The invention relates to the technical field of image fusion, and provides an image visibility monitoring method and system based on deep fusion. The method comprises the following steps: acquiring scene complexity characteristics of a traffic scene to be subjected to visibility prediction; obtaining passing area parameters of the passing scene, and performing focusing strategy adaptation analysis to obtain a target focusing strategy; controlling a high-position image acquisition device to obtain a focusing passing image; a focal segment fusion channel is pre-constructed, image depth fusion is carried out, and a target multi-focus image is obtained; performing traffic visibility analysis based on the target multi-focus image to obtain a traffic visibility index; and presetting a traffic safety threshold, judging the traffic visibility index by adopting the traffic safety threshold, and generating a short-term traffic analysis result. The technical problems of low image definition and poor visibility judgment condition in the prior art are solved, and the technical effects of improving the definition after image fusion and maintaining traffic safety are achieved.
本申请涉及图像融合技术领域,提供基于深度融合的图像能见度监测方法及系统。所述方法包括:获得待进行能见度预测的通行场景的场景复杂度特征;获得所述通行场景的通行区域参数,进行对焦策略适配分析,获得目标对焦策略;进行高位图像采集装置控制,获得对焦通行图像;预构建焦段融合通道,进行图像深度融合,获得目标多对焦图像;基于所述目标多对焦图像进行通行能见度分析,获得通行能见度指数;预设通行安全阈值,并采用所述通行安全阈值进行所述通行能见度指数判断,生成短期通行分析结果。本申请解决了现有技术中图像清晰度较低,能见度判断条件较差的技术问题,达到了提高图像融合后的清晰度,维护了交通安全的技术效果。
Image visibility monitoring method and system based on deep fusion
基于深度融合的图像能见度监测方法及系统
2024-04-12
Patent
Electronic Resource
Chinese
Highway visibility monitoring and early warning system based on deep learning
European Patent Office | 2022
|VISIBILITY DETERMINING SYSTEM, VISIBILITY DETERMINING METHOD, AND PROGRAM
European Patent Office | 2024
|