摘要
研究目标跟踪算法优化问题,针对高性能要求的目标跟踪中粒子滤波(PF)算法估计精度低、粒子退化的问题,提出了一种无迹粒子滤波(UPF)的改进算法。算法将实时观测信息融入无迹卡尔曼滤波(UKF)的重要性密度函数,并在UKF环节中Sigma点的选择过程中加入可动态调解参数的比例缩放因子,解决采样的非局部效应问题;同时在PF重采样算法中采用系统重采样的方法,根据有效粒子容量设置权值门限判定是否需要进行重采样,形成系统比例对称无迹粒子滤波算法(SPSUPF)。仿真结果表明,SPSUPF可以在耗时相当的情况下有效解决粒子退化的问题并提高滤波精度。
Considering the problem of poor tracking accuracy and particle degradation in the traditional Particle Filter （PF） algorithm for high - performance requirements of the target tracking, discussed a new improved nonlinear filtering algorithm based on Unscented Kalman Particle Filter （UPF）. The algorithm introduced the latest observation information into the Unscented Kalman Filter（UKF） to generate the importance density function in UKF links, added the scaling factor can dynamic mediation parameter to solve the problem of sampling the local effects; and used the system resample in the link of PF resampling. Then, according to the effective particle volume set weight threshold, we determined the need for resampling. Finally, we made a new algorithm called System Proportion Symmetry Un- scented Particle Filter（SPSUPF）. Simulation results show that the SPSUPF can effectively solve the problem of deg- radation of particles in the particle filter and improve filtering accuracy when the time - consuming considerable.
出处
《计算机仿真》
CSCD
北大核心
2014年第2期336-339,377,共5页
Computer Simulation
基金
国家自然科学基金(60875025/f030410)
