Color Constancy by Derivative-based Gamut Mapping

Arjan Gijsenij 1 Theo Gevers 1 Joost Van de Weijer 2
2 LEAR - Learning and recognition in vision
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : Color constancy aims to compute object colors despite differences in the color of the light source. Gamut-based approaches are very promising methods to achieve color constancy. In this paper, the gamut mapping approach is extended to incorporate higher-order statistics (derivatives) to estimate the illuminant. A major problem of gamut mapping is that in case of a failure of the diagonal model no solutions are found, and therefore no illuminant estimation is performed. Image value offsets are often used to model deviations from the diagonal model. Prior work which incorporated robustness to offsets for gamut mapping assumed a constant offset over the whole image. In contrast to previous work, we model these offsets to be position dependent, and show that for this case derivative-based gamut mapping yields a valid solution to the illuminant estimation problem. Experiments on both synthetic data and images taken under controlled laboratory settings reveal that the derivativebased and regular gamut mapping methods provide similar performance. However, the derivative-based method outperforms other methods on the more challenging task of color constancy for real-world images.
Type de document :
Communication dans un congrès
Peter Belhumeur and Katsushi Ikeuchi and Emmanuel Prados and Stefano Soatto and Peter Sturm. PACV - 1st International Workshop on Photometric Analysis For Computer Vision, Oct 2007, Rio de Janeiro, Brazil. INRIA, 8 p., 2007
Liste complète des métadonnées

Littérature citée [21 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/inria-00264813
Contributeur : Thoth Team <>
Soumis le : mardi 18 mars 2008 - 09:45:38
Dernière modification le : mercredi 11 avril 2018 - 01:59:31
Document(s) archivé(s) le : vendredi 28 septembre 2012 - 11:20:17

Fichier

p21.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : inria-00264813, version 1

Collections

Citation

Arjan Gijsenij, Theo Gevers, Joost Van de Weijer. Color Constancy by Derivative-based Gamut Mapping. Peter Belhumeur and Katsushi Ikeuchi and Emmanuel Prados and Stefano Soatto and Peter Sturm. PACV - 1st International Workshop on Photometric Analysis For Computer Vision, Oct 2007, Rio de Janeiro, Brazil. INRIA, 8 p., 2007. 〈inria-00264813〉

Partager

Métriques

Consultations de la notice

476

Téléchargements de fichiers

454