Offical Python codes for the paper "Hyperspectral super-resolution by unsupervised convolutional neural network and SURE", in Proceeding of IGARSS 2022, Kuala Lumpur, July 2022.
Authors: Han V. Nguyen
Email: [email protected]
Recent advances in deep learning (DL) reveal that the structure of a convolutional neural network (CNN) is a good image prior (called deep image prior (DIP)), bridging the model-based and DL-based methods in image restoration. However, optimizing a DIP-based CNN is prone to overfitting leading to a poorly reconstructed image. This paper derives a loss function based on Stein's unbiased risk estimate (SURE) for unsupervised training of a DIP-based CNN applied to the hyperspectral image (HSI) super-resolution. The SURE loss function is an unbiased estimate of the mean-square-error (MSE) between the clean low-resolution image and the low-resolution estimated image, which relies only on the observed low-resolution image. Experimental results on HSI show that the proposed method not only improves the performance, but also avoids overfitting.
Please cite our paper if you are interested
@inproceedings{nguyen2022hyperspectral,
title={Hyperspectral Super-Resolution by Unsupervised Convolutional Neural Network and Sure},
author={Nguyen, Han V and Ulfarsson, Magnus O and Sveinsson, Johannes R and Dalla Mura, Mauro},
booktitle={IGARSS 2022-2022 IEEE International Geoscience and Remote Sensing Symposium},
pages={903--906},
year={2022},
organization={IEEE}
}
The following folders contanin:
- data: The simulated PU dataset.
- models: python scripts define the model (network structure)
- utils: additional functions
Run the jupyter notebook and see results.
- Pytorch 1.8
- Matplotlib
- Numpy, Scipy, Skimage.