Reconstructions of large cultural heritage sites, at multiple time periods, facilitate public awareness, enable the visualisation of regeneration proposals, and assist archaeologists in establishing the validity of a particular hypothesis within context. This paper focuses on the reconstruction of sites that no longer exist, where archived cartography and archaeologist's sketches provide an invaluable resource conveying the layout of an area. Whilst three-dimensional models are used in a broad range of applications their construction typically involves a labour intensive process and this paper presents a set of techniques to aid the reconstruction of environments from maps. In particular, the approach considers that an environment will exhibit a substantial amount of similarity, which is exploited to reduce the modelling time. The concept of similarity permits a dominant set of k building footprints to be identified from a map. A set of models representing the k dominant footprints are created and, based upon both the image based and geometry based metrics discussed in this paper, are aligned to the closest matching footprint in the archived map. Any building that is not sufficiently close to any of the k dominant footprints is labelled as being visually distinct and requires further modelling. To evaluate the technique a reconstruction of 19th Century Koblenz is undertaken, where 2300 building footprints are extracted, classified and aligned to one of 51 dominant building footprints in under fifteen minutes.
[1]
David B. Arnold,et al.
Rapid procedural-modelling of architectural structures
,
2001,
VAST '01.
[2]
Luc Van Gool,et al.
Procedural modeling of buildings
,
2006,
SIGGRAPH 2006.
[3]
Andrew Jones,et al.
Digital Reunification of the Parthenon and its Sculptures
,
2003,
VAST.
[4]
A. M. Day,et al.
Exploring cultural heritage sites through space and time
,
2008,
JOCCH.
[5]
Pascal Müller.
Procedural modeling of cities
,
2006,
SIGGRAPH Courses.
[6]
Ramakant Nevatia,et al.
Automatic description of complex buildings from multiple images
,
2004,
Comput. Vis. Image Underst..
[7]
Tony Craig,et al.
Design and use of a virtual heritage model to enable a comparison of active navigation of buildings and spaces with passive observation.
,
2007
.
[8]
Gabriele Guidi,et al.
Virtualizing ancient Rome: 3D acquisition and modeling of a large plaster-of-Paris model of imperial Rome
,
2005
.
[9]
Ming-Kuei Hu,et al.
Visual pattern recognition by moment invariants
,
1962,
IRE Trans. Inf. Theory.