A person re-identification algorithm based on pyramid color topology feature

Due to illumination variations, person re-identification algorithms based on color features are not robust in practical applications. Different persons may have similar statistical distribution in terms of color histogram, while same person may have different statistical distributions. On the other hand, the color spatial distribution can remain stable, even for illumination variations. In our previous work, a new feature, namely the weighted color topology (WCT), is proposed to describe the color spatial distribution. However, since WCT is extracted on a single scale of region-division, it only represents part of color spatial distribution information. In this paper, a pyramid color topology (PCT) is proposed to extract WCTs on different scales of region-division. PCT can achieve a full description of multi-scales topology information. Based on PCT, a person re-identification algorithm is implemented. The experimental results demonstrate that the proposed algorithm can improve the recognition performance compared with both WCT and the state-of-the-art algorithms.