Fast Camera Calibration for Rail Profile Measurement Based on Structured Light

The paper presents a fast camera calibration method for rail profile measurement based on computer vision. A fast calibration pin-plate target is designed according to real rail profile. The target pins and rail profile are photographed by structured light. The local least-square method is used to fit horizontal lines and vertical lines near the known target pin points in the image. The landmark is the point of intersection of the two fitting lines, which is unique and accurate. The camera calibration parameters are calculated by physical coordinates and pixel coordinates. The rail profile is measured by the computer vision and the tool minprof respectively, which is used as a measuring standard with higher accuracy to validate the proposed method. The experiment shows that the measured average errors of the rail profile are less than 0.3 mm, which are good enough for rail profile measurement.