| 引用本文: | 张艺杰,雷明.量测划分的PHD航迹关联算法[J].哈尔滨工业大学学报,2018,50(4):62.DOI:10.11918/j.issn.0367-6234.201612028 |
| ZHANG Yijie,LEI Ming.PHD track association algorithm based on measurement partition[J].Journal of Harbin Institute of Technology,2018,50(4):62.DOI:10.11918/j.issn.0367-6234.201612028 |
|
| 摘要: |
| 高性能空中机动目标跟踪是现代预警雷达的核心任务之一.为了提高概率假设密度(PHD)滤波器的多目标跟踪性能，提出了一种基于量测划分的PHD航迹关联算法.考虑到传统PHD滤波无法给出目标各自的航迹属性信息，而且当跟踪区域存在大量杂波时，滤波效率严重降低；同时，对于密集邻近多目标跟踪，传统方法还存在不同目标跟踪的量测不匹配问题.针对上述问题，提出了分类检索的椭球门限量测划分方法，并将其应用到改进的高斯项权值重新分配的PHD航迹关联算法中，首先对每一步得到的量测集合通过分类检索的椭球门限方法，划分为已存在目标、新生目标和杂波的量测子集，再分别对各类目标使用对应的量测子集进行跟踪滤波，这样去除了杂波对真实目标的无效更新计算，提高了航迹关联算法的计算效率；其次，通过引入一种新的权值分配规则，调整邻近目标的高斯成分的权值大小，大幅减小了邻近目标的状态提取误差，提高了相邻目标的跟踪估计精度.大量数值仿真表明，所提方法明显改进了滤波计算效率和邻近目标跟踪精度.
|
| 关键词: 多目标跟踪 概率假设密度 量测划分 权值分配 航迹关联 |
| DOI：10.11918/j.issn.0367-6234.201612028 |
| 分类号:TP391 |
| 文献标识码:A |
| 基金项目:国家自然科学基金(61271317)； 航天支撑技术基金(15GFZ-JJ02-07) |
|
| PHD track association algorithm based on measurement partition |
|
ZHANG Yijie,LEI Ming
|
|
(School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai 200240, China)
|
| Abstract: |
| High performance aerial maneuvering target tracking is one of the core tasks of modern early warning radar. To improve the tracking performance of the probability hypothesis density(PHD)filter, a new PHD track association algorithm based on measurement partition is proposed. The traditional PHD filter cannot give the individual trajectories of each target, and the filtering efficiency will decrease drastically when there are a large number of clutters in the tracking environment. Besides, the traditional methods also have measurement mismatch problem for adjacent multi-target tracking. To solve the above problems, the measurement set at each time step can be divided into existing targets, new targets and clutter measurement subsets by the sorting ellipsoidal gate method, which targets can be updated by corresponding measurement subsets, thus reducing redundant computing time and improving computational efficiency. Furthermore, the algorithm can adjust the weight of Gaussian component when targets are close to each other, by introducing a new weight distribution scheme, thus greatly reducing the state extraction error of adjacent targets and improving adjacent targets estimation accuracy. The simulation results show that the proposed method can improve the filtering efficiency and the adjacent targets tracking accuracy.
|
| Key words: multiple target tracking probability hypothesis density measurement partition weight distribution track association |
