Unsupervised Super-Resolution of Hyperspectral Remote Sensing Images Using Fully Synthetic Training

2024-12-09

Workshop on Hyperspectral Image and Signal Processing

AI Summary

This work addresses the challenge of low spatial resolution in hyperspectral remote sensing imagery and the impracticality of obtaining real high-resolution ground-truth labels required by existing super-resolution methods. To overcome this limitation, the authors propose an unsupervised super-resolution approach that first performs spectral unmixing to extract abundance maps and endmembers. Leveraging the dead-leaf model, they generate synthetic abundance data that faithfully preserves realistic spatial statistics, which is then used to train a neural network for abundance super-resolution. The high-resolution hyperspectral image is finally reconstructed by recombining the enhanced abundances with the original endmembers. Notably, this study is the first to achieve fully unsupervised training using entirely synthetic abundance data, thereby eliminating reliance on real high-resolution labels. Experimental results demonstrate that the proposed method significantly improves reconstruction quality under unsupervised conditions, validating the effectiveness and potential of the synthetic abundance strategy.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Unsupervised & Self-Supervised Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Considerable work has been dedicated to hyperspectral single image super-resolution to improve the spatial resolution of hyperspectral images and fully exploit their potential. However, most of these methods are supervised and require some data with ground truth for training, which is often non-available. To overcome this problem, we propose a new unsupervised training strategy for the super-resolution of hyperspectral remote sensing images, based on the use of synthetic abundance data. Its first step decomposes the hyperspectral image into abundances and endmembers by unmixing. Then, an abundance super-resolution neural network is trained using synthetic abundances, which are generated using the dead leaves model in such a way as to faithfully mimic real abundance statistics. Next, the spatial resolution of the considered hyperspectral image abundances is increased using this trained network, and the high resolution hyperspectral image is finally obtained by recombination with the endmembers. Experimental results show the training potential of the synthetic images, and demonstrate the method effectiveness.

Problem

• hyperspectral image super-resolution
• unsupervised learning
• training data scarcity
• remote sensing

Innovation

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