改进联合多站多伯努利滤波器的水下多目标跟踪  

Underwater multi-target tracking using improved multi-sensor multi-Bernoulli filter

作　　者：王冠群 张春华[1,2,3] 尹力 李宇[1,2] 张扬帆[1,2] WANG Guanqun;ZHANG Chunhua;YIN Li;LI Yu;ZHANG Yangfan(Institute of Acoustics,Chinese Academy of Sciences,Beijing,100190;Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing,Chinese Academy of Sciences,Beijing,100190;University of Chinese Academy of Sciences,Beijing,100049)

机构地区：[1]中国科学院声学研究所,北京100190 [2]中国科学院先进水下信息技术重点实验室,北京100190 [3]中国科学院大学,北京100049

出　　处：《声学学报》2021年第4期508-518,共11页Acta Acustica

基　　金：国防基础科研计划重大项目(JCKY2016206A003);国防预研项目(3020201020501);中国科学院青年英才计划项目(QNYC201727)资助。

摘　　要：针对水下多站多目标跟踪问题,基于多站多伯努利滤波器思想,提出改进联合多站多伯努利滤波器的水下多目标跟踪方法.该方法有两处改进:首先在目标数变化场景,为了获得新生目标状态,提出基于节点分对组合随机观测新目标生成方法,该方法将观测节点两两分对,并进行观测对间组合,以相同概率随机选择观测对组合来生成新目标信息;其次为了输出连续、稳定航迹,提出基于统计双门限的航迹生成方法,该方法对超过第一门限的检测次数进行累加,并将结果与第二门限比较来控制目标航迹输出。仿真结果表明:(1)基于节点分对组合随机观测新目标生成方法能够正确生成新目标信息,与全局交叉定位目标生成方法相比,算法复杂度降低,并与节点数成线性关系;(2)基于统计双门限航迹生成方法能够实时输出稳定航迹,并且其最优子模式指派(OSPA)距离对比不使用该航迹生成方法时减少20%左右。Aiming at the multi-sensor multi-target tracking problem underwater,an improved method based on multi-sensor multi-Bernoulli filter is proposed.There are two improvements in this method.Firstly,in order to obtain the new target state under the situation of multiple targets going in and out randomly,a target generation method based on node-paired and combined random observation is proposed.This method divides the observed nodes into pairs and performs inter-pair combination.The observation pair combination is randomly selected with the same probability to generate new target information.Secondly,in order to get continuous and stable trace,a trace generation method based on statistical binary threshold is proposed.This method accumulates the number of detections exceeding the first threshold,and compares the result with the second threshold to control the target track output.The simulation results show that:(1)The new target generation method based on node-paired and combined random observation can get new target information correctly.The algorithm complexity is proportional to the sensors number and less than the global cross-target generation method:(2)the tracking method based on statistical binary thresholds trace generation can output a stable trace in real time,and its Optimal Sub-Pattern Assignment(OSPA)distance is reduced about 20%compared with the tracking method that do not use this trace generation.

关 键 词：交叉定位 子模式 多目标跟踪 节点数 实时输出 线性关系 算法复杂度 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置] TN929.3[自动化与计算机技术—控制科学与工程] TN713[电子电信—通信与信息系统]