Novel Approach based on Fuzzy Theory and Order Statistics in 2D/3D Processing of Images/Videos Employing Specific Hardware

An approach in filtering of color videos/images that are corrupted by impulsive or additive noises is justified. The difference of current approach in comparison with existing state-of-the-art techniques consists in use of fuzzy rules selecting high similarity pixels in the vicinity of the central one using existing correlation of different channels for an image in the RGB space as well as correlation in the neighboring frames of a video. Employing this correlation permits better preservation of the features, processing separately the areas with different texture. Numerous simulation experiments have confirmed the superiority of the designed techniques via objective criteria (PSNR, MAE, NCD, SSIM) as well as demonstrating better visual perception quality.