基于图像特征的高速铁路轨道焊缝缺陷检测  

High-speed railway track weld defect detection based on image features

作　　者：张祯祥 Zhang Zhenxiang

机构地区：[1]中铁十八局集团有限公司投资公司,天津300000

出　　处：《现代城市轨道交通》2023年第7期27-31,共5页Modern Urban Transit

摘　　要：针对传统焊缝缺陷检测方法误差大、速度慢和主观性强等问题,提出基于图像特征的高速铁路轨道焊缝缺陷检测技术。采集高速铁路轨道焊缝缺陷图像,通过图像降噪、图像增强环节预处理焊缝缺陷图像,提取焊缝缺陷图像的几何特征与纹理特征,基于特征驱动构建焊缝缺陷检测模型。文章所提方法应用于实际钢轨打磨工程后的试验结果表明,此方法平均检测精度较高、性能较好。Aiming at the issues of large error,slow and strong subjectivity of traditional welding defect detection methods,a welding defect detection method for high-speed railway track based on image features is proposed.This study collects the high-speed railway track weld defect images,preprocesses the weld defect image through image noise reduction and image enhancement,extracts the geometric and texture features of the weld defect images,and builds a weld defect detection model based on feature driving.The proposed method is applied to the actual rail grinding project.The experimental results show that this method has high average accuracy and good performance.

关 键 词：高速铁路轨道 焊缝 缺陷 检测 图像特征 

分 类 号：TG409[金属学及工艺—焊接]