多无人机协作侦察中的车辆再识别

Vehicle re-identification in multi-UAV collaborative reconnaissance

在多无人机协作侦察中,匹配识别不同无人机侦察发现的装甲车辆,属于车辆再识别任务。近年来,车辆再识别技术在智能交通领域得到了快速发展,但是该技术尚未在军事领域得到充分的关注和应用。本文围绕该技术的发展和军事应用展开研究,一方面,从数据集、特征表示、注意力学习、度量学习、属性学习等方面对民用车辆再识别技术进展进行综述,分析总结了现有技术的优点和劣势,提出了有待深入探索的研究方向;另一方面,以多无人机协作侦察任务为背景,探讨分析了车辆再识别在其中的作战运用方式,并针对战场上存在的计算资源有限、环境复杂多变等因素导致车辆再识别性能下降问题,提出了可能的解决思路。本文研究可为车辆再识别技术在多无人机协作侦察中的应用提供前瞻性参考。

Vehicle re-identification involves confirming whether armored vehicles captured by multi-UAV collaborative reconnaissance are the same.In recent years,vehicle re-identification has been applied to smart transportation.However,the technology has not received sufficient attention in the military field.This paper focuses on advances in vehicle re-identification and its applications for the military.We begin by surveying civilian research on vehicle re-identification in terms of datasets,feature representation,attention learning,metric learning,and attribute learning.We analyze the advantages and shortcomings of existing methods,and propose future research directions.Additionally,in the context of multi-UAV collaborative reconnaissance,we discuss applications for vehicle re-identification technology on the battlefield and suggest possible solutions regarding its limited computational resources and poor performance in complicated environments.We offer a forward-looking discussion on armored vehicle re-identification in multi-UAV collaborative reconnaissance.