Semi-local Total Variation for Regularization of Inverse Problems

Laurent Condat (1)

1 Equipe Image - Laboratoire GREYC - UMR6072

We propose the discrete semi-local total variation (SLTV) as a new regularization for inverse problems in imaging. We show that the corresponding optimization problems can be efficiently solved by a primal-dual algorithm, which is easy to implement. The SLTV favors piecewise linear images, so that the main drawback of the total variation, its clustering effect, is avoided.

Mots clés: total variation; non-local regularization; inverse problems; convex optimization; primal-dual splitting; proximal operator

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

Laurent Condat. Semi-local Total Variation for Regularization of Inverse Problems. 2011. ⟨hal-00608693v1⟩