A Collaborative Sparse Representation-Based Approach for Pattern Classification

Sparse representation-based classification (SRC) has been an important approach in pattern classification and widely applied to various fields of visual recognition. However, there are some practical factors to consider in real-world problems, such as pixel damage, block occlusion, illumination, and position change. In recent years, researchers have brought a large number of improvements and proposed various effective algorithms. We improve the degree of sparse representation (SR) based on the probabilistic collaborative representation framework in this paper. Furthermore, a new algorithm is proposed and successfully applied to face recognition. The effectiveness and efficiency of our method are demonstrated experimentally.