一种改进粒子滤波目标跟踪算法  

A novel particle filtering algorithm for target tracking

作　　者：李鹏[1] 肖兵[2] 

机构地区：[1]空军雷达学院研究生管理大队,武汉430019 [2]空军雷达学院预警监视情报系,武汉430019

出　　处：《信息技术》2010年第4期98-100,共3页Information Technology

摘　　要：使用粒子滤波进行目标跟踪时,是用状态转移概率作为重要性函数,由于没有利用最新的观测信息,使得滤波效果不够理想。针对粒子滤波算法的这一缺陷,提出了一种改进粒子滤波算法,该算法利用平方根无迹变换获得重要性函数,解决了粒子滤波算法以状态转移概率作为重要性函数的不足,同时保证了滤波过程中的稳定性。理论分析与仿真实例结果均表明,该算法提高了滤波的稳定性和精确性。When a target is tracked with particle filter,the state transition probability is used as the importance distribution function. Its effect is not good because of not including the recent observation. Aiming at the shortcoming of PF,a novel algorithm was proposed.This algorithm generates a proposal distribution by square-root unscented transform method and draws samples from it,and solves the various problems caused by traditional particle filter using transition prior density function as proposal distribution. The theory analysis and simulation results demonstrate that the novel PF algorithm improves stability and accuracy of filter.

关 键 词：粒子滤波 平方根无迹变换 目标跟踪 

分 类 号：TN953[电子电信—信号与信息处理]