A Study of Objectivization Method for Automatic Production of 3D LOD2 Building Model

摘要

With the fast urbanization of the world, the concept of "Smart City" has been presented. Fields of photogrammetry. remote sensing, and computer vision have developed various geoinformatics technologies for the reconstruction of 3D city model to support the realization of smart city. However, the production of a large quantity of 3D building models in 3D city model may face many problems, e.g. the efficiency, the cost. This paper discusses an automatic building objectivization method to generate 3D building models of large area by combining airborne LiDAR point cloud data. DEM. and airborne vertical and oblique images. The approach is developed based upon a conversion of each building outline shapefile (.shp) into object file (.obj) and a subsequent appropriate texture mapping. After that, building texture is baked into single image according to its corresponding material template library file (.mtl). The procedure can be divided into the following four stages. First, the elevation and ground floor elevation corresponding to each building in shapefile are determined based on LiDAR point cloud data within each building outline and a DEM. Second, individual 3D LOD1 building model with the standard file format of object file presented by triangulation networks can be transformed from the shapefile to object file, even 3D LOD1 building model with holes can be produced. Third, from aerial vertical and oblique images as well as the whole 3D LODI building models, the texture corresponding to each triangulation of 3D LODI building model is determined by the most approximate image for texture mapping. Finally, each complete 3D LOD2 building model is constructed by baking its corresponding texture images into the only one image. The relationship between object file and building textures is described by material template library file. In this study, the objectivization method is presented for automatically creating a large quantity of 3D LOD2 building models, and it is hoped to provide a new method to efficiently produce 3D LOD2 building models for use in the future.