| With the rapid development of high-speed railways and overloaded railways in our country, it's more and more important for rail wear and rail surface defects detection. The traditional manual testing can not meet the needs of detection in current situation, we must adopt a new efficient detection method.The nondestructive measurement of rail wear is realized based on the vision metrology and image processing method in this paper. The Zhang plane calibration method is selected to calibrate the camera from some commonly used camera calibration methods, which can solve the problem of perspective distortion in rail profile images.The algorithm of rail profile image processing is designed, and the steps of threshold segmentation, image thinning, contour reduction, image registration and wear calculation are deeply studied. After the verification of the algorithm, based on Visual C++ and OpenCV, the software of rail wear measurement is developed by using modular programming. Experimental results show that the rail wear can be well calculated by the software, and the accuracy can reach the sub pixel level.Meanwhile, in order to improve the engineering of the project, a real-time rail surface defects detection system has been developed based on secondary development of image acquisition card, and the system is under the foundation of previous efforts. After understanding the work process of the image acquisition card and its software development kit(SDK), the image processing modular is embeded in the SDK to realize the image acquisition and processing in real-time. In order to test the performance of the software, the scene of the fast relative motion of camera and track are simulated in the laboratory. Dynamic test results show that the accuracy of the software reached 94.5% and the average detection time is 20.24 ms. That means the speed of train is about 85.38 km/h in 1mm accuracy resolution and the system can satisfy the requirements of online detection of ordinary railway and subway. |
PDF Full Text Request