Heterogeneous Multidimensional Data Deblurring

Ferréol Soulez (1, 2) , Éric Thiébaut (1) , Alain Gressard (3) , Raphaël Dauphin (3) , Sébastien Bongard (4)

1 CRAL - Centre de Recherche Astrophysique de Lyon
2 LHC - Laboratoire Hubert Curien
3 Service de Cardiologie de l'hopital de la Croix Rousse
4 LBNL - Lawrence Berkeley National Laboratory [Berkeley]

We present a new scheme for deconvolution of heterogeneous multidimensional data (\eg spatio-temporal or spatio-spectral). It is derived, in a very general way, following an inverse problem approach. This method exploits the continuity of both object and PSF along the different dimensions to elaborate separable constraints. This improves the effectiveness and the robustness of the deconvolution technique. We demonstrate these improvements by processing real X-ray video sequences $(x,y,t)$ and astronomical multi-spectral images $(x,y,\lambda)$.

Domains: Signal and Image Processing; Signal and Image processing

Ferréol Soulez, Éric Thiébaut, Alain Gressard, Raphaël Dauphin, Sébastien Bongard. Heterogeneous Multidimensional Data Deblurring. 16th European Signal Processing Conference (EUSIPCO 2008), Aug 2008, Lausanne, Switzerland. pp.0. ⟨ujm-00293660⟩