Medical Image Noise Suppression Using Mediated Morphology

摘要

This paper introduces superior ability of mediated morphology in suppression of normal noise and its great unique ability in suppression of high density Gaussian and Salt & Pepper noises from medical images. The proposed morphological filter is based on mediated morphology, that its main operators are composed of a particular internal combination of Classical morphological operators and median operator. The mediated Morphological algorithm has been compared with classical Morphological filter, Weighted Morphological filter and linear averaging filter by applying over MRI images contaminated with three different types of noise; Salt & pepper, Speckle and Gaussian. Accurate Numerical descriptions convey that the proposed filter dominate all others, in different levels of noise. This dominance becomes more evident in high levels of noise. While due to increase of the noise level, the other filters gradually become useless; Mediated Morphological filter has still its effect. Furthermore as another advantage, it is not almost influenced by density of Salt & Pepper noise as well as variance of Gaussian noise.