Research On Airborne LiDAR Point Cloud Thinning For Highway Survey And Design

In the field of highway construction,from the initial mapping of roads to the reconstruction and expansion of highways,airborne LiDAR technology can provide high-precision terrain data,however,due to the narrowness of the highway survey area and the high density of LiDAR point cloud,the amount of data acquired by the airborne LiDAR is huge,the huge amount of data will affect the speed of DEM construction,ease of data exchange,and data storage speed in highway survey and design,therefore,it is useful for practical engineering application to thinning and simplify airborne LiDAR point cloud used for highway survey and design.With respect to the method of thinning LiDAR point clouds,how to ensure the rationality of the distribution of points while preserving the terrain feature points,avoid the occurrence of largearea point cloud holes,and improve the processing speed of the algorithm is still the focus and difficulty of current research.In this paper,the main research contents are as follows:(1)The advantages and disadvantages of the three kinds of massive point cloud data spatial indexing methods of regular grid,octree,and KD tree are summarized and analyzed.Combining the characteristics of airborne LiDAR point cloud data,the KD tree is used as the spatial index structure of the experimental data in this paper to improve the processing speed of the algorithm.(2)According to the working principle of airborne LiDAR,the main sources of gross errors are analyzed and summarized.The method based on hypothesis testing is used to eliminate gross errors.(3)Point cloud thinning algorithm based on mean curvature is proposed,and the marker method is used to solve the problem of point cloud voids,to improve the rationality of point distribution,and to ensure the accuracy of the point cloud after sparse dilution.(4)Designed and developed software for point cloud display,point cloud gross error elimination,point cloud evacuation,and other functions.Based on actual data,point cloud gross error elimination and lean extraction experiments were performed to compare the experimental results.