中文摘要：

为解决多摄像机协同行人关联问题,提出一种改进的灰度图像直方图匹配算法。利用K-means聚类算法对行人灰度信息进行聚类,获得更准确的主灰度信息;构造一个行人人体模型,将空间信息引入到主要灰度直方图中,对行人各个部分进行加权匹配,在整体和局部保证行人的一致性。实验结果表明,该方法能排除非期望的匹配干扰,其匹配度高于传统方法,减少了具有相似灰度信息的不同行人错误匹配,对行人轻微姿态变化有较好鲁棒性。

英文摘要：

An improved pedestrian matching approach of primary gray histogram was proposed to solve the problem of pedestrian association based on multi-camera collaboration.More accurate major gray was obtained by clustering pedestrian gray information using the K-means cluster method.A pedestrian model was constructed,and spatial information was used to establish primary gray histogram.In this way,pedestrian consistency in the whole and local was ensured.Experimental results show that the presented approach can eliminate undesired gray matching and its match degree is higher than traditional methods.It is capable of reducing false matching with similar information in different pedestrians and has good robustness to slight changes in illumination and pose.