Impulse noise filtering by using an adaptive single-linking pulse coupled neural network

This study describes a novel method, called single-linking pulse coupled neural network (PCNN), for Altering extreme impulse noise in a image. The proposed single-linking PCNN simplifies conventional PCNN and thus the related parameter can be adaptively selected and no iteration time needs to be determined, which a noisy image can be filtered by two times of firing process of the original image and the reversed image. The single linking PCNN first identifies noisy pixels and filters the noisy pixels by a median filter therefore the proposed method can filter impulse noise while keeping the fine information-bearing details. The proposed method can adaptively determine the filtering times based on the noise intensity. The method demonstrates better performance compared to conventional impulse noise filters when the noise intensity varies from 10%-60%. Experimental results on visual illustration and subjective indices show the effectiveness of the proposed method.