Skip to Main content Skip to Navigation
Theses

Paramétrisation et reconstruction des surfaces développables à partir d'images

Abstract : This thesis deals with the 3D reconstruction of deformable scenes from images. Our methods are based on a mathematical modeling of the scene. The reconstruction is expressed as a parameter estimation problem from images. More precisely, we focus on deformable objects that can be modeled by a developable or an inextensible surface. The developable surfaces are those surfaces are differentiable and isometric to the plane. Our first contribution is a generative model dedicated to those surfaces. The main differences with previous models are its generality and the explicit expression of the object boundary. Our second contribution is an algorithm to reconstruct our model from several synchronized images. Our third contribution is a method to reconstruct inextensible surfaces. It uses the constraints between the points of the surface to recover the 3D surface from a single image and a reference deformation.
Document type :
Theses
Complete list of metadatas

Cited literature [104 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00731113
Contributor : Camille Meyer <>
Submitted on : Wednesday, September 12, 2012 - 10:10:52 AM
Last modification on : Thursday, January 11, 2018 - 6:17:21 AM
Long-term archiving on: : Thursday, December 13, 2012 - 3:44:47 AM

Identifiers

  • HAL Id : tel-00731113, version 1

Citation

Mathieu Perriollat. Paramétrisation et reconstruction des surfaces développables à partir d'images. Automatique / Robotique. Université Blaise Pascal - Clermont-Ferrand II, 2008. Français. ⟨NNT : 2008CLF21876⟩. ⟨tel-00731113⟩

Share

Metrics

Record views

296

Files downloads

2582