Index model based on top-down greedy splitting R-tree and three-dimensional quadtree for massive point cloud management

Abstract

With the rapid development of three-dimensional (3-D) laser scanners, the correspondent point cloud accuracy and density are continuously improving. However, the point data volume becomes larger and larger, which brings new challenges for the point cloud data real-time processing on personal computers. To meet the managing requirement for real-time point cloud processing, we proposed a hybrid index model characterized by top-down greedy splitting (TGS) R-tree and 3-D quadtree, aiming at the balance improvement and the high index query efficiency. First, the large-scale point cloud data are divided into grids based on their spatial distribution, and then, the proposed TGS R-tree algorithm is applied to organize these grids. Second, a 3-D quadtree local index model is developed to manage local points in each grid. Experiments of five point cloud data scenarios are conducted to evaluate the proposed method, and the results show that the proposed model can meet the efficiency need of various 3-D point cloud managements, especially for those mainly distributed in XY plane, including the airborne LiDAR point cloud.