Cleaning images from impulse noise in a binary symmetrical channel

Abstract

The paper proposes a new method for cleaning uncoded images from impulse noise in a binary symmetric channel, where, when an error occurs, the information bit is distorted and the image pixels take incorrect values. The characteristic of such noise cnuorresponds to impulse noise, where impulse noise takes on random values and is randomly distributed over the image. Pixels are determined to be distorted by evaluating the difference between pixels within the local window. This estimate takes into account the brightness value and the distance of pixels within the local window. Image recovery is performed using adaptive median filtering.