Collaborative representation Bayesian face recognition

In recent years, a series of subspace methods, named the collaborative representation based methods, have aroused researchers’ interests. Inspired by the idea of sparse coding,Wright et al. proposed the sparse representation based classification (SRC) method. SRC encodes a query sample as a linear combination of all subjects’ training samples with sparsity constraint and then classifies it by evaluating which class has the minimal coding residual. This mechanism is named Collaborative Representation. When the training dictionary is overcomplete, SRC can achieve stateof- the-art recognition accuracy and is proven robust to occlusion and corruption. Based on SRC, Zhang et al. emphasized the collaborative representation mechanism’s importance and replaced the sparsity constraint by other loose constraints in SRC. A more general method, named collaborative representation based classification (CRC), was then proposed, which achieves similar recognition accuracy as SRC with much higher computational efficiency. CRC consists of two major parts: the collaborative representation model and the residual model.supported by National Natural Science Foundation of China,National Basic Research Program of China (973)中国科学引文数据库(CSCD)4048101-1-048101-36