Research On Strip Adjustment Method Of Unmanned Airborne Vehicles LiDAR Point Clouds

The Unmanned Airborne Vehicle LiDAR(UAV-LiDAR)system integrates a variety of sensors such as 3D laser scanner and navigation positioning and orientation system(POS).It can safely and efficiently obtain ground point cloud data.However,the point clouds obtained from the calibrated UAV-LiDAR system still has various errors,which will cause the relative error of the point cloud between strips that can manifest as differences in the coordinates of correspondences from different strips.If you want to cut the relative error in point clouds,you need to analyze the main sources of error and establish an effective mathematical model of adjustment.In order to improve the relative accuracy of UAV-LiDAR point cloud point clouds,this paper studies and discusses problems of point cloud registration,random error source of UAV-LiDAR point cloud,establishment and solution of fitting model for position and attitude errors in POS trajectory.The main work and conclusions of the paper are as follows:(1)In order to ensure the feasibility and effect of strip adjustment,it is necessary to select a fine registration method with higher efficiency and better accuracy.Therefore,the 3D Normal Distribution Transformation(3D-NDT)registration method and the Point-to-Plane Iterative Closest Point(P-P ICP)registration method are compared.In order to reduce the number of point clouds and improve the efficiency of registration,a point cloud preprocessing method is designed.Three sets of point cloud data are used for experiments,and the root mean squared errors(RMSE)of correspondences' distance is used to evaluate the overall accuracy of the registration.The results show that the efficiency and accuracy of point-plane ICP registration is higher than that of 3D-NDT.The results show that point-plane ICP is a more suitable registration method for UAV-LiDAR point cloud.(2)In order to specifically reduce the main random errors of the point cloud,the sources of the random errors of the UAV-LiDAR point cloud are analyzed from the point cloud acquisition process and the geometric positioning model,and each random error in the point clouds is obtains through calculation.The results show that the position and attitude errors in POS trajectory are the main source of random errors in point clouds.(3)In order to reduce the position and attitude errors in POS trajectory,a strip adjustment model is established.The observation equations of POS trajectory position and attitude are constructed,and the cubic spline function with time as independent variable is established.Observation equations and cubic spline functions are used to construct the strip adjustment model.Finally,with the minimum sum of squared distances of correspondences as the goal,the unknown parameters of the model are calculated,and the accuracy is evaluated.Experiments are conducted using the correspondences of three sets of data,and the overall accuracy of the adjustment was evaluated by the distance RMSE of the correspondences.The experimental results show that after strip adjustment,the RMSE increases to 9.24 cm,9.84 cm,and 13.14 cm.The results can quantitatively prove that this strip adjustment method can improve the relative accuracy of point clouds between two strips.