基于颜色和纹理特征的输电线路锈蚀区域检测  

Transmission Line Rusted Area Detection Scheme Based on Color and Texture Features

作　　者：戴玉静 吕东辉[1] 郭松鸽 

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《工业控制计算机》2018年第9期39-40,43,共3页Industrial Control Computer

基　　金：国家自然科学基金资助项目(U1736120)

摘　　要：输电线路中部件锈蚀会危害输电线路的安全运行。针对输电线路背景复杂、缺乏有效锈蚀缺陷检测手段的问题,提出了基于颜色和纹理特征的输电线路锈蚀区域检测方法。根据锈蚀图像的颜色特征与纹理特征,采用HSI颜色模型和灰度共生矩阵识别图像中的锈蚀区域。该方案在判断图像中是否存在锈蚀时正确率为100%,召回率为94%;在识别锈蚀区域时正确率为93.10%,错误率只有6.33%。实验结果表明,所提方法具有高效、准确与非人工检测的优点。This paper proposes a detection method for rusted areas of transmission lines based on dual characteristics of color and texture.According to the rusted color features and texture features,the HSI color model and the gray level co-occurrence matrix are used to identify the rusted areas in the image.This scheme judges whether the correct rate of rust in the image reaches 100%,the recall rate reaches 94%,the rust rate of the identified rusted area reaches 93.10%,and the error rate is as low as 6.33%.

关 键 词：输电线路 锈蚀 颜色特征 纹理特征 HSI颜色模型 灰度共生矩阵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]