基于支持向量机的铁塔构件防腐涂层缺陷识别  

Defect identification of anti-corrosion coating of tower components based on the support vector machine

作　　者：陈诚 李布辉 张庆 周昊 李殿亮 CHEN Cheng;LI Buhui;ZHANG Qing;ZHOU Hao;LI Dianliang(Jiangsu Power Design Institute Co.,Ltd,China Energy Engineering Group,Nanjing 211102,China)

机构地区：[1]中国能源建设集团江苏省电力设计院有限公司,江苏南京211102

出　　处：《武汉大学学报(工学版)》2024年第S2期197-201,共5页Engineering Journal of Wuhan University

摘　　要：输电杆塔在电力行业内广泛运用,其安全性鉴定主要从结构本体与防腐涂层两方面进行展开。为提高鉴定效率,快速且准确识别防腐涂层剥落缺陷,结合图像处理技术、图像特征提取技术和支持向量机分类算法进行输电杆塔镀锌层剥落缺陷自动检测。在图像处理阶段,采用高斯滤波去除噪声同时保留图像细节,基于自适应阈值分割法对图像进行背景分离;在特征提取阶段,使用边缘检测算子识别缺陷轮廓,并从几何特征与纹理特征两方面提取8个缺陷图像轮廓特征;在缺陷分类阶段,基于支持向量机分类算法构建图像缺陷分类器,识别图像中的缺陷部位,整体识别率为77%,最后输出缺陷图像轮廓特征的因子重要度。本文所采用的方法能够快速识别铁塔镀锌层剥落缺陷。尤其在较小的缺陷识别上,远高于现有检测效率。Transmission towers are widely used in the power industry,and their safety assessment is mainly carried out from two aspects:the structural body and anti-corrosion coating.In order to improve identification efficiency,this paper quickly and accurately identifies peeling defects of anti-corrosion coatings,image processing technology,image feature extraction technology,and support vector machine classification algorithm are combined to automatically detect peeling defects of galvanized coatings on transmission towers.In the image processing stage,Gaussian filtering is used to remove noise and preserve image details,and the image is segmented based on adaptive threshold segmentation method.In the feature extraction stage,8 defect image features are extracted from both geometric and texture features,and their Fisher scores are calculated.Finally,all feature vectors are used as inputs,and a multi-classification model is built based on the support vector machine algorithm for recognition,with a recognition rate of 77%.The method used has a certain degree of recognition for the rapid identification of peeling defects in the galvanized layer of iron towers.

关 键 词：镀锌层缺陷 图像处理 特征提取 支持向量机分类器 

分 类 号：TM762[电气工程—电力系统及自动化]