基于图像处理和Hough变换的轮对踏面轮廓检测方法

Tread profile of wheel detection method based image processing and Hough transform

轮对作为列车走行部的重要组成部分,对其踏面轮廓的在线实时检测意义重大。为实现非接触式在线测量,基于图像处理的自动检测方案可以实时准确的检测轮对踏面轮廓,以激光作为光源照射轮对,同时用面阵CCD采集踏面光带图像,通过分析光带图像特点以及Hough变换获取轮对轮廓,提出了对Hough变换方法的改进,包括前期图像预处理、边缘检测和直线圆弧信息聚类存储算法。在实验室模拟了现场试验,分别对改进方法和未改进方法进行了对比试验,同时将基于该方法下提取的轮廓测量的几何参数与传统测量的几何参数结果进行了对比,结果证明提出的方法有效地提高了测量的效率、实时性和准确性。

Wheel is an important part of the vehicle,the wheel tread profile of the online real-time detection is of great significance.To realize the non-contact on-line measurement,automatic detection scheme based on image processing can measure wheel tread profile and geometric parameters timely and accurately.The laser irradiate the wheel,there is a band on the tread of the wheel.It is the wheel tread profile.Through analyzing the image taken by CCD,it can achieve the line and curve information in the tread profile with image processing and Hough transform.Some steps are proposed to improve the method,including image pro-processing、edge detection and clustering storage algorithm.To analyzing the method,taking experiments compare with traditional method.The result proves the method in this paper can detect the wheel tread profile and geometric parameter accurately and effectively.