DCT Local Adaptive Filtering of Images Corrupted by Fluctuative Noise with a Priori Unknown Statistical Properties

Abstract

A DCT based locally adaptive filter has been proposed and applied for processing of real and imaginary components of bispectrum for solving a problem of unknown signal shape (waveform) reconstruction. The bispectrum is a 2-D complex function and in its real and imaginary components a strongly non-stationary noise is present. Thus, the filter designed can be also effectively applied to processing different kind of images for which non-stationary noise with a priori unknown local statistical properties is present. The modifications of DCT based filters have been successfully used for removal of pure multiplicative, speckle and film-grain noise.