| 摘要: |
| 目标分群问题也称为群形成过程，是态势估计中的一个重点和难点问题，属于信息融合中的高层融合范畴。现有的目标分群方法大多是将空间位置相近的目标聚为一类，极少有分群方法考虑到进行战场态势分析所需的位置、速度、航向、属性和类型等多维特征。为此，首先根据海战场态势分析需求，基于目标的特征属性定义了多维度欧式距离；然后综合利用分类与密度聚类两种分析方法，提出了一种战场实体目标分群方法,以综合实现战场实体目标的实时分群处理；最后，将该算法应用于一个军用场景的数据集上进行仿真实验。仿真结果表明了该方法在海战场环境中实现目标分群处理的可行性和有效性。 |
| 关键词: 态势分析 多维特征 目标分群 密度聚类 分类分析 |
| DOI： |
|
| 基金项目:山东省自然科学基金项目(ZR2020MF154）；山东省重点研发计划(2020CXGC010701,2020LYS11）；山东省高等学校青创科技支持计划(2019KJN031, 2020KJN006） |
|
| A Battlefield Entity Target Clustering Method Based on Multi-Dimensional Characteristics |
| LI Xueteng,ZHU Luying,LIN Xueyuan,WANG Haipeng,TAN Haiqi |
| (Unit 91827 of the PLA, Weihai, Shandong 264200, China;Department of Electrical and Electronic Engineering, Yantai Nanshan University, Yantai, Shandong 265713, China;Naval Aviation University, Yantai, Shandong 264001, China) |
| Abstract: |
| Target clustering problem, also known as group formation process, is a key and difficult problem in situation estimation and belongs to the high-leve fusion field of information fusion. Most of the existing target clustering methods cluster the targets with similar spatial positions into one class, and only a few of them consider the multi-dimensional characteristics required for battlefield situation analysis, such as the position, speed, course, attribute and type of the target. Therefore, this paper firstly defines the multi-dimensional Euclidean distance based on the characteristic attributes of the target according to the requirements of sea battlefield situation analysis. Then, by using the two analysis methods of classification and density cluster, a battlefield entity target clustering method is proposed to realize the real-time battlefield entity target cluster processing. Finally, the proposed algorithm is applied to a military scenario data set for simulation experiment.The simulation results show the feasibility and effectiveness of this method in target clustering processing in naval battlefield environment. |
| Key words: Situation analysis Multi-dimensional characteristics Target clustering Density clustering Classification analysis |
