基于无迹粒子PHD滤波的序贯融合算法

针对在杂波、漏检和非线性情况下，粒子概率假设密度滤波（particle probability hypothesis density filter, P-PHDF）算法估计精度不高、滤波发散及粒子退化等问题，提出了一种基于无迹粒子概率假设密度滤波（unscented particle PHDF, UP-PHDF）的序贯融合算法。利用无迹粒子滤波（unscented particle filter, UPF）实现PHDF，由UKF算法得到更好更优的重要性密度函数并从中采样，使粒子的分布更接近多目标概率假设密度分布；另外，为进一步提高滤波算法的性能，实现基于雷达和红外传感器的UP-PHDF序贯融合算法，通过两传感器交替滤波保证目标状态的可观测性。在复杂环境下，仿真结果表明该算法的估计精度和稳定性明显优于单传感器P-PHDF算法。

Sequential fusion algorithm based on unscented particle probability hypothesis density filter

In the case of clutter,missed detections and no-linear,the single sensor particle probability hypothesis density filter(P-PHDF) algorithm will result in many problems,such as low accurate,filter divergence and particle degradation.To overcome these problems,an unscented PHD filter(UP-PHDF) for multi-sensor multi-target tracking based on sequential fusion is proposed.Firstly,the unscented particle filter(UPF) is employed to fulfill PHDF,and the unscented Kalman filter(UKF) method is applied to generate and s...