基于ET-GM-PHD的多传感器多目标跟踪算法  

Multisensor multiple target tracking algorithm based on ET-GM-PHD

作　　者：刘丽娟 刘国栋 

机构地区：[1]江南大学 轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《传感器与微系统》2013年第10期126-128,133,共4页Transducer and Microsystem Technologies

摘　　要：针对单传感器跟踪系统的缺陷和扩展多目标跟踪问题的复杂性,提出了基于高斯混合扩展目标概率假设密度(ET-GM-PHD)的多传感器多目标跟踪方法。该算法的最大优点在于它可以避免传统多目标跟踪算法中的数据关联过程,不用进行复杂的测量与扩展目标之间的关联过程即可同时获得扩展目标的个数和状态估计。在线性高斯假设条件下给出了扩展目标PHD滤波器的GM实现,并通过仿真实验比较了单传感器ET-GM-PHD滤波器与多传感器ET-GM-PHD滤波器的跟踪性能,验证了该算法的跟踪性能和精度。Aiming at complexity of single sensor tracking system defects and extended multiple target tracking problem,propose multisensor muhitarget tracking method, based on Gaussian mixture extended target probability hypothesis density（ET-GM-PHD）. The biggest advantage of this algorithm is that it can avoid data association process in traditional multiple target tracking algorithm, need not to carry on process between complicated measurement and extended target correlation can also obtain the number of extended target and state estimation. Under the condition of linear Gaussian hypothesis, GM realization of extended target PHD filter is given, and through simulation experiment, compare tracking performance of single sensor ET-GM-PHD filter and multi-sensor ET-GM-PHD filter, verify tracking performance and precision of the algorithm.

关 键 词：扩展目标 概率假设密度滤波器 高斯混合方法 数据关联 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]