The usual Non-Local Total Variation (NLTV) term penalizes discrepancy between some specified pairs of pixels, a weight value is computed between these two pixels to measure their dissimilarity. In this presentation, we describe a stable scheme to regularize weight values in the NLTV model, allowing to restore them when they are difficult to define. We also show that the proposed model better recovers thin structures on inpainting, zooming and denoising problems.

This presentation is part of Minisymposium “MS78 - Recent developments in variational image modeling”
organized by: Sonia Tabti (Université de Caen, CNRS) , Rabin Julien (CNRS, Normandie Univ.) .

Authors:

Zhi Li (Department of Computational Mathematics, Science and Engineering (CMSE) Michigan State University )

Francois Malgouyres (Institut de Mathématiques de Toulouse, Université Paul Sabatier)

Tieyong Zeng (Department of Mathematics, The Chinese University of Hong Kong, Shatin, N.T.)

Keywords:

image restoration, non-local total variation, nonconvex minimization, proximal alternating linearized minimization