| HAL : hal-00204984, version 1 |
| Fiche détaillée | Récupérer au format |
|
|
| Multiscale Modeling and Simulation: A SIAM Interdisciplinary Journal: A SIAM Interdisciplinary Journal 6, 2 (2007) 547-576 |
|
|
|
|
| Denoising of frame coefficients using l^1 data-fidelity term and edge preserving regularization |
|
|
| Sylvain Durand 1Mila Nikolova 2 |
|
|
| (2007) |
|
|
| We consider the denoising of a function containing smooth regions and edges. Classical ways to solve this problem are variational methods amd shrinkage of a representation of the data in a basis or a frame. We propose a method which combines the advantages of both approaches. followinig the wavelet shrinkage method of Donoho and Johnstone, we set to zero all frame coefficients correponding to noise and some coefficients, erroneously set to zero, leading to Gibbs-lika oscillations in the estimate. We design a specialized objective function allowing all these coefficients to be selectively restored, without modifying the other coefficients which are well nealy faithful, using regularized in the domain of the restored function. We analyse the well-posedness and the main properties of this objective function. |
|
|
|
|
|
|
|
|
|
|
| 1 : | Laboratoire Amiénois de Mathématique Fondamentale et Appliquée (LAMFA) |
| CNRS : UMR6140 – Université de Picardie Jules Verne | |
| 2 : | Centre de Mathématiques et de Leurs Applications (CMLA) |
| CNRS : UMR8536 – École normale supérieure de Cachan - ENS Cachan | |
|
|
|
|
|
|
|
|
| Domaine | : | Sciences de l'ingénieur/Traitement du signal et de l'image Informatique/Traitement du signal et de l'image |
|
|
| wavelets – variational method – denoising |
|
|
| Liste des fichiers attachés à ce document : | |||||
|
|
|
| hal-00204984, version 1 | |
| http://hal.archives-ouvertes.fr/hal-00204984 | |
| oai:hal.archives-ouvertes.fr:hal-00204984 | |
| Contributeur : Sylvain Durand | |
| Soumis le : Mercredi 16 Janvier 2008, 10:44:40 | |
| Dernière modification le : Mercredi 16 Janvier 2008, 11:20:22 | |