Dimension reduction using collaborative representation reconstruction based projections

Abstract

This paper develops a collaborative representation reconstruction based projections (CRRP) method for dimension reduction. Collaborative representation based classification (CRC) is much faster than sparse representation based classification (SRC) while owning the similar recognition performance to SRC. Both CRC and SRC utilize the class reconstruction error for classification. First, CRRP characterizes the between-class/within-class reconstruction error using collaborative representation; Second, CRRP seeks the projections by maximizing the between-class reconstruction error to the within-class reconstruction error. So the proposed method is called CRRP. The experimental results on AR, Yale B and CMU PIE face databases demonstrate that CRRP is an effective dimension reduction method.