基于自适应遗传PHD滤波的多群目标跟踪方法

Multi-group-target tracking approach based on adaptive genetic PHD filter

针对标准的概率假设密度(PHD)滤波算法在杂波环境下对群目标跟踪误差较大的问题,提出一种基于自适应遗传PHD滤波的多群目标跟踪方法。该方法在PHD粒子滤波的基础上,利用选择概率减少了新生粒子的数量。为了有效抽取交叉粒子,在时间更新阶段引入当前量测与群目标间的马氏距离。为了提高预测粒子的鲁棒性,推导出自适应交叉与变异操作方案。仿真实验表明,所提出的方法能有效跟踪杂波环境下的多群目标,具有目标总数估计稳定、运动状态估计准确的特点。

To deal with the problem of larger error in multi-group-target tracking using the standard probability hypothesis density(PHD)filter,this paper presented a multi-group-target tracking approach based on the adaptive genetic PHD filter.Firstly,it reduced the number of newborn particles by using the selection probability.For the sake of extracting crossover particles,this work introduced the Mahalanobis distance between the current measurement and group-target in time-update.As for both diversity and robustness of predicted particles,it derived the adaptive crossover and mutation operations.The simulation experiment indicates that the proposed approach can effectively track multi-group-target with stable cardinality estimation and accurate state estimation.