针对在杂波环境下,一般的高斯混合粒子PHD出现滤波精度不高、滤波发散的问题,提出了一种基于无迹变换的高斯混合粒子PHD。该算法在高斯混合粒子PHD预测的基础之上,采用无迹变换进行重要性采样,结合观测值对采样粒子进行更新,获得重要性密度函数,然后对PHD进行更新。最后,将该算法与高斯混合粒子PHD进行比较;仿真结果表明,该算法在有效提高高斯混合粒子PHD精度的同时,还能提高系统的稳定性。

Considering lower estimating accuracy and filtering divergence of traditional GMP-PHD algorithm in clutter environment, a modified GMP-PHD based on unscented transform was proposed. On the basis of the GMP-PHD prediction, the algorithm applies unscented transform to importance sampling, updates the sampling particles combined with observation values to get the importance density function, and then update the GMP-PHD function. The performance of the proposed algorithm was compared with traditional GMP-PHD algorithm. The simulation results show that the proposed algorithm can promote not only the accuracy but also the stability of the system.