Pansharpening: an Application to Satellite Image Fusion

Description | Many Earth observation satellites provide continuously growing quantities of remote sensing images useful for a wide range of both scientific and everyday tasks. Most of them, such as Ikonos, Landsat, Quickbird, and Pléiades, decouple the acquisition of a panchromatic image at high spatial resolution from the acquisition of a multispectral image at lower spatial resolution. The wide range of wavelengths acquired by the panchromatic represents an accurate description of the geometry, while each spectral component covers a reduced bandwidth range leading to a detailed colour description useful for material classification.

Imaging sensors are subject to compromises due to various technical and economical constraints. The bandwidth capacity as well as the onboard storage are important limiting factors too.

Pansharpening refers to the fusion of a panchromatic (PAN) and a multispectral (MS) image simultaneously acquired over the same area. This can be seen as a particular problem of data fusion since one would aim at combining the spatial details resolved by the PAN (but not present in the MS) and the several spectral bands of the MS image (against the single band of the PAN) in a unique product. The PAN image has a high spatial resolution and poor spectral resolution (for instance, a grey level image) while the MS images have a low spatial resolution but a high spectral resolution (for instance, the blue, green, red and infra-red images). The resulting image Pansharpened is expected to have the spatial resolution of PAN and the high spectral resolution of MS.

How can we have a first solution to the Pansharpening problem? That is, interpolate the MS images from the geometry provided by the PAN image.

Some methods:
- component substitution (CS), i.e., passing to an image system that we can decompose into intensity + colour. Then substitute the intensity by the PAN image and reverse the process.
- modelling from a functional. In this case, we could apply a gradient descent type algorithm for minimization

Mathematical background | Basic programming (C++, Phyton); Mathematical optimization, interpolation, modelling.