A new public Alsat-2B dataset for single-image super-resolution

TL;DR

This paper introduces Alsat2B, a new public dataset of paired low and high-resolution remote sensing images created via pan-sharpening, to facilitate the development of super-resolution methods without relying on down-sampling.

Contribution

It presents a novel dataset for remote sensing super-resolution, addressing the scarcity of high-res image pairs and enabling more realistic training scenarios.

Findings

Super-resolution methods show promising results on the dataset.

The dataset highlights challenges requiring advanced super-resolution techniques.

Performance varies across different methods, indicating room for improvement.

Abstract

Currently, when reliable training datasets are available, deep learning methods dominate the proposed solutions for image super-resolution. However, for remote sensing benchmarks, it is very expensive to obtain high spatial resolution images. Most of the super-resolution methods use down-sampling techniques to simulate low and high spatial resolution pairs and construct the training samples. To solve this issue, the paper introduces a novel public remote sensing dataset (Alsat2B) of low and high spatial resolution images (10m and 2.5m respectively) for the single-image super-resolution task. The high-resolution images are obtained through pan-sharpening. Besides, the performance of some super-resolution methods on the dataset is assessed based on common criteria. The obtained results reveal that the proposed scheme is promising and highlight the challenges in the dataset which shows the need for advanced methods to grasp the relationship between the low and high-resolution patches.