一种新的未知杂波环境下的PHD滤波器

发布时间：2018-06-16 12:41

本文选题：多目标跟踪 + 概率假设密度　；参考：《西安电子科技大学学报》2014年05期

【摘要】：针对多目标跟踪中概率假设密度(PHD)滤波器在杂波模型与先验知识不匹配情况下滤波性能急剧下降的缺点,将增广状态空间引入PHD滤波器,提出了一种新的未知杂波环境下的PHD滤波器.该滤波器利用增广状态空间区分目标状态空间与杂波状态空间,通过量测对杂波模型进行估计,不需要杂波先验知识,避免了因杂波强度的先验知识选择不当而造成PHD滤波器跟踪性能下降的问题.仿真结果表明,该算法在未知杂波环境下,具有稳定的跟踪效果;在保证实时性的前提下,其跟踪精度与传统PHD滤波器在杂波模型匹配情况下相当.
[Abstract]:Aiming at the shortcoming of the probability assumption density (PHD) filter in multi-target tracking that the filtering performance of the filter drops sharply when the clutter model does not match the prior knowledge, the augmented state space is introduced into the PhD filter. A new PhD filter in unknown clutter environment is proposed. The filter uses the augmented state space to distinguish the target state space from the clutter state space, and estimates the clutter model by overmeasurement without the need of prior clutter knowledge. The problem of poor tracking performance of PhD filter caused by improper prior knowledge selection of clutter intensity is avoided. Simulation results show that the algorithm has a stable tracking effect in unknown clutter environment, and its tracking accuracy is equivalent to that of traditional PhD filter in clutter model matching under the premise of ensuring real-time performance.
【作者单位】：西安电子科技大学电子工程学院;中国人民解放军95972部队;
【基金】：国家自然科学基金资助项目(61372003);国家自然科学基金青年基金资助项目(61101246,61301289) 中央高校基本科研业务费专项资金资助项目(K5051202014) 国家留学基金资助项目(201206965015)
【分类号】：TN953

【参考文献】