基于颜色直方图匹配的非刚性目标跟踪算法

Non-Rigid Target Tracking Based on Color Histogram Matching

基于目标颜色直方图的相关匹配,用Bhattacharyya系数度量模板与候选目标直方图的相似度,结合Kalman预测和新三步搜索算法加快最优匹配的搜索速度,为避免静止背景对直方图统计的影响,加入运动分割信息,并对遮挡情况做出判断和处理.实验结果表明这种算法是十分有效的.

Based on color histogram correlation matching, the authors use Bhattacharyya coefficient to measure the similarity between the template and the possible target＇s histogram, combine Kalman predic tion and new-three-step-search algorithm to quicken the searching of optimal matching. To avoid the background influence to the histogram calculation, moving segmentation information of the target is used, and effective occlusion detection and handling method is also proposed. Experimental results show the good performances of the algorithm.