多帧高斯混合概率假设密度的多目标跟踪算法

针对低检测概率下多目标跟踪时,概率假设密度滤波器难以正确估计当前目标个数以及目标状态问题,提出一种基于多帧融合的高斯混合概率假设密度滤波算法。根据不同时刻目标权值构造目标多帧权值记录集及目标状态抽取标志。当某些时刻目标被漏检时,依据目标状态抽取标志,并结合目标多帧权值记录集中权值信息估计丢失目标的状态。仿真实验表明,算法有效地提高了低检测概率下现有相关算法的目标状态和数目估计精度。

Multi-frame Gaussian Mixture Probability Hypothesis Density Algorithm for Multi-target Tracking

For the problem of the probability hypothesis density (PHD) incapable of correctly estimating target states and their number in low probability of detection multi-target tracking scenario, a multi-frame fusion-based Gaussian mixture PHD algorithm is proposed. According to the weights of targets at each time step,both the multi-frame weight recorder and state extraction flag of individual targets are constructed. When the targets are undetected at some times,the states of these lost targets are extracted based on multi-frame weight recorder and state extraction flag of each target. Simulation results show that the proposed algorithm effectively improve the performance of the existing related algorithms in terms of target states and their number in low detection probability scenarios.