Airborne LIDAR Point Cloud 3D Filtering Method Based On Binary Voxel Data Structure Model

Airborne light detection and ranging(LIDAR)is a kind of active earth observation system(EOS),provides a new type of data for digital elevation model(DEM)acquisition,which is dense and high-precision 3D point cloud data.DEM can be obtained from LIDAR point cloud data only by filtering.And the accurate filtering result is the premise of producing high precision DEM.In the existing classical filtering algorithms,the common methods to express the point cloud data include raster grid,Triangulated Irregular Network(TIN)and point cloud.The first two are the 2.5D data representation,the same plane coordinates can only correspond to an elevation value,when the 3D point cloud data is expressed by those method,may cause the information loss,and then affects the accuracy of the subsequent filtering algorithm based on those data representations;and based on point cloud data,the spatial structure information is difficult to use,lead to filtering algorithms design difficult based on point data.Therefore,it is necessary to construct a more simple real 3D data structure.On the basis of the structure developing the 3D filtering to make full use of the spatial neighborhood information of the point data and improve the filtering precision.The main contents and innovation of this paper can be summarized as follows:At first a binary voxel-based data(BVD)model is proposed.The model's construction is to divide the entire scene volume into a 3D regular grid(the 3D sub-volumes,called voxels),remapping the original data points to 3D voxels and assigning the voxel value 1 or 0 when a voxel contains LIDAR points or not.Compared with raster grid,TIN,point cloud and other data structures,the advantage of voxel representation lies in that the implicit notion of adjacency and the convenient information extraction.Based on this model,a 3D filtering algorithm based on voxel is proposed.The proposed filtering algorithm selects the lowest voxels with value 1 as ground seeds and labels their 3D connected voxel set as ground voxels.The proposed algorithm makes better use of the implicit adjacency relation in BVD structure model.ISPRS benchmark dataset are used to compare the performance of V3F with those of eight other publicized filtering methods.Results indicate that the V3GF improves on Axelsson's performance on five samples in terms of total error.The average Kappa coefficients for sites with relatively flat urban areas,rough slope and discontinuous surfaces are 92.49%,72.23%and 61.27%,respectively.