Image denoising using FREBAS multiresolution image analysis

This paper describes a 2-step image denoising technique based on the FREBAS transform that we have proposed as a new multiresolution image analysis. Noise removal is performed by applying a Wiener filter function, which is designed using a FREBAS prefiltered image, in the FREBAS transformed domain. Comparisons of image SNR improvement with previous single FREBAS denoising method and other methods such as nonlinear digital filter and Wiener filtering in the Wavelet domain were made. It was demonstrated that the proposed method provides the highest SNR improvement and the resultant image was free from serious image degradation.