by Poullis, Charalambos, Gardner, Andrew and Debevec, Paul
Abstract:
For realistic simulations, architecture is one of the most important elements to model and render photorealistically. Current techniques of converting architectural plans or survey data to CAD models are labor intensive, and methods for rendering such models are generally not photorealistic. In this work, we present a new approach for modeling and rendering existing architectural scenes from a sparse set of still photographs. For modeling, we use photogrammetric modeling techniques to recover a the geometric representation of the architecture. The photogrammetric modeling approach presented in this paper is effective, robust and powerful because it fully exploits structural symmetries and constraints which are characteristic of architectural scenes. For rendering, we use view-dependent texture mapping, a method for compositing multiple images of a scene to create renderings from novel views. Lastly, we present a software package, named Fac¸ade, which uses the techniques described to recover the geometry and appearance of architectural scenes directly from a sparse set of photographs.
Reference:
Photogrammetric Modeling and Image-based Rendering for Rapid Virtual Environement Creation (Poullis, Charalambos, Gardner, Andrew and Debevec, Paul), In Proceedings of the 24th Army Science Conference, 2004.
Bibtex Entry:
@inproceedings{poullis_photogrammetric_2004,
address = {Orlando, FL},
title = {Photogrammetric {Modeling} and {Image}-based {Rendering} for {Rapid} {Virtual} {Environement} {Creation}},
url = {http://ict.usc.edu/pubs/PHOTOGRAMMETRIC%20MODELING%20AND%20IMAGE-BASED%20RENDERING%20FOR%20RAPID%20VIRTUAL%20ENVIRONMENT%20CREATION.pdf},
abstract = {For realistic simulations, architecture is one of the most important elements to model and render photorealistically. Current techniques of converting architectural plans or survey data to CAD models are labor intensive, and methods for rendering such models are generally not photorealistic. In this work, we present a new approach for modeling and rendering existing architectural scenes from a sparse set of still photographs. For modeling, we use photogrammetric modeling techniques to recover a the geometric representation of the architecture. The photogrammetric modeling approach presented in this paper is effective, robust and powerful because it fully exploits structural symmetries and constraints which are characteristic of architectural scenes. For rendering, we use view-dependent texture mapping, a method for compositing multiple images of a scene to create renderings from novel views. Lastly, we present a software package, named Fac¸ade, which uses the techniques described to recover the geometry and appearance of architectural scenes directly from a sparse set of photographs.},
booktitle = {Proceedings of the 24th {Army} {Science} {Conference}},
author = {Poullis, Charalambos and Gardner, Andrew and Debevec, Paul},
month = dec,
year = {2004},
keywords = {Graphics}
}