A Feature Fusion Algorithm for Human Matching between Non-Overlapping Cameras

Human matching is fundamental in human tracking over non-overlapping cameras. Fusing multiple features is an efficient way to increase the ratio of matching. In this paper, we present an algorithm of iterative widening fusion (IWF) to fuse the multiple features, including color histogram, UV chromaticity, major color spectrum histogram and scale-invariant features (SIFT). Also, the Bayesian framework, as a classical fusion method, is compared with the IWF algorithm. The experimental results indicated that the IWF algorithm obtained the matching accuracy better than Bayesian framework in most cases.