A Camera Calibration Method with Large Oblique Angle Machine Vision

In order to ensure the safety of the subway when it stops at the platform, it is necessary to detect the limits of the subway platform. However, the currently used station limit detection tools have large detection errors and low efficiency, and cannot meet the detection requirements. In this paper, a non-contact platform limit detection system based on machine vision is studied, and a new set of comprehensive calibration methods is proposed for the calibration of the camera in the system at large angles. After collecting the images of large multi array points set by camera, the effective calibration area is selected and the reference circle center is extracted from the connected domain and centroid center algorithm. After collecting the images of large multi array points set by camera, the effective calibration area is selected and the reference circle center is extracted from the connected domain and centroid center algorithm. The base circle center is used as the reference point to carry out perspective transformation, and the change matrix is obtained and tested. Finally, the maximum error of pixel calibration is 1.639pixel. It is proved by laboratory and field experiments that the measurement system of this method can obtain the contour geometry of the platform section accurately, and the gauge accuracy is within ±1mm after repeated measurements, and the accuracy of platform height and platform distance can be within ±3mm.