Research Of Building Extraction Based On Air-Borne LIDAR Cloud Data

Buildings as the most important entities in large numbers in urban areas are the important basis of urban planning, energy needs assessment, and urban population statistics. Combining laser measurement technology, inertial navigation technology, differential positioning technology and computer technology, airborne LIDAR technology could obtain accurate three-dimensional information directly and quickly. Compared with traditional photogrammetry methods, airborne LIDAR technology has a unique advantage, which plays a more and more important role in surveying and mapping of city. In summary, the study on how to make use of airborne LIDAR technology for the reconstruction of buildings has important practical significance.In reality, the shape of common buildings is regular, most of which is rectangular or a complex of rectangular. Based on the analysis of the extractive methods of buildings, a framework of automatic extraction and reconstruction for common buildings is designed and the correctness and effectiveness of the framework is verified.The major works implemented in this dissertation are as follow:1. The components positioning principle of airborne LIDAR system, and the characteristics of airborne LIDAR point cloud data are systematically introduced. Then the dissertation summarizes the data processing of Airborne LIDAR cloud data, which provides a theoretical support to further study.2. Through the analysis of the current algorithm of main filtering algorithms, progressive morphological filter is chosed to achieve the filtering processing. During the filtering process, the window size increases gradually and all nonground objects of various sizes detected. It proves in experiments that the algorithm can eliminate building measurements to obtain DEM data.3. Based on regular grid data structure, seed point is selected buy calculating the minimum deviation through plane-fitting. And then carry out the region-growing operation. At last, building measurements are detected.4. A building footprint automatic extraction framework is presented based on airborne LIDAR point cloud data. Firstly, Raw footprints are derived by connecting the boundary points for building patches. Then, the Douglas-Peucker algorithm is employed to generalize line segments in the footprint. At last, the simplified outline is adjusted using the dominant directions estimated. This extraction framework works well for buildings with two dominant directions in experiment.5. A quick reconstruction method of the building whose roof is a gable is put forward. And through plane property analysis, parallel and perpendicular properties of a plane are enforced using the variation of elevation values. Then plane-fitting is replaced by line-fitting, and the reconstruction is simplified. Through experiments, the method presented in this dissertation proved effectual.