Remote sensing image super-resolution: Challenges and approaches

Remote sensing has a growing relevance in the modern society with the development of image processing of satellite imagery. However, due to the limitations of the current imaging sensors and the complex atmospheric conditions, we are facing great challenges in the remote sensing applications due to the limited spatial, spectral, radiometric and temporal resolutions. Therefore, super-resolution techniques have attracted much attention by which the low quality low resolution remote sensing images are enhanced. In this paper, we discuss the challenges in remote sensing image super-resolution and thereafter review the relevant approaches. More specifically, the different categories of remote sensing techniques, i.e., the learning-based, interpolation based, frequency domain based, and probability based methods, are reviewed and discussed. Furthermore, the super-resolution applications are discussed and insightful comments on future research directions are provided.