Image filtering based on discrete cosine transform

Abstract

Quite often acquired images are noisy and filtering is a standard operation to improve image quality. "Good" filters should preserve details, edges and texture features and, at the same time, suppress noise. Recently it has been demonstrated that transform based filters are able to satisfy these requirements well enough and much attention has been paid to wavelet based filter design. In this paper it is shown that another kind of transform based filters, namely, discrete cosine transform (DCT) filters is an effective tool that competes with the state of the art methods. An additive noise case is basically considered although quite easily the DCT based filter can be modified for coping with other types of noise. © 2007 Begell House, Inc.