基于深度学习的激光干扰成像评估方法

可见光主动成像系统被广泛应用于成像侦察领域，激光压制干扰是一种可以利用的反制手段。合理地对激光干扰效果进行评估，对于干扰一方具有重要意义。为了更客观地评价激光干扰对成像侦察的影响，从可见光成像侦察的特点出发，对激光干扰可见光成像效果进行划分，确定了干扰效果评估需求。建立了可见光图像的激光干扰效果评估模型，结合基于YOLOv4的目标检测置信度及WFSIM算法的三个图像特征表征指标，形成了对应的评估体系。构建了一套参数可调的激光干扰图像采集及评估分析实验系统，对人体目标进行图像采集，验证了提出的评估体系及方法。

Evaluation method of laser disturbing imagingbased on deep learning

Visible light active imaging system is widely used in the field of imaging reconnaissance.In the practical application,oppressive laser dazzling can make a real threat to it.Reasonable evaluation of laser dazzling effect is of great significance for the interference side.In order to more objectively evaluate the impact of laser dazzling on imaging reconnaissance,based on the characteristics of visible light imaging reconnaissance,the laser interference visible light imaging effect is divided and the interference effect assessment requirements are determined.A laser interference effect evaluation model for visible images is established,and the corresponding evaluation system is formed by combining the target recognition confidence of YOLOv4 and three image feature indexes of the WFSIM algorithm.Thus,an experimental system with adjustable parameters for laser interference image acquisition and evaluation and analysis is constructed,and image acquisition of human targets is performed to validate the proposed evaluation system and method.