Solving a New Class of Variational Models for Image Decomposition via Projection

In this paper, we propose a new class of variational models for image decomposition into structure and texture or noise, which is based on Besov spaces. They can be seen as generalizations of Daubechies-Teschke's work. And we, inspired by Lorenz, give proof for the general characterization of the solution of these models based on the orthogonal projections onto the convex set, as well as some material examples of the proposed models. Finally, we present numerical examples on denoising and decompositions of images