCA把每类训练样本分解成低秩数据矩阵和稀疏误差矩阵,构建每类原始人脸图像矩阵和低秩数据矩阵之间的低秩映射矩阵。将任意幅测试图像通过每类低秩映射矩阵进行投影,得到测试图像在不同类下的低秩数据矩阵和稀疏误差矩阵,为提取稀疏误差矩阵包含的人脸类别信息,对误差图像计算平滑度和边缘检测,设计了利用平滑度和边缘检测的加权和的判别规则。在AR人脸数据库和Extended Yale B数据库上进行实验,表明了提出的方法的有效性,识别精度得到了有效的提高。Traditional face recognition method which affected by varying illumination, occlusion, disguise, as well as shadow and noise, runs an unstable performance. In this paper, we proposed an algorithm which combined Robust Principal Component Analysis （RPCA） and low -rank projection for face recognition. We divided each type of training sample into sum of low - rank matrix and sparse error matrix by RPCA, and constructed low - rank projection between training face matrix and the separated low - rank matrix. Then each test face image can obtain a low - rank matrix and a sparse error matrix corresponding to each face category by low - rank projection. In order to obtain discriminating information of the sparse error matrix, we calculated the smoothness and carried out the edge detection for the sparse error image, what is more, we made sum of weighted smoothness and edge information a criterion for classification. Experiment results based on face database of AR and Extended Yale B testify the effectiveness of the proposed method with an improved recognition rate.

关 键 词：低秩投影 稀疏误差矩阵 平滑度 边缘检测 人脸识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]