Abstract

In this paper, we address the person re-identification problem, discovering the correct matches for a probe per-son image from a set of gallery person images. We fol-low the learning-to-rank methodology and learn a similari-ty function to maximize the difference between the similarity scores of matched and unmatched images for a same per-son. We introduce at least three contributions to person re-identification. First, we present an explicit polynomi-al kernel feature map, which is capable of characterizing the similarity information of all pairs of patches between t-wo images, called soft-patch-matching, instead of greedily keeping only the best matched patch, and thus more robust. Second, we introduce a mixture of linear similarity func-tions that is able to discover different soft-patch-matching patterns. Last, we introduce a negative semi-definite regu-larization over a subset of the weights in the similarity func-tion, which is motivated by the connection between explicit polynomial kernel feature map and the Mahalanobis dis-tance, as well as the sparsity constraint over the parameters to avoid over-fitting. Experimental results over three public benchmarks demonstrate the superiority of our approach. 1 1.