PHD滤波器在多目标检测前跟踪中的应用

摘要

检测前跟踪(TBD)用于对低信噪比目标的雷达检测与跟踪.同时,传统的概率假设密度(PHD)滤波器是解决多目标跟踪问题的有效方法,但它不适用于多目标TBD问题.本文通过分析多目标跟踪问题中PHD滤波器的适用模型和假设,提出了针对TBD的“标准”多目标观测模型,并对噪声进行了“泊松化”,设计出一种能解决多目标TBD问题的PHD滤波器,从而使得PHD滤波器可以应用在多目标TBD问题之中,并充分发挥其处理多目标问题的优势.数值仿真结果表明,本文算法在估计的准确度和精度上都要优于经典的多目标粒子滤波器.%Tracking-before-detectian (1BD) is well suitable far radar detection and target tracking of low-observable objects. Probability hypothesis density (PHD) filter is regarded as an efficient solution to multitarget tracking problem.However,PHD filter is hard to use in multitarget TBD problem directly. By discussing the applicable model and hypothesis,a "standard" multitarget measurement model for TBD and "Poisson" noise are presented. Consequently, a PHD filter application to multitarget TBD problem, with analytical weighting coefficient,is deduced and can exploit the power of PHD fully. Numerical simulations show our approach has better performance than multitarget particle filter.