Railway track gauge inspection method based on computer vision

Abstract

Railway track gauge irregularity severely reduces the service life of rail and vehicle, and even result in vehicle falling off rail or wheel trapping, which causes driving accidents. In this paper, a dynamic inspection method of track gauge based on computer vision is presented. According to the method, the inspection system could be constructed by using four CCD (Charge-coupled Device) cameras and two red laser sector lights. The inspection principle and corresponding calibration method of inspection system are analyzed. In order to get the gauge points, several image processing technologies, including image component extraction, differential, adaptive iteration threshold, dilation and thinning are adopted. At last, the experiment results prove that this inspection method is capable of fast obtaining track gauge value with high accuracy and repeatability, and meeting the requirement of dynamic inspection for track gauge.