中文摘要：

针对人脸图像中表情变化、遮挡、光照的问题,本文提出了一种新颖的基于低秩分块稀疏表示的人脸识别算法。该算法采用了一种新的结构不相关的低秩矩阵恢复方法,同时采用离散余弦变换方法联合处理人脸图像中遮挡、掩饰和光照的问题,对处理过的图片采用一种独特的重叠分块方法,利用冗余信息有效地提高了算法的识别率。在分类阶段,利用Alignment pooling的方法,有效地提高了识别速度。该算法在标准人脸数据库上进行了多次实验,实验结果表明：与现有人脸识别算法相比,算法的识别准确率和计算效率都得到了一致提高。

英文摘要：

Aiming at the problem of human faces with varying expression and illumination,as well as occlusion and disguise,a face recognition algorithm is proposed based on local structural sparse representation.This algorithm combines low-rank matrix recovery with structural incoherence and discrete cosine transform（DCT）method to remove occlusion,disguise and illumination variations in face image.Meanwhile,the partial information is fully utilized by using sparse codes of local image patches with spatial layout.In the classification stage,the algorithm effectively improves the recognition rate based on a novel alignment pooling method.Extensive experiments are conducted on publicly available face databases.Compared with the related state-of-the-art methods,the experimental results demonstrate the accuracy and efficiency of the proposed method.