基于多尺度Retinex算法的电力巡检可见光遥感图像去雾自动化增强  

Automatic Enhancement of Visible Light Remote Sensing Image Dehazing in Power Inspection Based on Multi-scale Retinex Algorithm

作　　者：陈云龙 姜钦霞 邓飞凤 郑云梅 CHEN Yunlong;JIANG Qinxia;DENG Feifeng;ZHENG Yunmei(Yichun Power Supply Company,State Grid Jiangxi Electric Power Company,Yichun 336000,China;Beijing Jianghe Huiyuan Technology Co.,Ltd.,Beijing 100084,China)

机构地区：[1]国网江西省电力有限公司宜春供电分公司,宜春336000 [2]北京江河惠远科技有限公司,北京100084

出　　处：《自动化与仪表》2024年第12期88-91,95,共5页Automation & Instrumentation

基　　金：国网江西省电力有限公司科技项目(5218M0240005)。

摘　　要：电力巡检可见光遥感图像往往受到大气层粒子的影响,而大气环境的不稳定性对于图像的去雾过程有着较大的影响,导致处理后图像的峰值信噪比水平不高。为了缓解这一问题,提出了基于多尺度Retinex算法的电力巡检可见光遥感图像去雾自动化增强方法。根据可见光遥感图像的成像原理建立大气散射模型。滤波处理后,采用二维傅里叶变换的方法分离雾区像素点,并通过非线性映射处理实现去雾过程。在此基础上,引入多尺度Retinex算法实现图像的自动化增强处理。经过测试可知,该方法在实践应用中表现出了较高的峰值信噪比水平,图像处理质量较高,满足了电力巡检可见光遥感图像的应用分析需求。Visible light remote sensing images of power inspection are often affected by atmospheric particles,and the instability of the atmospheric environment has a significant impact on the dehazing process of the images,resulting in a low peak signal-to-noise ratio level of the processed images.To alleviate this problem,a multi-scale Retinex algorithm based automatic dehazing enhancement method for visible light remote sensing images of power inspection is proposed.Establish an atmospheric scattering model based on the imaging principle of visible light remote sensing images.After filtering,the pixels in the fog area are separated using a two-dimensional Fourier transform method,and the dehazing process is achieved through nonlinear mapping processing.On this basis,the multi-scale Retinex algorithm is introduced to achieve automatic image enhancement processing.After testing,it is known that this method has shown a high peak signal-to-noise ratio level and high image processing quality in practical applications,meeting the application analysis requirements of visible light remote sensing images for power inspection.

关 键 词：图像去雾 图像自动化增强 遥感图像 多尺度RETINEX算法 可见光遥感图像 电力巡检 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]