基于区间平滑的粒子概率假设密度滤波算法

Particle Probability Hypothesis Density Filtering Algorithm Based on Interval Smoothing

针对杂波情况下,粒子概率假设密度滤波(P-PHDF)算法存在估计精度低、滤波发散以及粒子退化等问题,提出基于区间平滑的粒子概率假设密度滤波(RTSP-PHDF)算法。该算法利用区间平滑算法(RTS)引入观测值优化重要性密度函数,使粒子分布更逼近多目标概率假设密度的分布,进而得到优化的采样结果,改善滤波性能。仿真结果表明,与P-PHDF算法相比,该算法在有效提高估计精度同时,进一步提高了跟踪系统稳定性。

In the presence of clutters, particle-probability hypothesis density filtering （P-PHDF） algorithm may cause low estimation precision, filter divergence and particle degeneracy, to solve these problems, a Rauch Tung Striebel smoother particle-probability hypothesis density filtering （RTSP-PHDF） algorithm was proposed. This algorithm exploited Rauch Tung Striebel smoother and introduced observation optimization importance density function, making particle distribution closer to multiple targets probability hypothesis density distribution and thus deducing optimized samples with improved filtering performance. The simulation revealed that, in comparison with P-PHDF algorithm, the proposed algorithm had increased estimation precision with further improved stability of tracking system.