Extraction Of Building Roof Contours From Lidar Data

Buildings occupy a large amount of urban space and are one of the main components of city. The three-dimensional building information not only plays an important role in digital city construction, but also in further analysis and interactive operation in urban modeling. The technology of airborne Light Detection and Ranging (LiDAR) provides tools to rapidly obtain high-density and high-accuracy data, which is a new data source for digital city construction. However, how to rapidly extract building information from the massive LiDAR data is still a challenge. Building roof information is very important in building modeling, and this paper thus proposes several methods for extraction of building roof contours from LiDAR data. The main work is listed as below:(1) After summarization of the development, components, surveying principles, data organization and characteristics and main applications of LiDAR, a brief introduction of the basic concepts, principles and methods of LiDAR data filter is given;(2) Present a literature review on approaches to building extraction using LiDAR data, including building cloud, region growing, Hough transform, twice echo algorithm, TerraScan and multi-source data;(3) Realize and test the feasibility and reliability of Alpha Shapes algorithm, tube-based algorithm and boundary-regularized algorithm to extract building roof contours, and by investigation of data clustering approaches, the DBSCAN algorithm is adopted to realize LiDAR point cloud segmentation of buildings, which help improve the efficiency of subsequent building roof contours extraction;(4) Develop a software to extract building roof contours based on ArcGIS Engine9.3and Microsoft Visual Studio2008. The software have such functional modules as add data, construct Triangular Irregular Network (TIN), segment data, data clustering, extract contour lines and regularize contour lines etc. The software has then been tested by several data.Results of the above work indicate that, the Alpha Shapes algorithm is relatively effective in extraction roof contours of the discretely distributed buildings; the operation of clustering point sets before extraction building roof contours can not only help effectively avoid the trouble of setting Alpha value, but also help improve the efficiency of Alpha Shapes algorithm.