The Research Of High-speed Railway Catenary Detection System Based On Digital Image Processing

As the rate of china's railway electrification continuous improved and the locomotive speed increased, it is becoming more and more important to detect the locomotive parameters. Contact line height, dynamic pull-out value and the hard port are the three of many detection parameters. They reflect the catenary system's operational status from a side. The traditional detection methods for the three parameters are to install pressure sensors and acceleration sensors on the pantograph and to use the laser range-finder. These traditional detection methods will impact the pantograph's flow characteristics. So we urgently need a new method to adapt the high-speed railway's detection.This paper fully studies the dynamics and development treads of the detection technology, and combines with our existing catenary system's characteristics. At last, we give the non-contact detection method to contact line height, dynamic pull-out value and the hard port. Our method has no effect on the catenary system's flow characteristics. It is actually the secondary development program for the original detection method. Our main work is to process the video, which is got at the original detection system. After the processing, we will get the contact line height, the pull-out value and the hard port.Since the original video recording system is just to play back the operational status of the pantograph in a certain period of time, the installation of the camera position and angle is qualitative. If we want to use the video to detect the parameters, we need to calibrate the camera. And because the video uses natural light for lighting, there will be a great background noise. In this paper, we mainly use the differential operator to pre-process the image, and the method of coarse search, fine search and slope correction will be given combined with the characteristics of the video. And then we use the Matlab simulation tools to simulate the algorithm. At last, we design the detection software with Microsoft MFC framework. The software implements the files' reading and storing. We can also view the dynamic processing results.The paper describes in detail the process of camera calibration, the algorithm of image processing and the overall design of the software program. But we don't have the video, which has been calibrated. So the detection values will be given with pixel. Through the tracking of the detection results frame by frame, and to compare the true value's trend with our pixel value's trend, we can predict that this method is feasible.