Towards more reliable matching for person re-identification

Person re-identification is an important problem of matching persons across non-overlapping camera views. However, the re-identification is still far from achieving reliable matching. First, many existing approaches are wholebody- based matching, and how body parts could affect and assist the matching is still not clearly known. Second, the learned similarity measurement/metric is equally used for each pair of probe and gallery images, and the bias of the measurement is not considered. In this paper, we address the above two problems in order to conduct a more reliable matching. More specifically, we propose a reliable integrated matching scheme (IMS), which uses body parts to assist matching of the whole body. Moreover, a sparsity-based confidence is also presented for regulating the learned metric to improve the matching reliability. The experiments conducted on three publicly available datasets confirm that the proposed scheme is effective for person re-identification.