Person Re-identification by Features Fusion

Person re-identification has an important role in the public security. The methods based on body segmentation have a good performance in distinguishing different people, but it will bring lots of calculation work. To solve this problem, a method combined color distribution fields and Weber Local Descriptor (WLD) is proposed. The method here takes advantage of the color distribution fields to extract the color information; and differential excitation in WLD to extract the texture information in images; and the gradient orientation to extract by orientation in WLD. The experiments show the method in this paper can achieve a good identification results in common data sets.