摘要
现代战争中,具有侦察、识别、打击、评估多种功能于一体的远程多用途子母弹被广泛应用,随之将在地表产生大量的未爆子弹药,对战场机动、疏散展开及纵深攻击等作战行动产生极大的阻碍作用。地表未爆子弹药因其特殊的可视特性,利用成像技术和深度学习技术进行远距离、大面积、无接触式快速准确检测与识别,在军事上具有重要的应用价值。论文介绍了未爆子弹药检测常用方法及优缺点,概述了未爆子弹药在不同成像技术条件下的成像特点及其识别定位方法,分析了深度学习在未爆子弹药检测与识别领域的显著优势,并提出一种基于深度学习的无人机载式地表未爆子弹药快速检测与识别定位方法。
Long-range multi-purpose cluster munitions with reconnaissance,identification,strike and evaluation functions are widely used,and a large number of unexploded submunitions will be generated on the surface,which will greatly hinder the battlefield maneuver,deployment and depth attack.Due to its special visual characteristics,surface unexploded submunitions use imaging technology and deep learning technology for long-distance,large-area,non-contact rapid and accurate detection and recognition.This paper introduces the common methods of unexploded submunition detection and their advantages and disadvantages,summarizes the imaging characteristics of unexploded submunitions under different imaging technology conditions and their identification and positioning methods,analyzes the signifi-cant advantages of deep learning in the field of unexploded submunition detection and recognition.Then,a UAV-borne rapid detection,identification and positioning method based on deep learning for surface unexploded submunitions is proposed.
作者
陈栋
闫小伟
石胜斌
Chen Dong;Yan Xiaowei;Shi Shengbin(Laboratory of Guidance Control and Information Perception Technology of High Overload Projectiles,PLA Army Academy of Artillery and Air Defense,Hefei 230031,China)
出处
《航空兵器》
CSCD
北大核心
2023年第5期1-10,共10页
Aero Weaponry
基金
省部级基金项目(20191A030124)。
关键词
未爆弹药
深度学习
成像技术
识别定位
unexploded submunition
deep learning
imaging technology
identification and positioning
