摘要
针对未来智能化武器装备面向复杂战场环境下的作战需求,设计了一种鲁棒的长期目标跟踪方法。面对高对抗性、高不确定性、高动态性和强实时性的战场环境,以孪生网络跟踪器为基准跟踪框架,建立目标跟踪正确性判断机制,融合双向特征金字塔框架与重检测机制,构建目标跟踪方法架构,从而实现复杂环境下的长期目标跟踪。实验结果表明,提出的方法相对于基准跟踪方法的平均距离精度提高了6.9%,从而确保作战任务高效、可靠地完成。
A robust long-term target tracking method is designed to meet the operational requirements of future intelligent weapons and equipment under complex battlefield environment.Facing the battlefield environment with high antagonism,high uncertainty,high dynamic and strong real-time,a siamese network tracker is used as the baseline tracking framework,a target tracking correctness judgment mechanism is established,and the bidirectional feature pyramid framework and redetection mechanism are integrated to construct a target tracking method framework,so as to achieve long-term target tracking in complex environment.The experimental results show that the average distance accuracy of the proposed tracking method is improved by 6.9%compared with the that of benchmark tracking method,thus ensuring to complete the combat task sufficiently and reliably.
作者
张雷
何舒文
段晶晶
马增琛
张建伟
ZHANG Lei;HE Shuwen;DUAN Jingjing;MA Zengchen;ZHANG Jianwei(North Automatic Control Technology Institute,Taiyuan 030006,China;Key Laboratory of Intelligent Information Control Technology of Shanxi Province,Taiyuan 030006,China)
出处
《火力与指挥控制》
北大核心
2025年第9期45-53,共9页
Fire Control & Command Control
基金
国防科技创新重点专项基金资助项目。
关键词
复杂战场环境
长期目标跟踪
孪生网络跟踪器
双向特征金字塔
重检测机制
complex battlefield environment
long-term target tracking
siamese network tracker
bidirectional feature pyramid
redetection mechanism
