Voxel Based Building Extraction For Airborne LiDAR Data

As an important component of the city,building is of great significance to the establishment of 3D city modeling and digital city.Airborne LiDAR technology provides a new technical method for the urban spatial information,building information in particular.However,the point cloud data acquired by airborne LiDAR technology has the characteristics of mass,discrete and uneven distribution,which makes the building extraction algorithms existing the problems of low efficiency,complicated construction and loss of information in the data structure.Aiming at the above problems,this paper proposes an algorithm for building extraction based on voxel.Making the original discrete point cloud data as the research object,through the method of vector and the calculation method of residual estimates the minimum value of the selected element in building seed,seed voxel based region growing method,the vector angle and distance threshold as the criterion,to realize the accurate extraction of buildings;designing a voxel based connected component analysis(Voxel based Connected-Component Analysis,VCCA)algorithm for building segmentation;using Alpha-Shape algorithm for building outline extraction.In order to verify the accuracy of this method,eight kinds of scene points cloud filtering test data provided by ISPRS in the extraction of buildings,building segmentation,outline extraction and other experiments,and the quantitative and qualitative results verify the extraction of buildings.The experimental results show that the proposed algorithm is not limited by the terrain and the type and size of the buildings,and the Kappa coefficient can reach more than 80%,which verifies the feasibility and effectiveness of the algorithm.