激光摄像式传感器标定是采用激光摄像技术进行轨道几何参数检测的难点。建立基于非线性最小二乘法及高斯-牛顿最优化计算方法的激光摄像式传感器标定模型。在专用调整平台上采用棋盘格进行标定试验,获取200组标定数据。分别运用非线性最小二乘法及高斯-牛顿最优化计算方法,求解激光摄像式传感器标定模型参数。在上述200组标定数据基础上,另取200组数据共同参与标定误差分析。结果表明,高斯-牛顿最优化计算方法较非线性最小二乘计算方法标定精度在水平方向和垂直方向分别提高0.155 mm、0.150 mm。针对激光摄像式传感器在轨道检测中的应用,提出采用高斯-牛顿最优化计算方法进行标定。给出轨道检测中基于高斯-牛顿最优化计算方法的轨距及转向架倾摆角求解方法。
Utilizing laser and camera for track geometry inspection,calibration approach for laser photogrammetric transducers is a key issue.The calibration models constructed by the nonlinear least-square method and Gaussian-Newton optimization computational method are presented respectively.The transducer measurement models parameters are acquired by the above-mentioned approaches making use of 200 individual data in the checkerboard plane by the dedicated test platform.Moreover,the other 200 individual data are added for different algorithms calibration errors analysis.The experimental results show that Gaussian-Newton algorithm is more accurate than the nonlinear least-square algorithm in camera calibration.The horizontal and vertical accuracy can improve 1.55 mm and 1.50 mm by use of Gaussian-Newton algorithm.While using laser photogrammetric transducers in track inspection vehicle,Gaussian-Newton calibration algorithm is proposed.Track gauge and bogie tilting angle computational formulations are given by Gaussian-Newton algorithm in the track inspection system.