The Three-Dimentional Model Of Extraction Method Of Buildings Combined LiDAR Data And Remote Sensing Images

Airborne LiDAR is a kind of sensor devices,which is the integration of all kinds of new technology,such as laser ranging technology,INS technology and GPS technology etc. Compared with the traditional measurement mode, the Airborne LiDAR has some virtues,such as a shorter production cycle,carrying on the surveying and mapping to the digital city quickly,etc.The building was an important content of city3D Model of Building is an important content of the construction of "digital city".Buildings’shape of city are complex,so studying how to extract3d model of building rapidly and efficiently from LiDAR data rapidly and efficiently has an important significance and function for urban planning, and geographic information service.The paper analyses and summarizes the research status of using LiDAR data extract the point cloud data of buildings both China and foreign.Extracting the building’s elevation and bottom contour shape accurately is the basis of3d model of building extracted from LiDAR data. The key is to build DEM with high precision.During the process of converting the DSM to the DEM, the effect of topographic relief degree change need to be considered fully. According to different terrain types, different transformation methods should be used.The buildings floor outline extracted only by LiDAR data is not accurate enough. It could be solved effectively with high resolution remote sensing image data.The first process of article was converting the LiDAR point cloud data to DSM by interpolation.During the area of city with small relief degree, grid numerical substitution method of small area was used to convert the DSM to the DEM. During the area of mountain with big relief degree,through setting the scope of "feature area" and " ground area" to obtain "the ground seed region".Comparing with "the ground seed region" and neighbor area’s elevation, then expand it to get " the growth area ground".The remaining space’s elevation could be determined by interpolated with spline function method, superposed the elevation of "the ground seed region" and "the growth area ground" to get the DEM of the area of mountain.Lastly, subtracting the DSM and DEM to get the regularization grid data(nDSM).Through filtering process of high threshold, area threshold and slope variable ratio threshold to extract building grid data with accurate elevation.With the help of a professional remote sensing image processing soft ware, a two-dimensional vector data of building outline shape was extracted using high resolution remote sensing image. Firstly, the image was segmented,secondly,the feature information of segmented region was extracted(including spectrum、geometry and texture characteristics). After picking up samples according to the feature inform ation, GLC tree classification was carried out to extracting building botto m contour shape. Finally, a three-dimensional building model was extract ed by overlay analysis of the two-dimensional vector data and raster data of building.The main technical characteristics of the paper,integrates LiDAR data and high-resolution image data to extract city building3D models, preliminary investigations and comparison with the actual buildings show that the three-dimensional model of the building extracted with the method discussed in this article is stereoscopic,with a exact bottom surface profile and accurate height value. So that the method can be applied to city building3D models of the fast modeling.