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

2026-01-30

AI Summary

This work addresses the challenge of lacking real high-resolution ground truth in hyperspectral image super-resolution by proposing an unsupervised approach. The method first performs hyperspectral unmixing to extract endmembers and abundance maps, then leverages a dead leaves statistical model to generate fully synthetic abundance data for training a neural network, thereby enabling super-resolution reconstruction without requiring real high-resolution labels. To the best of our knowledge, this is the first study to employ entirely synthetic abundance maps for hyperspectral image super-resolution, effectively circumventing reliance on ground-truth data. Experimental results demonstrate that the proposed method produces high-quality super-resolved images, confirming the feasibility and effectiveness of using synthetic data for unsupervised hyperspectral image reconstruction.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Other Foundations of Computer Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

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.

Problem

• hyperspectral super-resolution
• unsupervised learning
• synthetic data
• spatial resolution enhancement
• spectral preservation

Innovation

• unsupervised learning
• hyperspectral super-resolution
• synthetic data
• abundance maps
• dead leaves model