Abstract: Small sample size is one of the most challenging problems in face recognition due to the difficulty of sample collection in many real-world applications, and the performance of cooperation representation will be severely affected. Multi-scale patch based on collaborative recognition algorithm can integrate the classification results with different scales effectively, but the sub-block calculations are independent of each other in the classification framework, which ignore the structural relationship among the blocks. The local structure method divides the image into multiple local areas, in which overlapped blocks are distributed in the same linear subspace. The subspace can react the structure relationship of the blocks, and improve the robustness under the circumstance of small sample. This paper proposes Local Structure based Multi-Patch Collaborative Representation（LS_MPCRC）, and the experimental results on YaleB and AR face image databases indicate that the proposed method has a good recognition performance in dealing with small sample size problem.

Key words: face recognition, collaborative representation, small sample size problem, multi-scale patch based collaborative representation, local structure

摘要： 人脸识别在实际应用中，往往存在无法获取足够多的训练样本的情况，而在小样本情况下，协作表示的识别性能会受到严重影响。多尺度块协作表示算法能有效集成不同尺度下的分类结果，但其分类框架中子块的计算是相互独立的，忽略了块之间的结构关系。而局部结构法将图像划分为多个局部区域，每个局部区域的重叠块分布在相同的线性子空间中，该子空间可以反应块之间的结构关系，能提高多尺度块协作表示在小样本下的鲁棒性。因此提出了基于局部结构的多尺度块协同表示算法（Local Structure based Multi-Patch Collaborative Representation，LS_MPCRC），在Yale B和AR人脸库上的实验结果证明，该算法在训练样本数目较少时具有优秀的识别性能。

关键词: 人脸识别, 协作表示, 小样本问题, 多尺度块协作表示, 局部结构