Organization And Visualization Of Massive Multi-source Point Cloud Data

LiDAR(Light detection and ranging) technology, including airborne and terrestrial laser scanning, is a new remote sensing technology for fast acquisition of accurate three dimensional spatial information of Earth’s surface and other targets. The three-dimensional point cloud is an important data source of digital surface models, digital elevation model, three-dimensional building model reconstruction and forests estimation. Compared to traditional measuring technology, LiDAR technology has advantages of high efficacy and low cost. Thus, LIDAR is currently used in the digital city, forestry management, transportation line and power line detection, cultural heritage documentation and other area.Point cloud management and visualization is a basic research question, which influence the following data processing and application significantly. In spite of many previous relate work, this problem has always been a challenge, especially for the terrestrial laser scanning data and combination with airborne data. In addition, the approach to manage other surveying point cloud data from construction projects is also an issue.This paper focuses on the management of airborne and terrestrial laser scanning point cloud data at the beginning, then extend to other types of measuring data. Plus, the paper also pays a lot of attentions on the combination methods of multi-source data. Finally, the implementation system of virtual globe platform and the application in a hydraulic engineering project are briefly disused. The main contributions of this thesis are as follows:1. Based on the analysis of multi-source point cloud data, we have proposed a pyramid data organization model of airborne LiDAR point cloud data. This pyramid model is a multi-resolution data structure which combines the global quadtree index with local KD-tree. We use quadtree index to organize the overall data, which construct the upper layers of the multi-resolution structure. However, we can’t guarantee all the leaf nodes contain limited points and thus KD-tree is used to index the local region where the spatial distribution is uneven. This is the basis for large-scale visualization of point cloud data.2. We have also proposed the multi-resolution data structure for terrestrial laser scanning point clouds and other surveying point cloud. The terrestrial laser scanning data is indexed by R-tree and plane features are also considered in the process of data partition, which is the most significant feature in TLS data sets. For other surveying point data, we build multi-resolution triangulation data structure to construct three dimensional models. The methods are useful for efficient rendering.3. The above data structures are used to implement a LiDAR data management and display module in the virtual earth system-GeoGlobe, which is developed as a spatial information sharing platform by Wuhan University. The integration methods of airborne and terrestrial point cloud on virtual globe are discussed. Two different rendering approaches based on LoD are also described.4. We examine the methods on virtual globe platform by airborne and terrestrial point cloud data in the City of Dunhuang(Gansu Province) and power line data in the City of Yichang(Hubei Province). The methods are also be used in a hydraulic engineering project.