To the best of our knowledge, there are no variational models devoted to the problem of removing noise from images corrupted by mixed AN and MN. The difficulties are not only to design an effective image restoration model, but also to develop an efficient numerical solution for the associated large system of nonlinear equations arising from high-resolution images. The aim of this project is to cover both aspects by designing a new TV-based image restoration model and developing a robust nonlinear multigrid (NMG) method.
We note that this paper is in the area of variational image restoration, while our previous works [25,26] are in the area of variational image registration. Although they share similar backgrounds (e.g. variational models and MG methods), they are totally different by their topics, models, equations and numerical algorithms. In [26], the new and existing higher order variational image registration models were discussed for two extreme cases of smooth and non-smooth deformation problems. In [26], our main contribution was to highlight the fact that an approximate Jacobian idea (in Taylor’s expansion) for all nonlinear terms in a diffusion registration model provides a better MG smoother than linearization of only one term which was the standard way of designing an MG method at that time.