融合边缘特征的MeanShift跟踪算法

MeanShift Tracking Algorithm with Integrated Edge Feature

经典MeanShift跟踪算法通常用颜色特征进行跟踪,算法简单、速度快、跟踪效果较好,但用单一的颜色特征描述跟踪目标,当背景颜色和目标颜色接近时易受干扰,导致跟踪失败,鲁棒性较差.文中提出了一种融合边缘特征的MeanShift跟踪算法,用颜色和边缘特征来共同描述目标,根据特征对目标与背景的区分能力进行自适应特征选择跟踪.实验结果表明:文中提出的算法抗干扰性能更强,具有更好的鲁棒性.

When tracking color features with classical MeanShift tracking algorithm,it is simple,fast and better,but if a single color feature tracks target and the background color and the target color are similar,it is susceptible to interfere and led to track failures and lower robustness.This paper presents a MeanShift tracking algorithm with integrated edge feature,which describes the target by using color and edge features together.It adapts to selection of features for tracking according to the discrimination of the target and background characteristics.The results show that the proposed algorithm is stronger in anti-interference performance and has better robustness.