基于mean-shift的多目标粒子滤波跟踪算法设计

Multi-target tracking method based on mean-shift and particle filter

针对视频序列中多目标的跟踪问题,提出一种基于mean-shift和粒子滤波的多目标混合跟踪算法,采用目标之间的距离门限和相似度门限作为转换条件,有效地结合Mean-shift算法的简单快速和粒子滤波跟踪算法的抗遮挡的优点。实验结果表明,该算法在对多目标跟踪时能在保证跟踪效果的同时减少跟踪的时间花费,有效提高了设计的跟踪系统的实时性。

Aiming at tracking multi-target in video sequence, a multi-target tracking algorithm based on mean-shift and particle filter is proposed, adopting distance limit and similar degree limit of object, it amalgamates the virtues of the two techniques. Simulation results show that it reduces the computational cost and ensures the performance simultaneously.