Synthetic Abundance Maps for Unsupervised Super-Resolution of Hyperspectral Remote Sensing Images

2026-01-30

AI Summary

This work proposes an unsupervised framework for hyperspectral image super-resolution that circumvents the need for hard-to-acquire ground-truth high-resolution labels. The method first performs spectral unmixing to obtain endmembers and abundance maps, then leverages synthetically generated abundance maps—based on the dead-leaf model—to train a neural network for abundance super-resolution. High-resolution hyperspectral images are subsequently reconstructed by combining the original endmembers with the enhanced abundance maps. By introducing synthetic abundance data for unsupervised training, this approach effectively enhances spatial resolution without requiring real high-resolution labels, while preserving both spectral fidelity and spatial details. Experimental results validate the efficacy of training with synthetic abundances and demonstrate the superior performance of the proposed method in unsupervised hyperspectral super-resolution tasks.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Hyperspectral single image super-resolution (HS-SISR) aims to enhance the spatial resolution of hyperspectral images to fully exploit their spectral information. While considerable progress has been made in this field, most existing methods are supervised and require ground truth data for training-data that is often unavailable in practice. To overcome this limitation, we propose a novel unsupervised training framework for HS-SISR, based on synthetic abundance data. The approach begins by unmixing the hyperspectral image into endmembers and abundances. A neural network is then trained to perform abundance super-resolution using synthetic abundances only. These synthetic abundance maps are generated from a dead leaves model whose characteristics are inherited from the low-resolution image to be super-resolved. This trained network is subsequently used to enhance the spatial resolution of the original image's abundances, and the final super-resolution hyperspectral image is reconstructed by combining them with the endmembers. Experimental results demonstrate both the training value of the synthetic data and the effectiveness of the proposed method.

Problem

• hyperspectral super-resolution
• unsupervised learning
• ground truth unavailability
• spatial resolution enhancement
• hyperspectral remote sensing

Innovation

• unsupervised super-resolution
• synthetic abundance maps
• hyperspectral image
• dead leaves model
• spectral unmixing