DEGREE:一种基于Delaunay三角的任意群目标外形识别方法  

DEGREE:A Delaunay Triangle-Based Approach to Arbitrary Group Target Shape Recognition

作　　者：李天成 严瑞波 成明乐 李固冲 Li Tiancheng;Yan Ruibo;Cheng Mingle;Li Guchong(School of Automation,Northwestern Polytechnical University,Xi’an 710129,China)

机构地区：[1]西北工业大学自动化学院,西安710129

出　　处：《航空兵器》2024年第2期123-130,共8页Aero Weaponry

基　　金：国家自然科学基金项目(62071389,62201316);陕西省自然科学基础研究计划(2023JC-XJ-22);国防技术重点实验室基金(JKWATR-210504);中央高校基本科研业务费专项资金。

摘　　要：集群目标相比单一甚至多目标表现出复杂时变集群特性,其外形估计与评价颇具挑战性。针对任意形状的集群目标外形估计与评价难题,本文提出了一种基于数据驱动的多传感器集群目标群形状建模与识别方法,以及一种群目标外形拟合度评判指标。所提算法由三个部分组成:首先,采用信息洪泛(Flooding)方法实现强连接的多传感器对视场中目标信息的采集与传播;其次,采用密度峰值聚类实现观测数据的聚类;最后,采用改进Delaunay三角网络算法实现群目标外形的拟合。所提群外形拟合度指标可用于对群目标外形估计准确度定量评价。通过与超曲面、随机矩阵等经典方法进行比较,证实了所提出算法的有效性和可靠性。Compared with the single or even multiple targets,the group targets exhibit complex and time-varying structure,making the group shape estimation and evaluation quite challenging.This paper proposes a data-driven multi-sensor target group shape modeling and recognition approach to arbitrary shape estimation for group targets,and a group target shape fitting evaluation metric.The proposed approach consists of three parts.Firstly,the information flooding method is used to realize the collection and dissemination of the target information in the field of view by strongly connected sensors.Secondly,a density peak clustering method is utilized to cluster the data set.Finally,an improved Delaunay triangular network algorithm is used to fit the shape of group targets.The proposed group shape fitting evaluation metric can quantitatively evaluate the accuracy of any group target shape estimate.The effectiveness and reliability of the proposed algorithm are verified in comparison with the classic target shape fitting methods such as the hypersurface and random matrices.

关 键 词：群目标 传感网络 Delaunay三角网络 超曲面 随机矩阵 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]