中文

强杂波下紧邻目标的变分跟踪与重检测:补充材料

信号处理 2024-04-25 v2

摘要

非齐次泊松过程(NHPP)是一种广泛使用的量测模型,允许目标随时间产生多个量测。然而,在紧邻目标密度高且杂波严重的场景下,基于该NHPP模型高效可靠地跟踪多个目标可能十分困难。因此,本文基于通用坐标上升变分滤波框架,提出了一种基于变分贝叶斯关联的NHPP跟踪器(VB-AbNHPP),该跟踪器能够通过可并行化实现高效地完成跟踪、数据关联以及目标和杂波强度的学习。此外,提出了一种变分定位策略,可在极端强杂波下从大范围监视区域中快速重新发现丢失的目标。该策略被集成到VB-AbNHPP跟踪器中,形成了一种能够自动检测并从中断跟踪中恢复的鲁棒方法。在准确性和效率方面,该跟踪器在挑战性场景中相比现有跟踪器表现出改进的跟踪性能。

关键词

引用

@article{arxiv.2309.01774,
  title  = {Variational Tracking and Redetection for Closely-spaced Objects in Heavy Clutter: Supplementary Materials},
  author = {Runze Gan and Qing Li and Simon Godsill},
  journal= {arXiv preprint arXiv:2309.01774},
  year   = {2024}
}

备注

Supplementary Materials, including Appendices C-F, begin on page 25, with pages 1-24 constituting the main article. Key updates from the first arXiv version include: added comparisons of the sum-product-algorithm-based tracker, included Appendix C for association update derivation, and made minor adjustments throughout the article to enhance presentation