Towards Transformer-based Homogenization of Satellite Imagery for Landsat-8 and Sentinel-2

TL;DR

This paper explores using transformer-based models to homogenize satellite imagery from Landsat-8 and Sentinel-2, aiming to improve compatibility and cloud-free image availability, but finds UNet outperforms transformers.

Contribution

It introduces a transformer-based approach for satellite image homogenization and compares its performance to a UNet model, revealing unexpected results.

Findings

Deep models outperform classical approaches.

UNet significantly outperforms the transformer model.

Transformers show potential but are not yet superior.

Abstract

Landsat-8 (NASA) and Sentinel-2 (ESA) are two prominent multi-spectral imaging satellite projects that provide publicly available data. The multi-spectral imaging sensors of the satellites capture images of the earth's surface in the visible and infrared region of the electromagnetic spectrum. Since the majority of the earth's surface is constantly covered with clouds, which are not transparent at these wavelengths, many images do not provide much information. To increase the temporal availability of cloud-free images of a certain area, one can combine the observations from multiple sources. However, the sensors of satellites might differ in their properties, making the images incompatible. This work provides a first glance at the possibility of using a transformer-based model to reduce the spectral and spatial differences between observations from both satellite projects. We compare the results to a model based on a fully convolutional UNet architecture. Somewhat surprisingly, we find that, while deep models outperform classical approaches, the UNet significantly outperforms the transformer in our experiments.