A method to pan-sharpen SAR amplitude data via electromagnetic reflectance similarities

Abstract

ABSTRACT For a long time, the idea of fusing synthetic radar imagery (SAR) and higher-resolution panchromatic imagery, with the aim of increasing the spatial resolution of radar imagery, has been conceived as impossible. The challenge is due to the different natures of microwave radiation and visible radiation, where the former is the spectrum of SAR imagery, and the latter is of panchromatic imagery. Average reflection amplitude, which constitutes pixel values on both panchromatic and SAR imagery, is taken as a common framework around which this study proposes a method that enables the translation of an array on a panchromatic spectrum into an array of C-band microwave spectrum in order to achieve a higher spatial resolution. In this regard, open-access Worldview-2 and Worldview-3 panchromatic and Sentinel-1 (amplitude-only) images are used. The outputs are qualified using spatial correlation coefficient (SCC), net histogram curve, and R2 difference, and Erreur relative globale adimensionnelle de synthese (ERGAS) as full-reference image quality assessment methods in various scenarios. The results are tenable enough to consider this method for ordinary earth observation and monitoring objectives.