Data fusion: taking into account the modulation transfer function in ARSIS-based pansharpening methods

Multispectral images provided by satellite have a poor spatial resolution while panchromatic images (PAN) exhibit a spatial r esolution two or four times better. Data fusion is a mean to synthesize MS imag es at higher spatial resolution than original by exploiting the high spatial resolu tion of the PAN. This process is often called pan-sharpening. The synthesized multis pectral images should be as close as possible to those that would have been acq uired by the corresponding sensors if they had this high resolution. The metho ds based on the concept "Amelioration de la Resolution Spatiale par Injecti on de Structures" (ARSIS) concept are able to deliver synthesized images with good spectral quality but whose geometrical quality can still be enhanced. We propose to consider the characteristics of the sensor to improve the geomet rical quality. We take explicitly into account the modulation transfer function (MTF) of the sensor in the fusion process. Though this study is limited in methods an d data, we observe a better restitution of the geometry and an improvement in t he majority of quality indices classically used in pan-sharpening. The communicati on also presents a means to assess the respect of the synthesis property from a MTF point of view.