一种改进的IMMPF目标跟踪算法

对多运动模型的目标进行跟踪,通常采用传统的交互式多模型粒子滤波算法。但是,该算法存在一些缺陷和不足。为此,提出了一种新的基于变速率模型和遗传算法的IMMPF目标跟踪算法。针对IMMPF算法对目标进行跟踪时可能出现的未知可变转弯速率,采用了一种更恰当的可变速率目标模型;对于IMMPF算法中的粒子多样性丧失问题,则将进化理论中的遗传算法引入到目标跟踪算法中,对采样进行优化,增加了采样粒子的多样性,使采样向后验分布取值较大的区域移动。仿真结果表明,提出的算法能更好的适应目标的机动运动,同时明显减少所需的采样数,取得了更好的跟踪性能。

An Improved IMMPF Object Tracking Algorithm

For tracking the object with multiple mobile models,traditional interactive multiple model particle filter algorithm is usually adopted.However,the performance of this algorithm is not so satisfactory.Thus,a new IMMPF algorithm based on variable rate model and genetic algorithm is proposed.Aiming at the possible unknown turning rate in the use of IMMPF,a more proper variable rate object model is employed.And for the problem of particle diversity comedown occurred in IMMPF algorithm,the genetic algorithm in evolutionary theory is introduced into the object tracking algorithm,this could raise the diversity of sample particles and make samples move towards the regions with large value of posterior density distribution.Simulation results show that the proposed algorithm is more suitable for the objects with maneuvering motion and could remarkably reduce the sample,and thus achieve much better tracking performance.