The methods developed will be evaluated on real remote sensing images provided by the French Space Agency (CNES Toulouse).
Over the last decades, an increasing number of airborne sensors have been acquiring remotely sensed images characterized by various spatial and spectral resolutions. Due to inherent physical constraints, a tradeoff between both resolutions must be found. However, current and upcoming Earth observation missions include several sensors that are able to jointly provide hyperspectral images and highly resolved multispectral or panchromatic images. Coupling both acquired images to obtain a unique high-resolution hyperspectral image is a challenging problem. We propose to formulate this task as a fusion problem that can be conveniently solved within a Bayesian framework. Recently developed efficient Markov chain Monte Carlo (MCMC) methods will be used to recover the estimated image.

The knowledge required for this work includes statistical signal/image processing, estimation, and detection.