The Design And Research Of Track Flatness Detection

After we stepped into the 21 st century, the railway transportation in our country has a great improvement. No matter the national railway system or the urban subway system, they all have been in a fast expansion. It has become more and more important to establish the nondestructive tests precisely and effectively on railway tracks. The traditional methods are not too much precise and there is an urgent need for a new detecting method. Recently, with the rapidly development of computer vision and laser detecting technology, they have show great potential in lots of application fields, such as biological medicine, industrial production, and so on. But there still exists some defeats. During the choice of light, laser is better than normal light, how to establish it. During the process of image collecting, how to calibrate the camera and how to control it precisely are to be improved. During the image process, the edge detection method can be improved due to the gray features of track surface images.To cure the above problems, this paper presents a track flatness detecting system based on computer vision theory by using laser light and high speed CCD camera. This system mainly includes the hardware that conveys the workbench which collects the track image and the software that controls the whole system. The mainly research contents in this paper are as follows:(1) Design the image collecting plan, system hardware platform and the motion control plan, design the PLC and the master computer control programs, in order to finish the real time image collection in high speed and high definition.(2) Transform the images under four different coordinate systems, and based on this we accomplished the calibration of camera internal parameters and external parameters by using the chess board.(3) Design the image process algorithm for track flatness images, mainly include using the histogram equation to enhance the dynamic range, median filtering to remove the Gaussian noise, propose and accomplish a new edge detection method based on ACO algorithm. And we accomplish the image matching based on the edge information; fulfill the 3D rebuild based on the triangle measure theory, and extract the flatness deep information to get the flatness detection done.This paper simulates the real situation of large track detecting trucks, verifies the system overall performances and the effectiveness of the algorithm. The input image definition of the system is comparatively high and the detection precision has been improved correspondingly. The new method proposed in this paper has shown a better performance compared with the traditional algorithms in detecting results, which in all will have some contributions to track flatness detection and the follow-up overall researches on the total detections of the tracks.