Procedural Placement of Existing Building Models in Virtual Cities

Abstract

Procedural urban scene generation can show a lot of insights about infrastructure, organization and development of a city, and find itself many applications in different fields. On the other hand, in many cases the available information from satellite images and terrestrial dataset is limited, leaving considerable blank spaces in the generated virtual city. How to fill the blank spaces in order to obtain a more realistic scene is thus an important problem to be solved by designers. Designers usually spent enormous time and effort to create three-dimensional (3D) urban models using 3D modeling software. Importing building models manually one by one to cover the blank spaces in a large urban setting can be labor-intensive and also means high cost of time. Enlightened by these observations, this study is aimed at the design and implementation of an algorithm focusing on 1) the identification and positioning of blank spaces in a virtual urban area and 2) the automatic filling of such blank spaces, importing 3D models of existing buildings from a database.