Combined de-noising and sharpening of color images in DCT domain

Abstract

In this paper we present a new DCT-based method for combined de-noising and sharpening of color images. Our solution combines alpha-rooting and thresholding the DCT coefficients to achieve both sharpening and noise reduction. In our approach, sharpening and de-noising are done in the YUV color space with Y, U and V components being processed differently. Moreover, the size of the sliding DCT transform is variable and adaptive to the local characteristics of the input image thus increasing the visual quality of the processed image. We describe our method in detail and we present experimental results, performed on color images, to compare its performance with other existing solutions.