Adaptive rank weighted switching filter for impulsive noise removal in color images

In this paper, a novel approach to the problem of impulsive noise removal in color digital images is presented. The described switching filter is based on the rank weighted, cumulated pixel dissimilarity measures, which are used for the detection of image samples contaminated by impulsive noise process. The introduced adaptive design enables the filter to tune its parameters to the amount of impulsive noise corrupting the image. The comparison with existing denoising schemes shows that the new technique more efficiently removes the impulses introduced by the noise process, while better preserving image details. An important feature of the new filter is its low computational complexity, which allows for its application in real-time applications.