基于PPP模型的多扩展目标跟踪的JPDA算法研究  

JPDA Algorithm for Multi-Extended Target Tracking Based on PPP Model

作　　者：杜浩翠 谢维信[1] 范建德 Du Haocui;Xie Weixin;Fan Jiande(ATR Key Laboratory, Shenzhen University, Shenzhen, Guangdong 518060, China)

机构地区：[1]深圳大学ATR国防科技重点实验室

出　　处：《信号处理》2019年第6期1079-1087,共9页Journal of Signal Processing

基　　金：深圳市基础研究项目(JCYJ20170818102503604)

摘　　要：针对多扩展目标跟踪问题,提出了基于泊松点过程( Poisson Point Process, PPP )模型的多扩展目标跟踪的联合概率数据关联( Joint Probabilistic Data Association, JPDA )算法。首先,采用PPP对扩展目标进行测量建模,其次以“多对一”关联模型思想提出一种JPDA算法,从而计算运动目标的当前有效量测的边缘关联概率,然后结合该边缘关联概率以概率数据关联( Probability Data Association, PDA )的方式分别更新每个扩展目标的运动参数和形状参数向量,最后通过仿真实现了当扩展目标相互靠近或出现交叉时的跟踪。实验结果表明,在高杂波环境下,本文所提出的算法在计算时间和跟踪稳定上具有较明显的优势。A Joint Probabilistic Data Association (JPDA) Algorithm based on Poisson Point Process (PPP) Model is proposed for multiple extended target tracking issues. Firstly, we employ the PPP model for extended object measurement modeling. Secondly, a JPDA algorithm is proposed based on the “many-2-1” association model idea, and then marginal association probabilities can be calculated by the current effective measurement of the moving target. Thirdly, the kinematic and shape parameters of each extended target are updated separately in a probability data association (PDA) fashion of incorporating the marginal association probabilities. Finally, the simulation is used to achieve tracking when two targets whose trajectories cross and two spatially close trajectories. The simulation experiments show that the proposed algorithm has obvious advantages in computational time and tracking stability in high clutter environment.

关 键 词：泊松点过程 多扩展目标跟踪 联合概率数据关联 “多对一”关联模型 边缘关联概率 形状参数 

分 类 号：TN713[电子电信—电路与系统]