Accéder directement au contenu Accéder directement à la navigation
Communication dans un congrès

Segmenting junction regions without skeletonization using geodesic operators and the max-tree

Abstract : In a 2D skeleton, a junction region indicates a connected component (CC) where a path splits into two or more different branches. In several real applications, structures of interest such as vessels, cables, fibers, wrinkles, etc. may be wider than one pixel. Since the user may be interested in the junction regions but not in the skeleton itself (e.g. in order to segment an object into single branches), it is reasonable to think about finding junction regions directly on objects avoiding skeletoniza-tion. In this paper we propose a solution to find junction regions directly on objects, which are usually elongated structures, but not necessary thin skeletons, with possible protrusions and noise. Our method is based on geodesic operators and is intended to work on binary objects without holes. In particular, our method is based on the analysis of the wavefront evolution during geodesic propagation. Our method is generic and does not require skeletonization. It provides an intuitive and straightforward way to filter short branches and protuberances. Moreover, an elegant and efficient implementation using a max-tree representation is proposed.
Type de document :
Communication dans un congrès
Liste complète des métadonnées

Littérature citée [23 références]  Voir  Masquer  Télécharger

https://hal-mines-paristech.archives-ouvertes.fr/hal-02430543
Contributeur : Beatriz Marcotegui <>
Soumis le : mardi 7 janvier 2020 - 13:29:16
Dernière modification le : mercredi 14 octobre 2020 - 03:52:38

Fichier

ISMM2019_Junction_regions.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Citation

Andrés Serna, Beatriz Marcotegui, Etienne Decencière. Segmenting junction regions without skeletonization using geodesic operators and the max-tree. 14th International Symposium on Mathematical Morphology and Its Applications to Signal and Image Processing, Jul 2019, Saarbrücken, Germany. ⟨10.1007/978-3-030-20867-7_35⟩. ⟨hal-02430543⟩

Partager

Métriques

Consultations de la notice

97

Téléchargements de fichiers

196