The Method Of Vehicle-Borne Laser Point Cloud Data Streaming Thinning

Vehicle-borne laser mapping system (VLMS) obtained accurate three-dimensional information by multiple sensors and data processing technology. However, the vehicle-borne LiDAR data were usually too large for subsequent data processing and storage, the compression of the vehicle-borne LiDAR data became necessary. The main contents of this thesis include:1. This thesis aimed to analyze the composition and principle of vehicle-borne laser mapping system. Vehicle-borne laser mapping system obtained geo-spatial three-dimensional (3D) coordinate information via navigation and positioning, integrated Geographic Positioning System (GPS), PS, INS, and the DR. The goal was to improve the accuracy of geo-spatial three-dimensional coordinates information.2. The features of vehicle-borne LiDAR data were different from the features of air-borne LiDAR data. On account of diversity of urban surface features, the features were divided mainly into three categories:Man-made, natural and moving or non-moving. The spatial distribution of LiDAR data was irregular and scattered. According to the scanning approach of VLMS, the vehicle-borne LiDAR data was arranged in a scan line by scan line alignment and had inherent topological coherence. If the objects were the nearest surface to the scanner with respect to other objects, the reflected laser point density was higher compared to other objects. Consequently, the vehicle-borne LiDAR data resulted in massive large volumes.3. The resulting vehicle-borne LiDAR data had a strong inherent spatial coherence. The method of streaming compute was highly effective in improving the quality of data processing, even with a large volume of data. When thinning LiDAR data by streaming computation, the first data region was divided into quad tree grids. The quad tree grids acted as a spatial index and were used in massive cloud data organization and management. As the cell and its data became stable, they formed a mesh stream which could be directly piped into following procedure as input data or stored to disk as a file. Then, based on the mesh stream, a triangulated irregular network (TIN) was built by inserting data point by point. Final triangles were identified while a TIN was built for further thinning process of the massive points.4. The thinning method based on the streaming computation was also proposed in this paper. The steaming compute of LiDAR data was integrated with algorithms to study the grid-based compression on TIN, slope, and on the improved slope.The acquisition of data by the VSML took place at Southwest Jiaotong University at Xipu. The study indicated that the thinning based on the streaming computation method could reserve the point of building profile as well as reduce the loss of DEM accuracy caused by compression. As a result, the streaming computation method was more efficient to process data than other algorithms.The study was valuable for digital cities and digital Earth.