On-line Calibration Method For The LASER Scanning 3D Imaging System On Railway Inspection

It is a great responsibility to ensure the railway system trouble-free and reasonable operation in Chinese rail transit field.The maintenance and repair of railway infrastructure such as rails,switches and sleepers is an important work.At present,special rail inspection vehicles are used in China,and laser scanning imaging and IMU sensor are used as the core rail inspection technology to complete the detection tasks of rail geometry and position defects.The static orbit profile measurement accuracy of the system is mainly determined by the line laser imaging process,so the internal parameter calibration of the optical plane of the sensor has a great influence.In order to eliminate vibration interference during moving detection,IMU is needed to obtain vehicle attitude information,which leads to the calibration problem of visual components and IMU external parameters.Based on the development of the latest generation rail inspection system of CRRC Zhuzhou,this paper solves the following problems:In order to improve the calibration accuracy of internal parameters of laser scanning sensor in rail inspection system,polynomial fitting optical plane calibration model is proposed in this paper.The model does not consider the internal and external parameters of the camera and the structural light system parameters.The polynomial fitting optimization algorithm can obtain the optical plane parameters,which can achieve higher calibration accuracy and stability than the traditional method.In the line scanning vision platform measurement experiment,the calibration model in this paper is 1/2 error of the traditional method under the same hardware.The calibration model achieves the relative size measurement accuracy of 2‰～6‰,which is of great significance to improve the imaging accuracy of the component.Aiming at the calibration problem of line scanning camera and IMU external parameters in the operation environment of rail inspection vehicle,an online calibration method of 3D components is proposed in this paper.In this scheme,the auxiliary camera on the component is used for calibration,and the principle of monocular imaging cross-ratio invariance is used.The image algorithms such as Hough gradient circle detection and light center thinning are combined to achieve calibration.The feature points can be located in IMU coordinate system with high-precision three dimensional calibrator.In line laser imaging profile,the first-order differential derivation and RANSAC linear fitting optimization algorithm are used to improve the positioning accuracy of feature points in the imaging coordinate system.Finally,the same feature points are expressed in different reference systems,and the external parameter calibration matrix is obtained by SVD algorithm.In the field test of orbital inspection,the reprojection error of the external parameter calibration of the proposed scheme is 1/3 of the target ball method,and the accuracy is up to 0.1mm,which has the advantages of simple operation,high precision and on-line calibration.