Rank-wise Weighted Mean Filter for Gaussian Noise Removal

This paper presents a rank-wise weighted mean filter for restoring an image corrupted by Gaussian noise. The method is two stages: ranking of image pixels and noise removal. First, the rank of each pixel is found using a threshold value. Finally, the weighted mean is calculated for each pixel using the rank of the surrounding pixels as the weights. Experiments and comparisons involving different sample images with varying noise variances show that our method yields better results than the other state of art methods, even at high standard deviation of Gaussian noise.