Skip to Main content Skip to Navigation

Modèles déformables et multirésolution pour la détection de contours de traitement d'images

Abstract : Deformable models or active contours are used for edge detection in image analysis and picture processing.

We purpose in this thesis, to study these models in a multiresolution environment.

First, we study active contours at high resolution, and we prove the existence of solutions for closed snakes and open snakes with free boundaries. We also present a new model wich is called: The Deformable Bubble. This model has advantage to be discrete and less sensitive to noise and texture. Moreover it works by small deformations, allowing hence to track the evolution of the curve.

Then, we study a lot of multiresolution technics and show advantages and disavantages of each one. We also establish a relation between multiresolution and energy minimization.

Finaly, we present an original approach combining active contours and multiresolution. It consists to reduce image until some level of the pyramide, to detect edges at this level, to take (in several ways) the obtained contour from low resolution to higher resolution until the bottom of the pyramide is reached.
Document type :
Complete list of metadatas
Contributor : Thèses Imag <>
Submitted on : Monday, October 17, 2005 - 11:31:15 AM
Last modification on : Friday, November 6, 2020 - 3:52:51 AM
Long-term archiving on: : Friday, April 2, 2010 - 10:46:37 PM


  • HAL Id : tel-00010656, version 1




Youssef Elomary. Modèles déformables et multirésolution pour la détection de contours de traitement d'images. Interface homme-machine [cs.HC]. Université Joseph-Fourier - Grenoble I, 1994. Français. ⟨tel-00010656⟩



Record views


Files downloads