论文部分内容阅读
用多激光线代替单激光线应用于钢轨表面缺陷在线检测,解决了单激光线模式下由于钢轨跳动造成“伪缺陷”的问题.对相机采集到的多激光线单幅图像进行了中心线提取、差值算法检测、缺陷判别及缺陷图像拼接等处理,实现了缺陷的在线检测,并获得缺陷的完整区域图像.该方法利用多激光线单幅图像光带间的互相关和自相关深度信息提取图像的特征区域进行检测,避免了单激光线模式的深度拼接步骤,从根本上解决了由于钢轨跳动造成的“误检”问题.该方法已经在线应用于钢轨轨底的表面缺陷检测,相对于单激光线检测具有更高的准确率.
Multi-laser line instead of single laser line is applied to on-line detection of rail surface defects, which solves the problem of “fake defect ” caused by rail jump in single laser line mode. The multi-laser line single image collected by the camera is centered Line extraction, differential algorithm detection, defect identification and defect image stitching, etc. The defect detection is realized online and a complete image of the defect is obtained. This method uses the cross-correlation and auto-correlation between single light bands of multiple laser lines, Depth information to detect the feature area of the image to avoid the single laser line mode of the deep splicing step, fundamentally solve the problem of “false detection” caused by the rail jitter.This method has been applied online on the surface of the rail rail bottom Defect detection, relative to the single laser line detection has a higher accuracy.