An iterative adaptive fuzzy filter using alpha trimmed mean elimination of high density salt and pepper noise in images
V. Sujith, B. Karthik · AIP conference proceedings · 2022
This work deals with the removal of pepper and salt noise in two stages (detection and denoising).The detection stage uses an adaptive fuzzy detector. This is trailed by the denoising stage using slanted mean filter. The key steps include the initialization of parameters namely the window half size parameter (M) and number of good pixels (N).Once initialized Gaussian Membership function is applied. Then if the value obtained so is less than the threshold it is assumed to be noisy. The cardinality of the good pixels around the noisy pixel is found. If the cardinality so obtained is less than the previously initialized parameter N, then relaxation is done on the parameters namely chosen window size and then the initially set N good pixel demand. The denoising uses weighting function based on the inverse distance weighting. Here lesser heaviness is allocated to the pixel far away from the center pixel and higher weight for the pixelearlier to the centerpixel