Enhancing super-resolution in remote sensing: Integrating GIS data with CNN-based SRGAN models for improved image reconstruction

In the field of remote sensing (RS), image super-resolution (SR) techniques play a crucial role across various applications. Traditional SR methods face challenges when applied to long-term coverage datasets with limited spatial resolution. However, recent advancements in deep learning have opened u...

Full description

Saved in:
Bibliographic Details
Published in:E3S web of conferences 2024, Vol.592, p.5006
Main Authors: Sharshov, Ivan, Shoshina, Kseniya, Vasendina, Irina, Aleshko, Roman, Berezovsky, Vladimir
Format: Article
Language:English
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In the field of remote sensing (RS), image super-resolution (SR) techniques play a crucial role across various applications. Traditional SR methods face challenges when applied to long-term coverage datasets with limited spatial resolution. However, recent advancements in deep learning have opened up new possibilities for improving the spatial resolution of RS data. While many convolutional neural network (CNN)- based approaches have achieved excellent performance in developing efficient end-to-end SR models for natural images, they have been less frequently applied to satellite image upscaling with high scale factors. This paper introduces a novel CNN block that enhances the performance of SRGAN-based models. Experimental results show that these architectures benefit from additional data, especially when low-resolution images provide insufficient feature information.
ISSN:2267-1242
2267-1242
DOI:10.1051/e3sconf/202459205006