A new pan-sharpening method based on cartoon-texture decomposition and compressed sensing

Abstract

A novel pan-sharpening method based on cartoon-texture decomposition and compressed sensing is proposed in this paper. The textures of low resolution multispectral and high resolution panchromatic images are used to reconstruct the high resolution multispectral textures by employing the compressed sensing approach. Experiments on two data sets which are collected by WorldView-3 and QuickBird satellites show the effectiveness of the proposed method from the objective and visual perspectives.