Accès direct au contenu

CMLA

Version anglaise

aide

Accueil > Publications > Thèses du CMLA

Recherche - Valorisation

Geometry-aware patch modelling for 3D reconstruction and image denoising

le 20 juin 2016
14h00

Soutenance d'HDR de Gabriele Facciolo (CMLA)

Facciolo_Gabriele.jpg

Facciolo_Gabriele.jpg

The objective of this report is to provide an overview of my postdoctoral research activities. A part of this work is motivated by studies proposed by CNES (Centre National d'Etudes Spatiales) in the context of the MISS (Mathématiques de l'Imagerie Stéréoscopique Spatiale) project for the development of a low baseline stereo pipeline for the Pleiades satellites.

Concretely, I discuss the study of the mathematical and geometrical issues linked to the automated 3D reconstruction from high resolution satellite images, and the development of the theory and practice of reliable stereo matching algorithms. The algorithms for 3D reconstruction and the satellite stereo pipeline (S2P) developed in this context have been adopted as the CNES's official stereo pipeline.

A second line of work covered in this report concerns image denoising. As part of a collaboration with DxO Labs this work aims at inventing and industrializing the state-of-the-art in denoising algorithms. The contributed algorithms are currently beating the stateoftheart.


 
Type :
Thèses - HDR
Lieu(x) :
Campus de Cachan
Bâtiment Laplace, Salle Renaudeau, R-de-Ch.













Présentation de Gabriele Facciolo


Gabriele FACCIOLO
  • Universitat Pompeu Fabra, BCN, Spain
  • University of the Republic, Faculty of Engineering, Montevideo, Uruguay
  • CMLA, ENS Cachan, CNRS, France

Jury de thèse

Garant :                  
Jean-Michel MOREL (CMLA, ENS Cachan)

Rapporteurs :                   
Julie DELON (univ. Paris Descartes)
Laure BLANC-FÉRAUD  (univ. Nice Sophia Antipolis)
Luis ALVAREZ  (univ. Las Palmas de Gran Canaria)

Examinateurs:                 
Mila NIKOLOVA (CMLA, ENS Cachan)
Pascal MONASSE (Ecole des Ponts ParisTech)
Andrès ALMANSA (Telecom ParisTech)
Guillermo SAPIRO (Duke University)

Recherche d'une actualité

Recherche d'une actualité