A Parallel Fuzzy Method to Reduce Mixed Gaussian-Impulsive Noise in CT Medical Images

Abstract

A fuzzy logic filter to remove mixed Gaussian and impulsive noise from CT medical images is presented. The method performs a weighted average operation where a fuzzy rule-based model is defined to compute the coefficients. In addition, a parallel filter based on this method is presented. Implementation of the parallel algorithm on multi-core platform using OpenMP is presented. Efficiency is measured in terms of execution time and in terms of PSNR over medical CT images. Experimental results show that mixed Gaussian-impulsive noise is reduced efficiently and image details are preserved. The multi-core implementation allows reducing mixed Gaussian-impulse noise in real-time.