Post-doc proposal Blind and unsupervised detector calibration for satellite imaging

Preliminary works have already been performed between IRCCyN, IRAP and CNES. In particular, the paper [1] proposes a blind estimation method under the assumption of a purely linear response. A possible direction of the proposed project could be to generalize the latter method. On the one hand, an affine model would be more realistic. On the other hand, a totally unsupervised method would be preferable. Finally, a dedicated, fast optimization method should be proposed. The obtained performance in terms of estimation quality will be compared on real satellite images with existing destriping methods.