Skip to Main content Skip to Navigation

Imagerie Mathématique: segmentation sous contraintes géométriques ~ Théorie et Applications

Abstract : In this thesis, we are concerned with the issue of image segmentation under geometrical constraints. This problematics has emerged while analyzing classical methods of edge detection. Indeed, these classical tools (deformable models, geodesic active contours, fast marching, etc...) prove to be fruitless in several scenarios: when image data are missing or of poor quality. In medical imaging for instance, occlusion phenomena can occur: two organs can partly hide each other (e.g of the liver). Besides, two adjacent objects can own homogeneous intrinsic texture so that it is hard to clearly identify the interface beween both items. The classical definition of an edge which is features as the locus of connected points for which the image gradient varies abruptly can no longer be applied. To finish with, in some fields of research and/or for post-processing needs, one can have at one's disposal, in addition to image data, geometrical data to be integrated in the segmentation process.
To cope with these hindrances, we propose to design segmentation models that integrate geometrical constraints while satisfying the classical criteria of detection with in particular, the regularity that implies on the contour.
Document type :
Complete list of metadatas
Contributor : Carole Le Guyader <>
Submitted on : Friday, April 15, 2005 - 7:39:26 PM
Last modification on : Wednesday, May 22, 2019 - 10:12:02 AM
Long-term archiving on: : Friday, April 2, 2010 - 10:12:04 PM


  • HAL Id : tel-00009036, version 1


Carole Le Guyader. Imagerie Mathématique: segmentation sous contraintes géométriques ~ Théorie et Applications. Mathématiques [math]. INSA de Rouen, 2004. Français. ⟨tel-00009036⟩



Record views


Files downloads