Enhancing SRTM digital elevation models with deep-learning-based super-resolution image generation

Digital elevation models are responsible for providing altimetric information on a surface to be mapped. While global of low and medium spatial resolution available open source by several space agencies, the high-resolution ones, which utilized in scales 1:25,000 larger, scarce expensive. Here we address this limitation utilization deep learning algorithms coupled with SISR techniques digital obtain better quality versions from lower inputs. The development GAN-based methodology enables improvement initial low-resolution images. A dataset different pairs was created objective allowing study carried out, promoting emergence new research groups area as well enabling comparison between results obtained. It has been found that increasing number iterations performance generated model improved image increased. Furthermore, visual analysis against high- ones showed great similarity first two.