3D Modeling From Images Using Geometric Constraints

Marta Wilczkowiak 1
1 MOVI - Modeling, localization, recognition and interpretation in computer vision
GRAVIR - IMAG - Graphisme, Vision et Robotique, Inria Grenoble - Rhône-Alpes, CNRS - Centre National de la Recherche Scientifique : FR71
Abstract : Image-based 3D modeling is often an underconstrained and noise-sensitive process. Incorporation of geometrical constraints increases the robustness of this process, even when dealing with small image sets. This relies on a human operator and his intuitive knowledge of the main properties of the objects commonly present in scenes. In this thesis we propose various approaches for the camera and model constraints satisfaction. In particular, we propose an original method for the satisfaction of constraints in an exact way. Contrary to most existing work, each proposed method is completed by algorithms dealing with problems engendered by insufficient user-provided input. The introduced methods are validated by reconstruction of 3D models from small image sets, and even from single images. The images used for reconstructions are taken from various sources, such as the Internet, postcards or architectural drawings.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00006129
Contributor : Marta Wilczkowiak <>
Submitted on : Wednesday, May 19, 2004 - 7:47:15 PM
Last modification on : Wednesday, April 11, 2018 - 1:52:17 AM
Long-term archiving on: Friday, April 2, 2010 - 8:51:46 PM

Identifiers

  • HAL Id : tel-00006129, version 1

Collections

INRIA | IMAG | CNRS | UGA

Citation

Marta Wilczkowiak. 3D Modeling From Images Using Geometric Constraints. Human-Computer Interaction [cs.HC]. Institut National Polytechnique de Grenoble - INPG, 2004. English. ⟨tel-00006129⟩

Share

Metrics

Record views

258

Files downloads

871