HSR_Tucker: Hyperspectral Super-Resolution with Coupled Tucker Approximation

This repository contains the implementation for the methods proposed in the preprint:

@unpublished{prevost:hal-01911969,
 TITLE = {{Hyperspectral super-resolution with coupled Tucker approximation: Identifiability and SVD-based algorithms}},
 AUTHOR = {Pr{\'e}vost, Cl{\'e}mence and Usevich, Konstantin and Comon, Pierre and Brie, David},
 URL = {https://hal.archives-ouvertes.fr/hal-01911969},
 YEAR = {2019},
 HAL_ID = {hal-01911969},
 HAL_VERSION = {v3},
}

It also reproduces the results of the experiments from the preprint.

Installation

TensorLab v.3.0 should be installed and should be on the MATLAB path.

The datasets used in the examples can be downloaded here.

For some of the experiments, the following packages need to be downloaded:

• Codes for the STEREO method for some comparisons with comparisons with the STEREO methods
• HySure package for comparisons with HySure
• Export Fig to save some figures in PDF

Contents

• src contains the source codes for the methods
• demos contains the codes for the reproducivle examples