【作者】 万毅；

【导师】 明德烈；

【作者基本信息】 华中科技大学 ， 控制科学与工程， 2019， 硕士

【摘要】 现代战争正朝着科技化、信息化的方向不断发展。如何科学地了解战场的变化情况引起了各国军方的高度重视,对敌方目标进行打击效果评估已成为军事行动中必不可少的环节。论文针对战场打击实际情况,在目标变化检测算法、毁伤评估模型等方面进行了深入研究,获得了较好的实验效果。论文主要工作如下:提出了一种分预检测和再检测两部分的图像变化检测算法框架,在预检测部分,通过比较打击前后两张图像纹理特征的相似性来获取差分图像,结合FCM聚类将差分图像分为三类区域。在再检测部分,重点研究了预检测结果中未知类区域的分类问题,提出了基于PCANet模型的变化检测算法。该方法结合预检测中变化类和未变化类区域图像作为训练样本,将训练好的PCANet模型用来实现打击前后图像的变化检测,解决了图像部分实际变化区域难以检测的难题。实验表明论文提出的变化检测算法相比传统算法获得了更好的检测结果。提出了一种结合层级目标体系的模糊评估模型。针对目标毁伤信息种类众多的问题,对不同类型目标建立起层级目标体系知识库,综合统计分析目标的物理性和功能性毁伤信息,全面表征目标毁伤情况。根据目标各结构确定相应的子结构权重以及评估准则,同时基于模糊评估概念对目标毁伤情况进行综合毁伤评估,解决了如何对毁伤评估进行定量分析的难题。实验表明论文提出的毁伤评估方法与专家意见相比获得了较高的相符率结果。基于目标变化检测算法和层级目标体系模糊评估模型,开发了一套打击效果评估软件,该软件具备了对多种典型目标进行打击效果评估的功能。更多还原

【Abstract】 Modern warfare is developing in the direction of high technology and information.How to scientifically and timely understand the changes in the battlefield has aroused the attention of the military of all countries.Battle damage assessment has become an indispensable part of military operations.The paper has carried out in-depth research on the target change detection algorithm and damage assessment model for the actual situation of the battlefield strike,and obtained good experimental results.The details are as follows:An algorithm framework of image change detection algorithm based on pre-detection and re-detection is proposed.In the pre-detection part,This method obtains difference images by comparing the similarity of texture features of two images taken before and after attack.The pixels of the texture difference image are divided into changed,unchanged and unknown classes by FCM algorithm.In the re-detection part,the classification problem of unknown pixel points in pre-detection results is studied,and a change detection algorithm based on PCANet model is proposed.This method combines the images of the changed and unchanged class in the pre-detection to create the sample images to train the PCANet model.Finally,the trained PCANet model is used to detect the change between images taken before and after attack.This method solves the problem that the actual change area of the image part is difficult to detect.Experiments show that the proposed change detection algorithm obtains better results than traditional algorithms.A fuzzy evaluation model combining hierarchical target system is proposed.Aiming at the problems of the target damage information,the target knowledge base of hierarchical target system is established to comprehensively analyze the physical and functional damage information of the target to comprehensively characterize the damage situation.The comprehensive damage assessment of the target based on the concept of fuzzy evaluation solves the problem of how to quantitatively analyze the damage assessment.Experiments show that the damage assessment method proposed by this paper has a high match rate result with expert opinions.Developed the comprehensive damage assessment software based on the change detection algorithm and the fuzzy evaluation model combining hierarchical target system,and it can make the damage assessment of various typical targets.更多还原