Graph-based ordering scheme for color image filtering

Olivier Lezoray (1) , Cyril Meurie (2) , Abderrahim Elmoataz (1)

1 Equipe Image - Laboratoire GREYC - UMR6072
2 INRETS/LEOST - Laboratoire Electronique, Ondes et Signaux pour les Transports

This paper presents a graph-based ordering scheme of color vectors. A complete graph is defined over a filter window and its structure is analyzed to construct an ordering of color vectors. This graph-based ordering is constructed by finding a Hamiltonian path across the color vectors of a filter window by a two-step algorithm. The first step extracts, by decimating a minimum spanning tree, the extreme values of the color set. These extreme values are considered as the infimum and the supremum of the set of color vectors. The second step builds an ordering by constructing a Hamiltonian path among the vectors of color vectors, starting from the infimum and ending at the supremum. The properties of the proposed graph-based ordering of vectors are detailed. Several experiments are conducted to assess its filtering abilities for morphological and median filtering.

Mots clés: Ordering of vectors; color; filtering; mathematical morphology; median; graph; Hamiltonian path

Domaines: Traitement des images [eess.IV]

Olivier Lezoray, Cyril Meurie, Abderrahim Elmoataz. Graph-based ordering scheme for color image filtering. International Journal of Image and Graphics, 2008, 8 (3), pp.473-493. ⟨hal-00329511⟩