基于多源信息融合的巡飞弹对地目标识别与毁伤评估  

Ground Target Recognition and Damage Assessment of Patrol Missiles Based on Multi-source Information Fusion

作　　者：徐艺博 于清华 王炎娟[2] 郭策 冯世如 卢惠民 Xu Yibo;Yu Qinghua;Wang Yanjuan;Guo Ce;Feng Shiru;Lu Huimin(College of Intelligence Science and Technology,National University of Defense Technology,Changsha 410073,China;Beijing Aerospace Control Center,Beijing 100094,China;Shenzhen TFC Technology co.,Ltd,Shenzhen 518000,China)

机构地区：[1]国防科技大学智能科学学院,湖南长沙410073 [2]北京航天飞行控制中心,北京100094 [3]深圳天富创科技有限公司,广东深圳518000

出　　处：《系统仿真学报》2024年第2期511-521,共11页Journal of System Simulation

摘　　要：面向利用多枚巡飞弹对地面高防御移动目标进行打击的任务场景,提出一种基于多源信息融合的巡飞弹对地移动目标识别与毁伤评估方法。基于IoU判定实现红外图像与可见光图像的多源信息融合;提出一种基于YOLO-VGGNet的两阶段紧耦合的巡飞弹对地移动目标毁伤评估方法,利用卷积神经网络深度语义信息提取的优势,引入红外毁伤信息,实现对地面移动目标的在线实时毁伤评估。。实验结果表明:基于多源信息融合的目标识别算法有效提升了巡飞弹对地面移动目标识别的有效性;基于YOLO-VGGNet的在线实时毁伤等级评估方法较传统基于图像变化检测与基于两阶段卷积神经网络的方法评估准确率分别提升19%和10.25%。For the multiple patrol missiles to attack the high defense capacity targets,a mobile ground target detection and damage assessment method based on multi-source information fusion is proposed.The multi-source information fusion of infrared images and RGB images is carried out by using IoU determination.A novel two-stage tightly coupled damage assessment method based on YOLO-VGGNet of patrol missiles to mobile ground targets is proposed.This method can fully use the advantage of deep semantic information extraction of CNNs and introduce the infrared damaging information simultaneously to achieve the online and real-time damage assessment of mobile ground targets.The results of simulation experiments show that the target recognition algorithm based on multi-source information fusion significantly improves the detection of mobile ground targets of patrol missiles.Compared with the traditional image-change-detection-based method and the two-CNN learning-based method,the real-time and online damage level assessment method based on YOLO-VGGNet improves the accuracy by 19%and 10.3%,respectively.

关 键 词：多源信息融合 毁伤评估 卷积神经网络 YOLO-VGGNet 在线实时评估 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]