基于压缩感知的人脸识别算法实现  

Implementation of face recognition algorithm based on compressed sensing

作　　者：孙海威[1] 曹曦文 张鑫晟[1] 孔尧[1] 宋雪桦[1] 

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212000

出　　处：《信息技术》2018年第3期155-158,共4页Information Technology

摘　　要：为了可以更好地处理人脸识别中数据量大、高维度、非线性等问题,文中提出了一种基于压缩感知的人脸识别算法。首先采用随机测量矩阵将人脸稀疏表示,其次使用主成分分析法对人脸进行特征提取,最后运用改进的梯度投影法来确定测试人脸图像的分类。实验在ORL人脸数据库中进行验证,表明本文提出的算法具有更短的识别时间和更高的识别率。In order to deal with the problem ol large amount ol data, high dimensionality and nonlinearity in lace recognition,in this paper,a face recognition algorithm based on compressed sensing is proposed. First ly, the sparse face is represented by the random measurement matrix. Secondly, the principal component analysis （PCA） is used to extract the features of the face. F in a l ly, the improved gradient projection method is used to determine the classification of the tested lace images. The experiments on ORL lace database show that the proposed algorithm has shorter recognition time and higher recognition rate.

关 键 词：压缩感知 人脸识别 稀疏表示 梯度投影 主成分分析法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]