Based On Machine Vision Trucks Cross Bar Geometry Detection

Abstract:Cross-bar support bogie’s property is crucial to improve the railway wagon running speed and load capacity.With the development of rail freight in China, cross bar’s repair and maintenance, also become more and more important. However, the existing artificial detection efficiency became inefficiency, it has been difficult to meet the requirement of railway freight transportation high-speed development. Aiming at this problem, this paper proposes a plan about trucks cross bar geometry automatic detection which based on machine vision, that can effectively improve the efficiency of cross-bar overhaul.This paper makes deep analysis on machine vision related theory, and carried out the measurement experiment about the length of cross-bar. Using relationship ensured CCD cameras to take the photos of cross-bar edge, then deal them with median filtering, gray scale conversion,Canny edge detection, and so on. We can get the center of cross-bar edge in pixel coordinates. According to the conversion from imaging model to camera geometry, we can get the center of cross-bar edge in the world coordinate system, which can draw on both ends of the cross-bar, so we can get the length of cross-bar and the difference between two cross-bar. This paper also mentioned the cross-bar flatness detection.In this paper, I made a brief introduction on the basic theory, and processed a detailed description about the selection of hardware and software, processing. Finally, according to the design plan, I did some experiment about cross-bar geometry detection. The results has been analyzed.Experimental results show that this cross-bar geometry detection plan is feasible. This plan can effectively improve the efficiency and precision of cross-bar geometry measurement, and also can do repeat measurement. Strict camera parameter calibration is needed to ensure the accuracy of measurement, it is also the key to the whole machine vision measurement system. In the end, I summed up the work of the whole paper, and pointed out the further work.