Super-résolution non supervisée d'images hyperspectrales de télédétection utilisant un entraînement entièrement synthétique
This addresses the challenge of enhancing spatial resolution in hyperspectral images for remote sensing applications where ground truth data is often unavailable, representing an incremental improvement over supervised methods.
The paper tackles the problem of hyperspectral single image super-resolution without high-resolution ground truth data by proposing an unsupervised learning approach using synthetic abundance data, achieving effective results as demonstrated in experiments.
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.