An Effective Modified Median Filter Based on Neuman’s Cellular Automata for Low SPN Attack Detection and Restoration in Medical Images

Abstract

Medical images such as brain MRI (Magnetic Resonance Imaging) scans are prone to various noise attacks. The presence of noise perplexes the identification, which may lead to inaccurate analysis and diagnosis. Under the framework of Median Filter (MF), its variant Switching Median Filter (SMF) and Cellular Automata (CA), we proposed effective Salt & Pepper Noise (SPN) filters. They work similar to those of MF and SMF except the use of Neumann Neighborhood (NN) and CA framework. Filters 1 and 2 processes all pixels in the noisy image uniformly whereas filters 3 and 4 first performs noise check and then restores the noisy pixels. Peak Signal to Noise Ratio (PSNR) performance parameter were used to test the proposed filters and the standard traditional filters. Experimental results reveal that the performance of the proposed filters is better. The proposed filters are able to restore low SPN effectively while preserving image details. The purpose of the proposed study is to present CA-based MF’s so that CA-based MF’s could also be explored like that of MF discussed in general image processing domain. Further, proposed filters proved computationally efficient as they required to compute lesser number of steps than standard MF and SMF by exploiting NN.