Denoising by Utilizing Mean Deviation of Image's Smooth Region

摘要

This paper presents a denoising method to preserve the image fine details and edges while effectively reducing the noise level. The proposed denoising is performed in a two-phase process. The image is restored by reducing noise to a certain level that will still preserve the image fine details and edges. Further denoising is performed only in the smooth regions to remove the remaining noise. The performance of the proposed denoising method is investigated by the visual effects and objective image quality evaluation. Simulation results demonstrate that the proposed method provides high performance for all evaluation criteria in comparison to the conventional approaches.