Pyramidal perceptual filtering using Moon and Spencer contrast

Abstract

A perceptual multiresolution filtering method (Moon and Spencer, 1945) based on human perception models is proposed. The main idea is to detect and remove the irrelevant structures at different scales using the just-noticeable contrast notion and luminance adaptation. The processing is done in the Laplacian and Gaussian pyramid decompositions of the image. The consistency of the method is demonstrated on a gray-level image.