Reconstruction Of Buildings With Multi Right Angles And Multilayers From LIDAR

3D-reconstructing is an important part of survey. In fact, traditional ways of acquiring the space data have lots of inherited shortcomings, so they can not adapt to the demand of 3D-reconstrucingt increasely. Finding a new method of acquiring space data is necessary. The data of LIDAR(Light Detect And Ranging),which has inborn 3D property, is making a more and more important pole in 3D-reconstruction.In this paper, make a focus on reconstructing multi-orthogonal and hierarchical buildings from the DSM derived from LIDAR.During the progress, hierarchy and structure of buildings are taken into account and some contribution are made on reconstructing the multi-orthogonal and hierarchical buildings. Because the roofs of buildings is very complex, they can not be described by a general model. nd the flat roof and gable roof are the most popular types ,which are the models adopted in this paper. In order to improve the efficiency and robustness of reconstructing, how to find proper way to deal with local data(finding of ROI) is important.The contents and contribution of this paper just as followed:1) Regarding the contour as the bridge between DSM and the model of buildingsIn this paper, regard the contour as the bridge between DSM and the model of buildings, The contour,which describe the figure of buildings in every hierarchy of buildings and afford the elements to reconstruct, indicates the altitude of buildings in every level ,so it is possible to reconstruct the multi-hierarchical buildings. For a complete buildings, contour is enclosed,which avoid grouping the scatted edges and get an enclosed area of interest, so the problem of localization is resolved and improve the efficiency and robustness.2) Use the mode of LS as the tool for normalizing the shape of buildings' plane.In reality, the buildings are deliberately designed and built,so the adjacent angle of neighbour edges are orthogonal in many cases. In this paper, By means of the model of LS adjusting and additional conditions such verticality , parallelity and dependent point so on, adjust the figures of building at every level ,and exactly locate the corners of building,then accomplish reconstruction of orthogonal buildings. 3)Chose the building's modle through analyzing the normal' direction in ROIThrough the normal's direction, the model of building is automatically chosed . If it is gable, extract the two slopes according to the peak of directions; extract the ridge of house by intersecting slopes; acquire the height and vertexes of house by intersecting slope and vertical wall.4) Adopt TIN to reconstruct the roofBecause TIN can perfectly describe patches, feature points(such as vertex, and some points of ridge) are extracted and used to construct TIN , then the roof is described by the patches.5) Make use of some simple measures for filter "false " buildingsThe clues in LIDAR are mainly geometric information .Some "false" buildings(such as trees and cars) may be regarded as buildings because they are above the ground. So some measures such as: size ,shape ,area ,ratio of area and circumference,elevation are adopted to filter the "false" buildings.The experiment shows the method of automatically extracting multi-orthogonal and hierarchical houses from the DSM,which is depicted in this paper, is effective.