基于粒子滤波器的多机动目标跟踪贝叶斯滤波算法研究

A Bayesian Approach to Tracking Maneuvering Multiple Targets Using Particle Filters

提出了一种新的基于粒子滤波器的贝叶斯滤波算法,用于在非线性非高斯假设下跟踪多机动目标.对目标动态行为的已知描述构成了贝叶斯的先验知识.近来时序蒙特卡罗技术的发展,特别是粒子滤波器算法,使采用一个目标状态的集合对贝叶斯模型的后验知识进行建模和跟踪成为可能,这个集合可以看作是这个后验密度函数的采样集合.这种新的贝叶斯滤波算法是粒子滤波器与划分采样技术和假设计算的有机结合.在与SIR/MCJPDA算法的比较仿真研究中,证明该算法能够提高系统的跟踪性能.

A new Bayesian approach to tracking multiple maneuvering targets under nonlinear and nonGaussian assumptions is presented. The prior is a description of the dynamic behavior we expect for the targets. Resent advances in sequential Monte Carlo techniques, specifically particle filter algorithm,allow us to model and track the posterior distribution defined by Bayesian model using a collection of targets states that can be viewed as samples from the posterior of interest. The proposed algorithm is a combination of the partition sampling technique and hypothesis calculations with the particle filter. In a simulation comparison with SIR/MCJPDA, it is proved that our approach yields performance improvements.