一种基于信息融合的粒子滤波跟踪算法

摘要

Aiming at the problem of tracking failure based on traditional algorithm in complex background, this paper proposed a particle filter tracking algorithm based on information fusion. This method combined targets' gray scale with gradient, which efficiently improved the reliability of target description. On this basis, the method deduced the likelihood function of information fusion to fuse the two features, so that the fusion algorithm selectd proper fusion weight for each cue in terms of current situation , achieved the goal of complementing each other's advantages. Experimental results show that the proposed method has better robust and tracking accuracy.%针对粒子滤波在复杂背景下容易造成跟踪目标丢失的问题,提出一种基于多特征信息融合的粒子滤波算法.该方法同时利用灰度和梯度信息描述目标,有效提高了复杂场景下对目标描述的可靠性；在此基础上,推导出多信息融合的观测似然函数,将两种信息融合在一起,使得融合算法能根据当前跟踪形势自适应调整各信息的加权,实现了信息间的优势互补.实验结果表明,该算法鲁棒性较高,明显提高了跟踪精度.