ADAPTIVE SELECTIVITY REPRESENTATION: DESIGN AND APPLICATIONS

Abstract

This paper focuses on the development of a new two-dimensional representation for images that can capture different features in images, ranging from highly directional ones to fully isotropic ones. We propose a multiselectivity analysis, defined by combining an isotropic multiscale and multidirection decomposition. The result is new half-continuous frame for each selectivity level. The angular selectivity of these frames grows with selectivity level. This selectivity level can be adapted locally to the content of the image; so it can be seen as an adaptive selectivity representation, which present adaptively isotropic, directional and intermediary features in images. The numerical experiments presented in this paper demonstrate that the adaptive selectivity approach is very competitive in image denoising and enhancement.