顾祥岐,,
徐从安,
崔亚奇
海军航空大学信息融合研究所 烟台 264001
基金项目:国家自然科学基金(91538201, 61790550)
详细信息
作者简介:熊伟：男，1978年生，教授，博士生导师，研究方向为多传感器信息融合
顾祥岐：男，1995年生，硕士，研究方向为信息融合、雷达数据处理
徐从安：男，1987年生，讲师，研究方向为信息融合、大数据技术
崔亚奇：男，1987年生，讲师，研究方向为深度学习、多传感器信息融合
通讯作者:顾祥岐 guxiangqi1314@163.com
中图分类号:TN953计量
文章访问数:2550
HTML全文浏览量:598
PDF下载量:34
被引次数:0
出版历程
收稿日期:2019-07-08
修回日期:2020-03-22
网络出版日期:2020-04-09
刊出日期:2020-07-23
Tracking Method without Prior Information when Multi-group Targets Appear Successively
Wei XIONG,Xiangqi GU,,
Congan XU,
Yaqi CUI
Institute of Information Fusion, Naval Aeronautical University, Yantai 264001, China
Funds:The National Natural Science Foundation of China (91538201, 61790550)
摘要
摘要:针对多编队机动目标先后出现时的跟踪问题，该文提出了一种基于交互式多模型高斯混合概率假设密度滤波(IMM-GM-PHD)算法的无先验信息跟踪方法。首先，在IMM-GM-PHD算法预测过程完成的基础上，引入密度检测机制，利用相关域为所有预测高斯分量挑选有效量测，结合密度聚类(DBSCAN)算法检测是否出现新编队目标。其次，在IMM-GM-PHD算法状态更新完成的基础上，利用更新高斯分量的组成情况完成模型概率的更新。最后，在状态估计优化过程中，结合编队目标的特点，加入相似度判别技术，利用杰森-香农(JS)散度度量高斯分量间的相似度，剔除没有相似分量的高斯分量，进一步优化估计结果。仿真结果表明，该文方法能够快速有效地跟踪非同时出现的多编队机动目标，具有较好的跟踪性能。
关键词:多编队机动目标/
交互式多模型高斯混合概率假设密度滤波算法/
相关域/
密度聚类算法/
杰森-香农散度
Abstract:Considering the problem of multi-group maneuvering target tracking, a fast tracking method based on Interactive Multiple Maneuvering Gaussian Mixture Probability Hypothesis Density (IMM-GM-PHD) algorithm is proposed. Firstly, based on the completion of the IMM-GM-PHD algorithm prediction process, the density detection mechanism is added, and the correlation domain is used to select effective measurement for all predicted Gaussian components, and the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is combined to detect whether a new formation target appears. Secondly, based on the completion of the state update of the IMM-GM-PHD algorithm, the update of the model probability is completed by updating the composition of the Gaussian component. Finally, in the process of state estimation optimization, combined with the characteristics of formation targets, the similarity discrimination technique is added, and the Jensen-Shannon (JS) divergence is used to measure the similarity between Gaussian components, and the Gaussian components without similar components are eliminated, and the estimation results are further optimized. The simulation results show that the proposed algorithm can track multi-group maneuvering targets quickly and effectively, and has better tracking performance.
Key words:Multi-group maneuvering target/
Interactive Multiple Maneuvering Gaussian Mixture Probability Hypothesis Density (IMM-GM-PHD) algorithm/
Correlation domain/
Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm/
Jensen-Shannon (JS) divergence
PDF全文下载地址:
https://jeit.ac.cn/article/exportPdf?id=daf9b233-97ac-45b9-b1e0-0ba841ea7ef2
