Image contrast enhancement using a recursive rational filter

Abstract

Digital cameras, new generation phones and in general modern image acquisition devices, requires algorithms for image enhancement suitable to work in real time and possibly with a limited power consumption. Unfortunately most of the systems present in the literature either require a very high computational effort or do not provide good performances. In this paper a new method for the improvement of the visual quality of a digital image is introduced. The exposed method, substantially based on the "retinex" approach, adopts an innovative system for the extraction of the illumination component, which is essentially the bottleneck for real-time applications based on the retinex method. In the retinex theory, the illumination of an image is approximated by a narrow-band low-pass filtering of the original image which should maintain the abrupt transitions unchanged. In order to obtain this behavior a new typology of filters is introduced, i.e. the recursive rational filter (RRF), The adoption of such filters concurs to obtain the proper behavior using a very simple structure, suitable for an implementation in real-time devices