STUDY ON WEIGHTED MEAN FILTERING ALGORITHM BASED ON THE IMAGES WITH SALT AND PEPPER NOISE

摘要

Digital image restoration is the automatic restoration of the original image to the most possible degree with the aid of the currently available information. The reduction of salt-pepper noise is considered a leading task in de-nosing process, which forms the essential part of pre-processing. In this paper, a method is proposed for the de-nosing of salt-pepper noise with an adaptive weighted mean filter based on the decomposition of histograms. Traditional ways of filtering the salt-pepper noise are introduced first, followed by the study of weighted mean filter based on decomposition of histograms and adaptive weighted mean filter. The advantages and disadvantages of both are further discussed for the proposal of an adaptive weighed mean filter based on decomposition of histograms, according to which the weighting coefficient is determined by an orthogonal decomposition of the original image histogram using the properties of Gaussian distribution. The improvement is proved to be simple and effective.