A dimensionality reduction method based on structured sparse representation for face recognition

摘要

Face recognition (FR) has been one of the most fundamental problems in computer vision. Two issues are always concerned in a FR task: one is the dimensionality reduction (DR) of the features, and the other is the sparse representation for the samples. DR is an important step because it can not only reduce the storage space of face images, but also enhance the discrimination of the features. Meanwhile, sparse representation based classification (SRC) has been proved a powerful method to solve the problem of dimensionality. It simply considers the training samples as the dictionary to represent the testing samples. However, most of the SRC algorithms do not consider the structure of the dictionary. To consider these two aspects, in this paper, we proposed a FR method by combining a new DR model with the structured sparse representation (SSR). The key idea is projecting the images on a learned projection matrix, and performing the face classification by the SSR considering the structure information of the dictionary. The validity of the proposed method is verified by the evaluations on three databases.