DATA FUSION AND UNMIXING WITH THE REGULARIZED NON-NEGATIVE BLOCK-TERM DECOMPOSITION: JOINT PROBLEMS, BLIND APPROACH AND AUTOMATIC MODEL ORDER SELECTION - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

DATA FUSION AND UNMIXING WITH THE REGULARIZED NON-NEGATIVE BLOCK-TERM DECOMPOSITION: JOINT PROBLEMS, BLIND APPROACH AND AUTOMATIC MODEL ORDER SELECTION

Résumé

This paper introduces a family of coupled tensor optimization problems for joint super-resolution and unmixing in remote sensing. Using β-divergences allows the proposed methods to account for various noise statistics. A family of simple, efficient and flexible algorithms is proposed, that are capable of solving the two problems at hand. Moreover, the proposed algorithms are able to estimate the degradation operators mapping the HSI and MSI to the unknown SRI. We introduce penalized versions of our optimization problems, with an emphasis on the minimum-volume regularization approach. This approach is designed to efficiently identify the number of factors in the tensor decomposition, and to effectively manage scenarios involving potential rank deficiencies in the estimated mixing factors. It facilitates the computation of interpretable and meaningful tensor decompositions, and enhances the identifiability of the decomposition model. The proposed algorithms demonstrate competitive performance against state-of-the-art methods for joint fusion and unmixing, even in scenarios with various noise statistics and challenging cases, including partially unknown degradation operators, almost collinear materials, and estimation of the number of endmembers.
Fichier principal
Vignette du fichier
Paper_LongVersion_V2.pdf (2.47 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04302397 , version 1 (23-11-2023)

Identifiants

  • HAL Id : hal-04302397 , version 1

Citer

Clémence Prévost, Valentin Leplat. DATA FUSION AND UNMIXING WITH THE REGULARIZED NON-NEGATIVE BLOCK-TERM DECOMPOSITION: JOINT PROBLEMS, BLIND APPROACH AND AUTOMATIC MODEL ORDER SELECTION. 2023. ⟨hal-04302397⟩
43 Consultations
62 Téléchargements

Partager

Gmail Facebook X LinkedIn More