数字图像去雾技术在变电站雾天图像应用研究  

Research and Application of Digital Image Dehazing Technology in Foggy Images of Substations

作　　者：陆世豪 潘勇斌 祝云[2] 邓厚兵 戎春园 LU Shihao;PAN Yongbin;ZHU Yun;DENG Houbing;RONG Chunyuan(Nanning Monitoring Center of EHV Power Transmission Company,CSG,Nanning 530028,Guangxi,China;Guangxi Key Laboratory of Power System Optimization and Energy Technology(Guangxi University),Nanning 530004,Guangxi,China)

机构地区：[1]中国南方电网有限责任公司超高压输电公司南宁监控中心,广西南宁530028 [2]广西电力系统最优化与节能技术重点实验室(广西大学),广西南宁530004

出　　处：《电力大数据》2023年第8期1-9,共9页Power Systems and Big Data

摘　　要：随着变电站从有人值守向无人值守转变,运行人员更普遍地利用摄像头、机器人采集变电站表计、开关等设备的图像,以了解设备运行情况。然而,在雾天条件下,变电站采集的图像存在能见度低、不清晰问题,导致远程监控和操作无法有效开展,增加了电网运行安全风险。针对此问题,本文对雾天变电站图像进行了研究。首先梳理了电力系统中图像去雾算法的应用情况;然后总结了变电站雾天图像的特点,图像背景主要以灰、黄、黑、白等色彩为主;接着介绍了几种基于图像复原和图像增强技术的去雾算法的原理,并选取实际采集的变电站雾天图像,从主观和客观评价两方面出发,综合对比分析了几种去雾算法在变电站雾天图像去雾能力的优劣;最后结合变电站雾天图像特点及图像去雾技术发展趋势,提出了未来的发展方向,包括引入深度学习算法、提高算法效率等,进一步提高去雾效果、提高泛化能力等方面的改进。As substations transition from manned to unmanned operation,operators increasingly rely on cameras and robots to capture images of equipment such as meters and switches in substations to understand their operational status.However,in foggy conditions,the images captured in substations suffer from low visibility and blurriness,which hinders effective remote monitoring and operation,thereby increasing the operational safety risks of the power grid.In response to this issue,this paper conducts a study on foggy images in substations.Firstly,it reviews the application of image dehazing algorithms in power systems.Then,it summarizes the characteristics of foggy images in substations,where the image backgrounds are predominantly composed of gray,yellow,black,and white colors.Next,it introduces the principles of several dehazing algorithms based on image restoration and enhancement techniques.Using actual collected foggy images from substations as the basis,it comprehensively compares and analyzes the strengths and weaknesses of various dehazing algorithms in terms of their effectiveness in dehazing foggy images in substations,considering both subjective and objective evaluations.Finally,in conjunction with the characteristics of foggy images in substations and the development trends in image dehazing technology,future directions are proposed,including the introduction of deep learning algorithms and the improvement of algorithm efficiency,aiming to further enhance dehazing effects and generalization capabilities.

关 键 词：变电站 图像去雾 图像复原 图像增强 机器视觉 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]