基于无迹变换的概率假设密度滤波算法  

Probability Hypothesis Density Filter Based on Unscented Transformation and Its Application to Multi-target Tracking

作　　者：吕学斌[1] 游志胜[1] 周群彪[1] 陈正茂[1] 蔡葵[1] 

机构地区：[1]四川大学计算机学院图形图像研究所,成都610064

出　　处：《系统仿真学报》2009年第3期845-850,872,共7页Journal of System Simulation

基　　金：Supported by the National High-Tech Research and Development plan of China under Grant(2006AA12A104);国家自然科学基金资助项目(60705005;60572175);教育部博士点科研基金(20070610031)

摘　　要：基于有限集统计理论的概率假设密度滤波算法运用于多目标跟踪时,不再考虑数据关联问题,突破了传统的跟踪方法。但该滤波公式在非线性条件下没有解析解,在非线性高斯条件下提出了基于无迹变换的概率假设密度滤波算法,实现了算法在强杂波环境下的多目标跟踪。仿真实验比较了该算法与基于粒子滤波的概率假设密度滤波算法的跟踪性能,验证了该算法的跟踪性能和精度。同时分析指出了此算法的不足,以及结合无迹变换与粒子滤波的概率假设密度滤波算法的改进研究方向。The Probability Hypothesis Density （PHD） Filter based on Finite Set Statistics doesn＇t need data association for multi-target tracking, which breaks through the tradition tracking method. But there is no closed form solution to the PHD recursion under the nonlinear models. The Probability Hypothesis Density Filter based on Unscented Kalman filter algorithm was proposed for jointly estimating the time-varying number of targets and their states under clutter environment, Simulation result validated UKF-PHD performance and then compared UKF-PHD and PF-PHD performance. Lastly, the algorithm＇s lacks were pointed out and the direction was researched based on unscented particle filter.

关 键 词：概率假设密度滤波 无迹变换 随机集 多目标跟踪 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]