未知杂波状态下基于箱粒子滤波的PHD算法

A PHD Algorithm Based on Box Particle in Unknown Clutter Environment

针对未知杂波环境中,传统的多目标概率假设密度(PHD)滤波器跟踪精度无法保证,所需粒子支撑集过大导致效率低下的问题,引入了区间分析技术,提出了未知杂波状态下基于箱粒子滤波的PHD算法。该算法首先完成对雷达目标和杂波的混合空间随机有限集模型的构建,然后基于箱粒子滤波技术,利用量测数据建立未知杂波模型,推导出目标状态更新方程,并用多目标箱粒子PHD滤波递推地估计目标状态。仿真实验表明,当杂波环境与先验模型不匹配时,相较于多目标粒子滤波算法,该算法在保证目标跟踪性能的同时,有效减少了算法的计算时间。

In unknown clutter environment, traditional Probability Hypothesis Density （PHD） filter in multi-target tracking cannot guarantee a good performance, and multitude number of particles leads to time consuming and low efficiency. Aiming at the problems？ a new PHD filter tracking algorithm in unknown clutter environ-ment based on interval analysis was proposed. Firstly, radar targets and clutter disjoint union state space mod-eled were established in random finite set. Next, Using measurement model to set up clutter model and derived to multi-target updated state function based on box particles. Additionally, the state of multi-target was recur-sively estimated in utilization of PHD filter box particles. Simulation revealed that the proposed algorithm was a-ble to dramatically lower computational time with better tracking performance compared with traditional box particle filter.