| 摘要: |
| 由于高机动多目标跟踪场景中目标个数未知、运动形态复杂，采用传统的GMPHD滤波算法容易出现跟踪误差大、目标数目估计不准确、目标航迹难以区分等问题。针对以上问题，在GMPHD滤波算法的基础上，提出了一种自适应CS模型的标签化GMPHD滤波算法，借助标签化手段显式区分目标航迹，并对漏检目标航迹外推，同时将适用于单目标的自适应CS模型拓展到机动多目标场景，将目标机动性变化实时反馈到目标状态估计上。通过与扩展GMPHD、平方根容积GMPHD和自适应CS?GMPHD算法的仿真对比实验，验证了所提算法在高机动多目标场景下计算耗时低且跟踪精度高的性能优势。 |
| 关键词: 改进GMPHD滤波器 高机动多目标跟踪 标签化 自适应CS模型 |
| DOI：DOI:10.3969/j.issn.1672-2337.2024.05.002 |
| 分类号:TN953 |
| 基金项目:国家自然科学基金（No.61871307） |
|
| An Improved GMPHD High⁃Maneuverability Multi⁃Target Tracking Algorithm |
|
HAO Weibing， LI Ming
|
|
National Key Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, China
|
| Abstract: |
| Due to the unknown number of targets and complex motion patterns in the high?maneuvering multi?target tracking scene, the traditional GMPHD filtering algorithm is prone to large tracking errors, inaccurate target number estimation and difficult to distinguish target tracks. To solve the above problems, based on GMPHD filtering algorithm, a labeling GMPHD filtering algorithm for adaptive CS model is proposed. By means of labeling, the target track is explicitly distinguished and the missing target track is extrapolated. Meanwhile, the adaptive CS model suitable for single target is extended to maneuvering multi?target scenarios. The maneuverability change of the target is fed back to the target state estimation in real time. Compared with extended GMPHD, square root cubature GMPHD and adaptive CS?GMPHD, the simulation results show that the proposed algorithm has the advantages of low computation time and high tracking accuracy in high maneuvering multi?target scenarios. |
| Key words: improved GMPHD filter high maneuverability multi⁃target tracking labeling adaptive CS model |
