A Larger Coefficient Emphasis Framework for Sparse Coding in Face Recognition

摘要

Sparse representation based classification for face recognition has become a very popular topic in these years. In this paper, a test signal was represented as a sparse linear combination of the predefined dictionary with the sparse coefficients. A novel framework for the image reconstruction with sparse coding was proposed. It filtered the redundancy coding coefficients by selecting a number of largest coding coefficients called Larger Coding Coefficient Emphasis (LCE) to generate the new coding residual. So the novel coding residual was used to reconstruct the test image instead of the standard residual. This larger coefficient emphasis framework, which improves Sparse Representation Based Classification (SRC) and Robust Sparse Coding (RSC), is evaluated on the AR, extended Yale B and FERET face databases and the experiment results show its practical advantages compared with that of SRC and RSC in the face recognition.