Fast Multispectral and Hyperspectral Image Fusion via Hessian Inversion - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

Fast Multispectral and Hyperspectral Image Fusion via Hessian Inversion

Résumé

Multispectral and hyperspectral data fusion allows data restoration with increased spatial and spectral resolutions. A common approach, in particular with the presence of variant blur, is to solve an ill-posed inverse problem by minimizing a mixed criterion. This minimization usually requires an iterative gradient-based method. Still, this paper demonstrates the existence of a reachable explicit solution without the need to solve a Sylvester equation, allowing for a three order of magnitude timesaving factor wrt. iterative algorithms. We exploit this explicit resolution to accelerate the minimization by a Majorize-Minimize algorithm of a semi-quadratic convex edge-preserving criterion. We conduct experiments on realistic synthetic measurements for the James Webb Space Telescope and show that our proposed solution outperforms the state-of-the-art in computation time and quality of reconstruction.
Fichier principal
Vignette du fichier
1main.pdf (968.6 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04337970 , version 1 (12-12-2023)

Identifiants

  • HAL Id : hal-04337970 , version 1

Citer

Dan Pineau, François Orieux, Alain Abergel. Fast Multispectral and Hyperspectral Image Fusion via Hessian Inversion. 2023. ⟨hal-04337970⟩
36 Consultations
25 Téléchargements

Partager

Gmail Facebook X LinkedIn More