On a family of nonlinear cell-average multiresolution schemes for image processing: An experimental study

This paper is devoted to a new family of nonlinear cell-average multiresolution schemes and its applications to image processing. The algorithms are based on nonlinear reconstruction operators with several desirable features: the reconstructions are third-order accurate in smooth regions, the data used is always centered with optimal support and they are adapted to the presence of discontinuities.The goal is to obtain similar properties as linear multiresolution schemes but avoiding the classical Gibbs phenomenon of this type of reconstructions. Applications to image compression and denoising will be presented.