摘要
无砟轨道表面缺陷自动检测节点设置一般为单一目标结构,受区域限制,导致误检率增加。因此,本文提出基于视觉特征的铁路无砟轨道表面缺陷自动检测方法。根据实际的检测需求及标准,该方法采用多阶的形式采集无砟轨道缺陷特征打破检测区域的限制,设定多阶自动检测节点,建立ROI多维视觉检测矩阵,构建视觉特征无砟轨道表面缺陷自动检测模型,采用伪边缘处理实现自动检测。测试结果表明,与传统计算机视觉技术和传统图像识别技术无砟轨道表面缺陷自动检测组相比,本方法的误检率可以较好地控制在10%以下。
Currently,the setting of automatic detection nodes for surface defects of ballastless track is generally a single target structure,which is subject to regional constraints,leading to an increase in the rate of false detection.Therefore,this paper proposes a visual feature based automatic detection method for surface defects of railway ballastless track.According to the actual detection requirements and standards,this method adopts a multi-step form to collect the defect features of ballastless track,breaking the limits of the detection area,and setting multi-step automatic detection nodes.Establish a multi-dimensional ROI visual detection matrix,construct a visual feature automatic detection model for ballastless track surface defects,and use pseudo edge processing to achieve automatic detection.The test results show that compared with traditional computer vision technology and traditional image recognition technology,the error detection rate of this method can be better controlled below 10%.
作者
张宏伟
ZHANG Hongwei(Rail Transit Branch,Guiyang Vocational and Technical College,Guiyang,Guizhou 550081,China)
出处
《自动化应用》
2023年第15期144-146,共3页
Automation Application
关键词
视觉特征
铁路无砟轨道
轨道缺陷
自动检测
轨道缺陷标定
visual features
railway ballastless track
track defect
automatic detection
track defect calibration
