ABSTRACT
The purpose of the talk is going more deeply into (1) as a discrete model for image processing by itself, paying special attention to the influence of the nonlocal character in the implementation of several problems in image and the comparisonwith other discrete but local alternatives, see e. g. [13,14]. A first part of the talk will analyze general theoretical properties of (1). The study will focus on scale-space properties, [1,2], and the relation with semidiscrete models. In this latter case, models based on (3) will be emphasized. The second part will show several results of comparison between (1) and discrete, local techniques presented in the literature, in differentproblems of image restoration and classification.
