FUZZY BASED GAUSSIAN NOISE REDUCTION IN COLOR IMAGE

Abstract

An elementary problem of image restoration is to reduce the noise from an image as much as possible. The noise reduction procedures depends upon the type of noise degrading the image. In this work effort are made to reduce Gaussian noise. To do same efforts are made to develop fuzzy based filters for reduction of Gaussian noise in colour images. In the proposed method first we determine the level of Gaussian noise in the local window with the help of the standard deviation of local window in comparison to standard deviation of all local windows contained within image and mean of local window and its nearby windows. Then we compute adaptive distances between colour component pairs of the central pixel and the neighbourhood pixel and fuzzify these adaptive distances. These fuzzified adaptive distances used to determine the weights of everey color component of each neighbourhood pixels contained within local window. The weighted average of color components weights of all the neighbourhood pixels helps in computing the correction term for the Gaussian filter. The results on several color images prove the efficacy of the proposed fuzzy filters.