基于机器视觉的接触网悬挂缺陷检测方法  

Catenary Defect Inspection Method Based on Machine Vision

作　　者：殷超 YIN Chao(China Railway Hohhot Group Co.,Ltd.,Hohhot Inner Mongolia 010050,China)

机构地区：[1]中国铁路呼和浩特局集团有限公司,内蒙古呼和浩特010050

出　　处：《铁路技术创新》2025年第1期150-154,160,共6页Railway Technical Innovation

摘　　要：为有效利用接触网悬挂状态检测监测装置采集的高清现场图像,实现对接触网设备的自动化检测和数字化记录,提出一种基于机器视觉的非接触式接触网悬挂状态检测方法。对接触网悬挂全景图像进行相位校正及灰度处理,并建立以像素为单位的图像坐标系;利用加权中值滤波去除图像噪声干扰,采用基于Canny算子的改进边缘检测算法提取各零部件边缘特征;对图像进行基于相似度量的目标区域粗匹配和基于曲线特征投影的零部件边缘轮廓精细匹配,实现对接触网悬挂螺栓松脱、管帽缺失等缺陷的智能识别。该方法的识别准确率为81.6%,平均识别时间为1.174 s,可在一定程度上代替人工检测。To effectively utilize high-definition images captured by the catenary status inspection and monitoring equipment(referred to as“4C equipment”),a non-contact catenary status inspection method based on machine vision is proposed.This approach enables automatic inspection and digital recording of catenary equipment.The process begins with phase correction and grayscale adjustment of panoramic catenary images,followed by establishing an image coordinate system in pixels.A weighted median filter is employed to eliminate image noise.Additionally,an improved edge inspection algorithm based on the Canny operator is used to extract edge features from each component.To achieve intelligent defect identification,a rough matching of the target area is performed based on similarity measurement,followed by fine matching of component edge contours through curve feature projection.This method facilitates the intelligent identification of defects,such as loose catenary bolts and missing tube caps.The accuracy of this technique is 81.6%,with an average identification time of 1.174 seconds,providing a viable alternative to manual inspections to a significant extent.

关 键 词：接触网悬挂 4C装置 机器视觉 边缘特征提取 缺陷检测 

分 类 号：U226.8[交通运输工程—道路与铁道工程]