A Method to Better Account for Modulation Transfer Functions in ARSIS-Based Pansharpening Methods

Abstract

Massip P., Blanc Ph., Wald L., A method to better account for modulation transfer functions in ARSIS-based pan-sharpening methods. IEEE Transactions in Geoscience and Remote Sensing, 99, 1-9, 2011, doi : 10.1109/TGRS.2011 1 Abstract-Multispectral (MS) images provided by EarthObservation satellites have generally a poor spatial resolution while panchromatic images (PAN) exhibit a spatial resolution two or four times better. Data fusion is a mean to synthesize MS images at higher spatial resolution than original by exploiting the high spatial resolution of the PAN. This process is often called pan-sharpening. The synthesis property states that the synthesized multispectral images should be as close as possible to those that would have been acquired by the corresponding sensors if they had this high resolution. The methods based on the concept ARSIS are able to deliver synthesized images with good spectral quality but whose geometrical quality can still be improved. We propose a more precise definition of the synthesis property in terms of geometry. Then, we present a method that takes explicitly into account the difference in modulation transfer function (MTF) between PAN and MS in the fusion process. This method is applied to an existing ARSIS-based fusion method: ATWT-M3. Simulated images of the sensors Pleiades and SPOT-5 are used to illustrate the performances of the approach. Though this study is limited in methods and data, we observe a better restitution of the geometry and an improvement in all indices classically used in quality budget in pan-sharpening. We present also a means to assess the respect of the synthesis property from a MTF point of view.

A New Look at Image Fusion Methods from a Bayesian Perspective

“…Setting = , / , produces the M3 modulation coefficient [11,68] corresponding to the GS method in CS category (Table 1). We can derive the weight from the modulation coefficient,…”

“…In fact, the up-sampled images are spectrally consistent only if the sensor PSF is ideal and the nearest neighbor resampling method is applied. Massip et al [68] recently demonstrated that de-convoluting the MS images according to the sensor PSF before up-sampling could improve the fusion results, which in fact indicates that the assumption Hx̃ = x is more reasonable when the sensor PSF is more like ideal after de-convolution.…”

Bio-empirical mode decomposition: visible and infrared fusion using biologically inspired empirical mode decomposition

“…Several techniques offer the conversion of multispectral images into intensity hue saturation (IHS). 22,23 This method has the advantage of adding the spectral quality of the color image to high resolution details from the panchromatic image. Similar frameworks have been applied utilizing pyramid-based fusion method.…”

Scale-Aware Pansharpening Algorithm for Agricultural Fragmented Landscapes

“…The novelty of the method relies on the fact that filters matching the MTFs of the different channels of the imaging instruments are used to extract the spectral (low pass) and spatial (high pass) information from the merged image. However, Massip et al [19] concluded that the differences in sampling rates, pixel size, integration time and spectral sensitivity, between the MS and PAN sensors yield a difference in MTF. This difference must be taken into account when synthesizing MS images at the PAN high-spatial resolution.…”