针对视频序列中目标的跟踪问题，提出了一种基于模版匹配的多目标跟踪 算法。该算法提取目标模版灰度分布特征，并在目标匹配过程中采用Bhattacharyya系数在图像中寻找 相似性最高的区域。仿真结果表明，用该算法进行多目标跟踪时，跟踪性能远优于其它特征匹配算 法和多模型算法，而且计算量小，能保证跟踪的实时性。

Aiming at the tracking problem of targets in video sequence images, a multi-target tracking algorithm based on region matching is proposed. This algorithm can extract the grey level distribution features in a target region. In the matching process of targets, it uses the Bhattacharyya modulus to search the most similar regions in images. The simulation result shows that the tracking performance of this algorithm is much better than that of other feature matching and multi-model algorithms. The algorithm has a small computation load and can ensure the real-time performance in target tracking.