Blur reduction by a Multiplicative Multiresolution Decomposition MMD

Abstract

In this paper, a new blur reduction method based on the new concept of Multiplicative Multiresolution Decomposition is presented. This method turns on specific corrections of coefficients in MMD domain. To appreciate the proposed approach performances, one could consider the classical method based on Shock filters and blind deblurring. The conventional methods for assessment like PSNR and MSE are invalid since there is no correlation with human perception. Therefore one could propose to use metrics which quantify the quantity of blur in images as it is proposed in [1]. In addition, one could compare the results visually and by plotting an original and deblurred transitions.