Study Of The High Precision Dynamic Measurement Method Of Contact Rail Geometric Parameters

In recent years, urban rail transit has developed rapidly. In its power-supplying system, contact rail has been widely used. Because contact rail features low cost and low failure rate and beautiful characteristics. And the normal operation of the contact rail is the important indicator to ensure normal taking-flow. However, when the contact rail deviates from its normal position, the locomotive will be influenced in reliable taking-flow. And sometimes arcing occurs or the collecting shoe crashes the rail. As the contact rail is long and closed, measurement by person cannot guarantee the comprehensiveness and accuracy. So suing automatic testing device to measure geometric parameters of the contact rail is of great significance.This paper combined computer vision measurement technology and digital image processing technology to detect the contact rail. The detect system used laser camera image sensor to collect the image of contact rail. And through the triangle ranging to get the parameter. During the detection, image acquisition module combined with vibration compensation module, vehicle positioning module to precisely and quickly detect the geometric parameters of the contact rail.To solve the problem of poor accuracy and large time-consuming, this paper used comprehensive test to find corners. On the basis of analyzing the superiority and the inferiority of the current corner extraction algorithms, the comprehensive test algorithm combined the quickness of Harris method with the accuracy of Forstner method. And it used iterative and bilinear interpolation to detect corners. And through experiment, the reliability and accuracy has been verified. To save manpower, this paper found a corner automatic sorting method by the characteristic of corners, and in Matlab corners were sorted.In regard to the image distortion caused by camera lens distortion, this paper used distortion separation method to get the distortion coefficient. The laser plane perspective model is set up by transformation relations of the coordinate systems. And the internal parameters can be obtained through least square method after using calibration point data.Light bar of laser’intensity distributes like Gaussian. Based on this, this paper used image segmentation, image cropping and image smooth methods to improve the efficiency of image processing and reduce the image noise. This paper used edge detection operator to extract the laser image edge. And the light has been positioned and the computation has been reduced. Finally, this paper achieved accurate laser light extraction through sub-pixel algorithm, and improved the detection system precision.Through analyzing the test data, this paper has studied the quality of contact rail detection system on behalf of precision and reliability. Compare the data from person with the data from detection system, the detection system’ precision can reach to±3mm. And through different tests the system got the same data.