Single-Image Super-Resolution of Sentinel-2 Low Resolution Bands with Residual Dense Convolutional Neural Networks

Luis Salgueiro,Verónica Vilaplana,Javier Marcello

doi:10.3390/rs13245007

Abstract

Sentinel-2 satellites have become one of the main resources for Earth observation images because they are free of charge, have a great spatial coverage and high temporal revisit. Sentinel-2 senses the same location providing different spatial resolutions as well as generating a multi-spectral image with 13 bands of 10, 20, and 60 m/pixel. In this work, we propose a single-image super-resolution model based on convolutional neural networks that enhances the low-resolution bands (20 m and 60 m) to reach the maximal resolution sensed (10 m) at the same time, whereas other approaches provide two independent models for each group of LR bands. Our proposed model, named Sen2-RDSR, is made up of Residual in Residual blocks that produce two final outputs at maximal resolution, one for 20 m/pixel bands and the other for 60 m/pixel bands. The training is done in two stages, first focusing on 20 m bands and then on the 60 m bands. Experimental results using six quality metrics (RMSE, SRE, SAM, PSNR, SSIM, ERGAS) show that our model has superior performance compared to other state-of-the-art approaches, and it is very effective and suitable as a preliminary step for land and coastal applications, as studies involving pixel-based classification for Land-Use-Land-Cover or the generation of vegetation indices.

Highlights

Managed by the European Space Agency (ESA), the Sentinel-2 satellites play an important role in today’s remote sensing as they provide multispectral optical imagery that can be used for several applications such as land cover-land use monitoring, change detection, vegetation and soil analysis, and mapping of physical variables, among others.Some of its characteristics are its considerable surface coverage, the high revisit time [1]and the possibility of getting the data for free, democratizing images for research, as well as launching free and commercial products, making it increasingly useful for Earth observation data.Each satellite provides 13 bands: 4 high-resolution (HR) bands with 10 m/pixel, 6 low-resolution (LR) bands with 20 m/pixel, and 3 very low-resolution (VLR) bands with60 m/pixel
This paper proposes the use of Single-Image Super-Resolution techniques to enhance the spatial quality of the LR and VLR bands, and reach the resolution of HR bands
Due to the physical limitation of sensors and to minimize storage of data, only a small set of bands are provided with maximal spatial resolution

Summary

Introduction

Managed by the European Space Agency (ESA), the Sentinel-2 satellites play an important role in today’s remote sensing as they provide multispectral optical imagery that can be used for several applications such as land cover-land use monitoring, change detection, vegetation and soil analysis, and mapping of physical variables, among others.Some of its characteristics are its considerable surface coverage, the high revisit time [1]and the possibility of getting the data for free (available at [2]), democratizing images for research, as well as launching free and commercial products, making it increasingly useful for Earth observation data.Each satellite provides 13 bands: 4 high-resolution (HR) bands with 10 m/pixel, 6 low-resolution (LR) bands with 20 m/pixel, and 3 very low-resolution (VLR) bands with60 m/pixel. Managed by the European Space Agency (ESA), the Sentinel-2 satellites play an important role in today’s remote sensing as they provide multispectral optical imagery that can be used for several applications such as land cover-land use monitoring, change detection, vegetation and soil analysis, and mapping of physical variables, among others. Spatial resolution is a fundamental parameter for the analysis of remote sensing imagery. It is defined as the minimum distance in which two separated objects are distinguishable, and depends on several factors, for instance altitude, distance, and the quality of the instruments [5]. Another relevant feature, in line with spatial resolution, is the Ground

Methods

Discussion

Conclusion