基于向量线性表示的集群点目标同一性识别方法
Method of Vector Linear Representation-Based Point Cluster Target Registration
麻家骅 1魏东 2吴楚泽 1齐文元 2李元祥 1杨永胜1
作者信息
- 1. 上海交通大学 航空航天学院·上海·200240
- 2. 上海航天控制技术研究所·上海·201109
- 折叠
摘要
现代战争中,"集群攻击"已经成为一个重要的战术手段.防空导弹武器系统如何实现集群目标探测分辨和导弹选派是集群目标拦截面临的主要技术难题.针对集群目标探测分辨任务,提出结合地面雷达和弹上导引头探测信息的协同分辨策略.在此基础上,探讨向量线性表示与多视角间坐标系线性变换的关系,并提出了基于向量线性表示的集群点目标描述方法,以实现多视图间点目标的同一性识别.通过模拟无人机集群飞行态势进行仿真验证,结果表明,该方法在理想情况下对无人机这类稀疏集群目标的同一性识别准确率高达99.6％;在视图相对位姿误差引起拓扑畸变的情况下,其同一性识别准确率在99％以上;在上游弱小目标识别错误不改变集群有效拓扑的前提下,同一性识别准确率高达95％.
Abstract
Cluster attacks have become an important means in modern warfare. However,the main technical chal-lenges faced by cluster target interception were how to detect and plan to intercept them. In order to detect cluster targets,this paper fused images from ground radars and missiles,and localized targets using multi-view stereo. In view of this,we explored the relationship between vector linear representation and linear transformation of coordi-nate systems among multiple views firstly,and then proposed a target description method based on vector linear representation which can be used to identify point targets in clusters effectively. The experiments showed that the accuracy of identity for sparse cluster targets was as high as 99.6％ under ideal conditions. And in the case,where the existing topological distortion was caused by relative pose errors,the accuracy of identity was over 99％. What is more,with errors from upstream small and weak target recognition tasks,the accuracy of identity recognition can reach up to 95％ under the premise of no change in the available topology of the cluster.
关键词
稀疏集群目标/多视角/同一性识别/向量线性表示Key words
sparse cluster targets/multi-view/identity recognition/vector linear representation引用本文复制引用
基金项目
上海航天先进技术联合研究基金项目(USCAST2022-35)
出版年
2024
国家科技期刊平台
NSTL
万方数据