Human matching approach based on color spatial distribution model across non-overlapping camera views
非重叠监控视频中基于颜色空间分布模型的行人匹配方法

For the human matching requirements, this paper presented a special approach of human matching based on color spatial distribution model across non-overlapping camera views. First, it obtained the MC feature by clustering the target pixels through the nearest cluster method. Secondly, built the color spatial distribution model and transformed the MC feature. Finally, calculated the target pair similarity based on the built color spatial distribution model and got the matching result. Experimental results show that the presented approach can effectively eliminate the influence of the edge pixels which come with the detection and the segmentation. While it maintained a high matching rate to the same target in different cameras, it is capable of coping with the wrong matching by the difference of the color distribution. The results also show that the proposed approach is robust to the difference of the visual angle and the change of the object gesture.