Adaptive inverse perspective mapping transformation method for ballasted railway based on differential edge detection and improved perspective mapping model

Abstract

Inverse perspective mapping (IPM) can eliminate the perspective effect of the image, which is mainly used for rail-lane detection and foreign body intrusion detection in rail transportation. Most of the IPM algorithms set the IPM parameters in advance according to the intrinsic and extrinsic parameters of the camera system, which has the problem of poor generality. In order to solve the problem of existing algorithms, an adaptive IPM method suitable for ballasted rail is proposed. This method has two parts. The first part is to obtain a trapezoidal calibration region surrounded by rails and sleepers; The second part is through the improved perspective mapping model, the trapezoidal calibration region is solved to perform the adaptive IPM. Finally, we conducted two experiments. In calibration experiments, the error range of adaptive IPM is between 0.1% and 5.3%. In outdoor field experiments, the error of the camera's horizontal field of view, vertical field of view, and installation height obtained by the adaptive IPM is about 10%. Experiments show that adaptive IPM is better than other methods when only the camera pitch angle parameter is needed, and it has generality on different cameras and adaptability in different scenarios. This research has a certain reference value for future railway rail inspection and foreign body intrusion detection.