自适应目标新生强度的随机集跟踪算法研究

发布时间：2018-11-18 14:53

【摘要】：多目标跟踪问题是信息融合领域的重点和难点,由于具有很高的军用和民用价值,历来受到国内外学者的广泛关注和研究。随着基于随机集理论的多目标跟踪方法研究的深入,多目标跟踪领域得到了快速发展。早期的随机集跟踪方法假设新生目标强度是先验信息,但在真实的复杂场景中,目标新生强度是难以预先获得的。因此,需要在未知目标新生强度的条件下完成多目标的稳定跟踪。本文研究了随机集框架下未知目标新生强度的多目标跟踪问题,主要工作如下:首先,概述了随机集理论的基本概念以及相关滤波算法,详细介绍了PHD和CPHD两种滤波算法,并给出了其在线性高斯条件下的高斯混合实现。其次,介绍了传统的GM目标新生模型,并针对其不足,详细研究了自适应目标新生强度的PHD滤波器。对于检测时杂波和新生目标存在互相制约的问题,介绍了一种目标新生率的估计方法,能够减小杂波对目标新生检测的影响。由于在杂波环境下会出现目标新生时刻的确认滞后现象,不利于后续的航迹关联等处理,本文提出了自适应目标新生强度的PHD平滑器,将后向平滑算法与目标新生率估计相结合,经分析及仿真结果验证,该算法能够更加准确地估计新生目标的状态并获得新生时刻,可得到更好的跟踪效果。最后,研究了自适应目标新生强度的CPHD滤波算法,并结合仿真实验,分析对比了ATBI-CPHD滤波器和ATBI-PHD滤波器的跟踪性能,结果表明,前者对目标数目的估计更加准确。在未知杂波密度的条件下,提出了自适应目标新生强度CPHD滤波器的改进算法,并给出了其高斯混合实现形式。该滤波器能够在杂波密度和目标新生强度都未知的条件下完成多目标的稳定跟踪,不仅可以摆脱对新生目标强度作为先验信息的依赖,并且能够在线估计场景中的杂波密度。通过仿真实验,验证了改进算法的有效性和实用性。
[Abstract]:Multi-target tracking is an important and difficult problem in the field of information fusion. Because of its high military and civilian value, it has always been widely concerned and studied by scholars at home and abroad. With the development of multi-target tracking method based on random set theory, the field of multi-target tracking has been developed rapidly. The early random set tracking method assumes that the intensity of the new target is a priori information, but it is difficult to obtain the intensity of the new target in the real complex scene. Therefore, it is necessary to complete the stable tracking of multiple targets under the condition of unknown target strength. In this paper, we study the multi-target tracking problem of unknown targets in the frame of random set. The main work is as follows: firstly, the basic concepts of random set theory and related filtering algorithms are summarized, and two filtering algorithms, PHD and CPHD, are introduced in detail. Moreover, the mixed realization of Gao Si under the condition of linear Gao Si is given. Secondly, the traditional GM target newborn model is introduced, and the adaptive PHD filter with new strength is studied in detail. This paper introduces a method to estimate the rate of target birth, which can reduce the influence of clutter on the detection of new target. Due to the fact that the confirmation lag of the target birth time will occur in the clutter environment, which is not conducive to the subsequent track correlation processing, a PHD smoother with adaptive target regeneration strength is proposed in this paper. Combining the backward smoothing algorithm with the target birth rate estimation, the analysis and simulation results show that the algorithm can estimate the state of the new target more accurately and obtain the new time, and obtain better tracking effect. Finally, the CPHD filtering algorithm with adaptive target strength is studied, and the tracking performance of ATBI-CPHD filter and ATBI-PHD filter is analyzed and compared with the simulation experiment. The results show that the former is more accurate in estimating the number of targets. Under the condition of unknown clutter density, an improved algorithm of adaptive target freshly intensity CPHD filter is proposed, and its Gao Si hybrid realization form is given. The filter can not only get rid of the dependence on the intensity of the new target as a priori information but also estimate the clutter density in the scene online. The effectiveness and practicability of the improved algorithm are verified by simulation experiments.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713

【参考文献】