2026年 01期

摘要(Abstract)：

为了实现高质量的输电线路巡检图像水痕去除，有效消除水痕遮盖造成的不利影响，提出一种基于条件生成对抗网络的输电线路巡检图像水痕去除方法；以有水痕原始图像作为输入，在U型网络架构基础上构建结合注意力机制的生成器，以增强生成器对关键水痕特征的聚焦能力，同时通过判别器对无水痕生成图像的真实性进行监督，以提升训练稳定性与无水痕生成图像的质量；按照训练集和测试集中图像对数之比为4∶1划分985对有水痕原始图像与对应无水痕真值图像，并进行水痕去除效果验证及目标检测实验。结果表明：所提出的图像水痕去除方法生成的无水痕生成图像质量显著提升，与无水痕真值图像的结构相似性指数为0.867,比采用pix2pix网络、循环生成对抗网络生成的图像更接近无水痕真值图像；应用于目标检测算法检测鸟巢、悬浮物的平均精确率相比使用有水痕原始图像的分别提升了17.2、 19.1个百分点，有效改善了水痕遮盖的劣质输电线路巡检图像的检测效果。

关键词(KeyWords)：图像水痕去除方法;条件生成对抗网络;注意力机制;输电线路巡检图像;目标检测

基金项目(Foundation):国家自然科学基金项目(61903155)

作者(Author):罗龙,李岩,刘荣,赵云龙,齐鹏文,张梦华

DOI: 10.13349/j.cnki.jdxbn.20251124.001

参考文献(References)：

[1] 刘传洋，吴一全.基于深度学习的输电线路视觉检测方法研究进展[J].中国电机工程学报，2023,43(19):7423.

[2] XU J,ZHAO W,LIU P,et al.Removing rain and snow in a single image using guided filter[C]//2012 IEEE International Conference on Computer Science and Automation Engineering (CSAE),May 25-27,2012,Zhangjiajie,China.New York:IEEE,2012:304.

[3] HAO J Y,JIANG M R,HUANG Y Q,et al.Scatter inpainting algorithm for rain or snow removal in a single image[C]//2018 IEEE 4th International Conference on Computer and Communications (ICCC),December 7-10,2018,Chengdu,China.New York:IEEE,2018:1700.

[4] LEPCHA D C,GOYAL B,DOGRA A,et al.Image super-resolution:a comprehensive review,recent trends,challenges and applications[J].Information Fusion,2023,91:230.

[5] TIAN C W,ZHANG M H,ZUO W M,et al.A cross transformer for image denoising[J].Information Fusion,2024,102:102043.

[6] CHEN Z X,HE Z W,LU Z M.DEA-Net:single image dehazing based on detail-enhanced convolution and content-guided attention[J].IEEE Transactions on Image Processing,2024,33:1002.

[7] SONG Y D,HE Z Q,QIAN H,et al.Vision transformers for single image dehazing[J].IEEE Transactions on Image Processing,2023,32:1927.

[8] ANWAR T,MU C X,YOUSAF M Z,et al.Robust fault detection and classification in power transmission lines via ensemble machine learning models[J].Scientific Reports,2025,15(1):2549.

[9] WANG H,XIE Q,ZHAO Q,et al.RCDNet:an interpretable rain convolutional dictionary network for single image deraining[J].IEEE Transactions on Neural Networks and Learning Systems,2024,35(6):8668.

[10] FU X Y,HUANG J B,ZENG D L,et al.Removing rain from single images via a deep detail network[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),July 21-26,2017,Honolulu,Hawaii,USA.New York:IEEE,2017:1715.

[11] 杨青，于明，付强，等.基于倍频卷积和注意力机制的图像去雨[J].控制与决策，2023,38(12):3372.

[12] XIANG P,WANG L,WU F X,et al.Single-image de-raining with feature-supervised generative adversarial network[J].IEEE Signal Processing Letters,2019,26(5):650.

[13] SHEN Y Y,FENG Y D,WANG W M,et al.MBA-RainGAN:a multi-branch attention generative adversarial network for mixture of rain removal[C]//2022 IEEE International Conference on Acoustics,Speech and Signal Processing (ICASSP),May 23-27,2022,Singapore,Singapore.New York:IEEE,2022:3418.

[14] RAJ N B,VENKETESWARAN N.Single image haze removal using a generative adversarial network[C]//2020 International Conference on Wireless Communications Signal Processing and Networking (WiSPNET),August 4-6,2020,Chennai,India.New York:IEEE,2020:37.

[15] BOROUJENI P H,RAZI A.IC-GAN:an improved conditional generative adversarial network for RGB-to-IR image translation with applications to forest fire monitoring[J].Expert Systems with Applications,2024,238(Part D):121962.

[16] AZAD R,AGHDAM E K,RAULAND A,et al.Medical image segmentation review:the success of U-Net[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2024,46(12):10076.

[17] CHENG J,DENG C J,SU Y Z,et al.Methods and datasets on semantic segmentation for unmanned aerial vehicle remote sensing images:a review[J].ISPRS Journal of Photogrammetry and Remote Sensing,2024,211:1.

[18] 吕振鸣，董绍江，夏宗佑，等.基于改进CycleGAN的多失真类型水下图像增强[J].浙江大学学报(工学版),2025,59(6):1148.

[19] ISOLA P,ZHU J Y,ZHOU T H,et al.Image-to-image translation with conditional adversarial networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),July 21-26,2017,Honolulu,Hawaii,USA.New York:IEEE,2017:1125.

[20] 史加荣，王丹，尚凡华，等.随机梯度下降算法研究进展[J].自动化学报，2021,47(9):2103.

[21] 赵园喜，邱志斌，朱轩，等.基于SSIM-Sobel与多特征融合的绝缘子缺陷识别方法[J].智慧电力，2023,51(12):74.

[22] ZHAI W X,NI Z X,XU Z,et al.T2R-pix2pix:a method for constructing rural thematic road network based on pix2pix[J].Computers and Electronics in Agriculture,2025,230:109911.

[23] TODA R,TERAMOTO A,KONDO M,et al.Lung cancer CT image generation from a free-form sketch using style-based pix2pix for data augmentation[J].Scientific Reports,2022,12(1):12867.

[24] DAYARATHNA S,ISLAM K T,URIBE S,et al.Deep learning based synthesis of MRI,CT and PET:review and analysis[J].Medical Image Analysis,2024,92:103046.

[25] 魏业文，李梅，解园琳，等.基于改进Faster-RCNN的输电线路巡检图像检测[J].电力工程技术，2022,41(2):171.