一种改进Camshift算法的研究

摘要

针对基于颜色概率分布的连续自适应均值偏移算法Camshift（Continuously Adaptive Meanshift）在背景中出现相同颜色干扰时容易致使跟踪目标失败的问题，提出一种改进的Camshift 跟踪算法。该算法首先采用三帧差分法进行运动目标检测，然后利用最大类间方差法确定分割阈值，最后采用Camshift算法进行目标跟踪。该算法可以克服传统Camshift算法需要人工定位和容易发散的缺点。实验结果表明，该算法跟踪效果好，并且能够在一定程度上满足准确性的要求。%Continuous adaptive mean shift (Camshift)algorithm is based on the colour probability distribution,it may lead to tracking target failure when the background colour is similar to the target colour.In light of this issue,we propose an improved Camshift tracking algorithm.First,it uses three-frame difference method to detect the moving target.Secondly,it determines the segmentation threshold by the Otsu method.At last,it tracks target with Camshift algorithm.The algorithm can overcome the defects of traditional Camshift algorithm that it requires manually positioning and is prone to divergence.Experimental results show that the proposed algorithm has good tracking effect,it can meet the requirements of accuracy to certain extent.