Alpha-trimmed lexicographical extrema for pseudo-morphological image analysis

Abstract : The extension of mathematical morphology to colour, and more generally to multivariate image data, continues to be an open problem. As its underlying theory is defined in terms of complete lattices, the main challenge lies in introducing a complete lattice structure on the image intensity range, hence vectorial extrema computation methods are necessary. In this paper, we circumvent the need for a multivariate ordering, and propose a method for directly computing the multivariate extrema of vector sets. To this end the α-trimming principle is employed in combination with lexicographical ordering. The resulting pseudo-morphological operators, although deprived of important properties, present the advantage of a “collective” calculation, taking into account the distribution of vectors within the structuring element. They are tested against state of the art methodologies in applications treating noise reduction and texture classification, where they are shown to exhibit superior performances.
Type de document :
Article dans une revue
Journal of Visual Communication and Image Representation, Elsevier, 2008, 19 (3), pp.165-174. <10.1016/j.jvcir.2007.10.001>
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00512672
Contributeur : Sébastien Lefèvre <>
Soumis le : mardi 31 août 2010 - 11:20:26
Dernière modification le : mardi 31 août 2010 - 11:44:19
Document(s) archivé(s) le : mercredi 1 décembre 2010 - 02:43:49

Fichier

jvcir2008.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

Citation

Erchan Aptoula, Sébastien Lefèvre. Alpha-trimmed lexicographical extrema for pseudo-morphological image analysis. Journal of Visual Communication and Image Representation, Elsevier, 2008, 19 (3), pp.165-174. <10.1016/j.jvcir.2007.10.001>. <hal-00512672>

Partager

Métriques

Consultations de
la notice

91

Téléchargements du document

89