Occluded Face Recognition Based on Double Layers Module Sparsity Difference

摘要: Image recognition with occlusion is one of the popular problems in pattern recognition. This paper partitions images into some modules two layers and sparsity difference used to evaluate occluded modules. The final identification processed on unoccluded by sparse representation. Firstly, we partition four blocks representation performed each block, so block can be obtained; secondly, partitioned again Sparsity small module calculated as first step. Finally, corresponding detect modules; this paper, negative differences are considered selected Experiments AR Yale B database verify robustness effectiveness proposed method.