融合快速稀疏描述与协同描述的人脸识别

Integrating Fast Sparse Respresentation and Collaborative Representation for Face Recognition

快速稀疏描述分类法(FSRC)与协同描述分类法(CRC)是在压缩感知理论的基础上发展而来的,不同的侧重点限制了两者在人脸识别上的进一步提升。针对此,提出了融合快速稀疏描述与协同描述的人脸识别方法。首先,将人脸镜像图像引入样本库;然后,利用FSRC与CRC方法求解残差矩阵;最后,利用加权信息融合的方式将两者的残差矩阵进行权值加和,依据最小值所对应的位置信息求取识别率。公共人脸数据库的实验表明,所提方法优于FSRC,CRC及其他方法。

On the basis of compressed sensing theory,fast sparse representation classification(FSRC)and collaborative representation classification(CRC)were proposed.Different emphases restrict further improvement in face recognition.Focused on this,this paper proposed an improved method named integrated fast sparse representation and collaborative representation.Firstly,the face mirror image is introduced into the sample library.Then,the residuals matrix is solved with FSRC and CRC.Finally,the weights of residuals matrix get a summation by weighted fusion and the recognition rate is obtained according to the minimum value's position information.Experiments on different face databases show that the proposed method can get better recognition performance than FSRC,CRC and others.