Deep Hyperspectral Prior: Denoising, Inpainting, Super-Resolution

TL;DR

This paper introduces a deep prior-based method for hyperspectral image denoising, inpainting, and super-resolution that does not require training data, making it effective even with limited datasets.

Contribution

It extends the deep prior approach to hyperspectral images using 3D CNNs, enabling high-quality restoration without training data.

Findings

Performance comparable to trained networks

Effective with limited or no training data

Applicable to denoising, inpainting, and super-resolution

Abstract

Deep learning algorithms have demonstrated state-of-the-art performance in various tasks of image restoration. This was made possible through the ability of CNNs to learn from large exemplar sets. However, the latter becomes an issue for hyperspectral image processing where datasets commonly consist of just a few images. In this work, we propose a new approach to denoising, inpainting, and super-resolution of hyperspectral image data using intrinsic properties of a CNN without any training. The performance of the given algorithm is shown to be comparable to the performance of trained networks, while its application is not restricted by the availability of training data. This work is an extension of original "deep prior" algorithm to HSI domain and 3D-convolutional networks.