I. Introduction
Satellite remote sensing images have found widespread applications in Earth surface observations, including agricultural monitoring, change detection, and geophysical variable estimation [1], [2]. Owing to hardware limitations of the imaging sensors, as well as other practical concerns such as atmospheric absorption and satisfactory signal-to-noise ratio (SNR), the spatial resolution of multiband systems varies across the spectral bands. In such a scenario, many systems adopt ground sampling distances (GSDs) varying across the spectral bands in order to make an efficient use of the system resources and the communication bandwidth. Some well-known multispectral multiresolution satellites include the Moderate Resolution Imaging Spectroradiometer (MODIS) [3], Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) [4], as well as the very recent, Sentinel-2 launched by the European Space Agency (ESA) [5]. To facilitate the effectiveness of analyzing these images, super-resolving the lower resolution bands to make their spatial resolution the same as those with the highest resolution ones is of paramount importance.