BLIND DECONVOLUTION OF VIDEO SEQUENCES

Ferréol Soulez (1, 2) , Éric Thiébaut (1) , Yves Tourneur (3) , Alain Gressard (4) , Raphaël Dauphin (4)

1 CRAL - Centre de Recherche Astrophysique de Lyon
2 LHC - Laboratoire Hubert Curien
3 Centre Commun de Quantimétrie
4 Service de Cardiologie de l'hopital de la Croix Rousse

We present a new blind deconvolution method for video sequence. It is derived following an inverse problem approach in a Bayesian framework. This method exploits the temporal continuity of both object and PSF. Combined with edge-preserving spatial regularization, a temporal regularization constrains the blind deconvolution problem, improving its effectiveness and its robustness. We demonstrate these improvements by processing various real video sequences obtained by different imaging techniques.

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Ferréol Soulez, Éric Thiébaut, Yves Tourneur, Alain Gressard, Raphaël Dauphin. BLIND DECONVOLUTION OF VIDEO SEQUENCES. 15th International Conférence on Image Processing, Oct 2008, San Diego, United States. pp.0. ⟨ujm-00293656⟩