Automatic Generation of 3D Building Models Considering Roof Shape

Abstract

It costs a lot of money and time to create 3DCG models of a cluster of buildings in a residential area in order to replicate landscape of a city. One approach is to measure the real space with a laser scanner and reproduce it a 3DCG model, but this requires a costly machine and a high skill to operate the machine. Accordingly, aiming at the automatic generation of 3D buildings considering roof shapes using easily available map data and aerial photographs, this study developed a “RSD Tool” system which discerns roof shapes in aerial photographs and a “RMAC Tool” which automatically generates 3DCG models of buildings with the data of roof shapes and Fundamental Geospatial Data. Proving test showed that the RSD Tool can discern 60.8% of roof shapes precisely, and the RMAC Tool can replicate 51.0% of roofs automatically.