| 一种基于IS-MCMC的多目标跟踪算法 |
| |
| 引用本文： | 龙云利,徐晖,安玮.一种基于IS-MCMC的多目标跟踪算法[J].宇航学报,2011(9). |
| |
| 作者姓名： | 龙云利 徐晖 安玮 |
| |
| 作者单位： | 国防科学技术大学电子科学与工程学院; |
| |
| 摘 要： | 针对密集杂波环境下的多目标跟踪问题,提出了一种基于定向概率数据关联重要度采样的改进马尔可夫链蒙特卡洛(MCMC)算法(IS-MCMC)。首先,基于马尔可夫链蒙特卡洛方法求解具有最大后验概率的量测分配集,并根据量测与目标的定向关联概率进行重要度采样以提高收敛速度。其次,依据量测分配结果更新目标轨迹的感知概率和状态参数,并基于感知概率实现对轨迹起始、维持和终止的判断。为实现对目标的连续跟踪,采用滑窗方法,根据当前帧数据与上一窗口数据的处理结果构建初始量测分配集,利用IS-MCMC对当前加窗数据进行处理。不同杂波密度、目标密度和检测概率下的仿真实验表明,IS-MCMC相比MCMC在计算效率、轨迹跟踪性能和环境适应能力上更具优势。
|
| 关 键 词： | 多目标跟踪 马尔可夫链蒙特卡洛 定向概率数据关联 |
An Algorithm for Tracking Multiple Targets Based on IS-MCMC |
| |
| LONG Yun-li,XU Hui,AN Wei.An Algorithm for Tracking Multiple Targets Based on IS-MCMC[J].Journal of Astronautics,2011(9). |
| |
| Authors: | LONG Yun-li XU Hui AN Wei |
| |
| Institution: | LONG Yun-li,XU Hui,AN Wei(College of Electronic Science and Engineering,NUDT,Changsha 410073,China) |
| |
| Abstract: | An improved Markov Chain Monte Carlo(MCMC) algorithm based on directional probabilistic data association(DPDA) importance sampling(IS-MCMC) is presented for tracking multiple targets in dense clutter circumstance.First,the optimum assignment of total measurements is solved by using the MCMC method,and importance sampling is adopted to improve the convergence according to the directional association probability of the measurements to the targets.Then,states and perceivable ability of targets are updated to i... |
| |
| Keywords: | Multi-target tracking Markov Chain Monte Carlo(MCMC) Directional probabilistic data association(DPDA) |
| 本文献已被 CNKI 等数据库收录！ |
|
