Super-r\'esolution non supervis\'ee d'images hyperspectrales de t\'el\'ed\'etection utilisant un entra\^inement enti\`erement synth\'etique

TL;DR

This paper introduces an unsupervised hyperspectral image super-resolution method that leverages synthetic abundance data generated by the dead leaves model, enabling super-resolution without high-resolution ground truth.

Contribution

The proposed approach is the first to use synthetic abundance data for unsupervised hyperspectral super-resolution, avoiding the need for real high-res training data.

Findings

Effective super-resolution demonstrated on hyperspectral images.

Synthetic data training yields competitive results.

Method preserves spectral information well.

Abstract

Hyperspectral single image super-resolution (SISR) aims to enhance spatial resolution while preserving the rich spectral information of hyperspectral images. Most existing methods rely on supervised learning with high-resolution ground truth data, which is often unavailable in practice. To overcome this limitation, we propose an unsupervised learning approach based on synthetic abundance data. The hyperspectral image is first decomposed into endmembers and abundance maps through hyperspectral unmixing. A neural network is then trained to super-resolve these maps using data generated with the dead leaves model, which replicates the statistical properties of real abundances. The final super-resolution hyperspectral image is reconstructed by recombining the super-resolved abundance maps with the endmembers. Experimental results demonstrate the effectiveness of our method and the relevance of synthetic data for training.