改进的交互式多模型粒子滤波目标跟踪算法  

A Target Tracking Algorithm Based on Improved Interacting Multiple Model Particle Filter

作　　者：吕铁军[1,2] 蒋宏[1] 梁国威[3] 丁全心[3] 

机构地区：[1]北京航空航天大学飞行器控制一体化技术国防科技重点实验室 [2]94683部队 [3]火力控制技术国防科技重点实验室

出　　处：《弹箭与制导学报》2013年第3期9-11,16,共4页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：航空科学基金(20125151028)资助

摘　　要：针对现有的交互式多模型粒子滤波算法存在粒子退化现象,跟踪性能不高,计算量大影响跟踪的实时性等问题,采用改进的残差重抽样算法,并在滤波前后,对各模型粒子进行输入输出交互运算,得出一种改进的交互式多模型粒子滤波目标跟踪算法。该算法在解决了粒子退化现象的同时,避免了残留粒子重采样问题,在一定程度上降低了计算量,减小了系统估计误差,提高了跟踪性能。通过仿真,验证了该算法的良好性能。In view of the particles degeneracy phenomenon,not perfecting tracking performance and the high computational complexity which lower the real-time performance of tracking in the exacting interacting multiple model particle filter algorithm,an improved residual re-sampling algorithm is used and the particles of all models interact with computing before and after filtering,finally a new target tracking algorithm based on improved interacting multiple model particle filter is obtained.The algorithm not only solved the problems of particles degeneracy phenomenon and residual particles re-sampling,but improved the tracking performance,together with a lower computational complexity.Simulation results confirmed the proposed IMMPF algorithm.

关 键 词：交互式多模型 粒子滤波 目标跟踪 IMMPF 

分 类 号：TN953[电子电信—信号与信息处理]