Construction of Multi-resolution Spatial Data Organization for Ultralarge-scale 3D Laser Point Cloud

The high-precision laser point cloud data obtained by 3D laser scanning technology is an important source of 3D spatial data for smart city construction. The efficient data organization of TB/PB ultralarge-scale point cloud data for urban applications is the key to point cloud data processing and visualization. Toward solving the problems of data redundancy and low storage efficiency in existing spatial data organization, we propose a spatial data organization model of a multi-resolution ultralarge-scale point cloud based on an octree and its multi-resolution point cloud construction method based on the divide-and-conquer algorithm. Firstly, a multi-resolution spatial data organization model based on an octree without redundancy is designed, which makes it easy to quickly judge the visibility of the point cloud. To obtain a high-quality rendering effect, a double-shell Poisson disk sampling method is used as a point cloud filling method to ensure constant spacing between sampling points and improve the visualization quality of point clouds. Finally, the original point cloud is partitioned by a quadtree and a process is started for each quadtree node. We propose a parallel construction algorithm for a multi-resolution point cloud to improve the construction efficiency of the algorithm when dealing with massive point clouds. In this paper, a large number of urban street point cloud data are obtained and tested using a high-precision vehicle-mounted 3D laser sensor. Experiments show that the multi-resolution spatial data organization model based on an ultralarge-scale 3D laser point cloud is reasonable and that its algorithm is efficient.

Automated summary

This paper proposes a spatial data organization model of a multi-resolution ultralarge-scale point cloud based on an octree and its multi-resolution point cloud construction method based on the divide-and-conquer algorithm, which solves the problems of data redundancy and low storage efficiency in existing spatial data organization. The experimental results verify the rationality and efficiency of the proposed model and algorithm.