Towards the Automatic Integration of Ground Level Images into a Virtual Urban Environment

Robert Laycock and Dr Andy Day
School of Computing Sciences
University of East Anglia
Norwich NR4 7TJ
United Kingdom

e-mail: Robert.Laycock@uea.ac.uk, amd@cmp.uea.ac.uk

http://www.cmp.uea.ac.uk/research/rwm/rwm1.html

Abstract

In this paper we present a method which progresses towards automatically generating the texture maps for the building façades of an existing coarse three dimensional urban model. Rather than focusing on a method for achieving the exact texture map the algorithm quickly creates two texture maps. These represent the building façade’s wall material and building features, such as windows and doors. The paper describes two algorithms. The first defines an area in the image to be used as a repeatable wall pattern. It undertakes this procedure without prior knowledge of the wall material. The second algorithm presented creates an object map from the building image. The object map is an image containing the spatial arrangement of building features on a transparent background. These features are extracted from the building image automatically using a variant on an active contour model. The approach focuses on residential scenes where occlusions cause problems with existing automated techniques.