Measuring isotropic local contrast: a circular mask based approach

Abstract

Many image processing tasks depend on contrast measures, which can be used to compare and improve contrast enhancement algorithms. Contrast definitions are not always suitable for all situations. In particular an isotropic local contrast measure, that produces a flat response to sinusoidal gratings, can be difficult to obtain. In this paper we review the main existing contrast measures, and propose a new approach, denoted as Circular Mask Metric (CMM). It is based on band-pass filters and circular mask based local contrast computations. This approach has been applied on different test images and with three contrast enhancement methods, in order to show its potentialities for contrast enhancement algorithm testing and improving.