Implementation of Super Resolution Techniques in Geospatial Satellite Imagery

Abstract

The potential for more precise land cover classifications and pattern analysis is provided by technological advancements and the growing accessibility of high-resolution satellite images, which might significantly improve the detection and quantification of land cover change for conservation. A group of methods known as "super-resolution imaging" use generative modelling to increase the resolution of an imaging system. Super-Resolution Imaging, which falls under the category of sophisticated computer vision and image processing, has a variety of practical uses, including astronomical imaging, surveillance and security, medical imaging, and satellite imaging. As computer vision is where deep learning algorithms for super-resolution first appeared, they were mostly created on RGB images in 8-bit colour depth, where the sensor and camera are separated by a few meters. But no evaluation of these methods has been done.