Switching Median Filter Based on Iterative Clustering Noise Detection

Chingakham Neeta Devi, Keisham Pritamdas

Abstract: In this paper, a switching median filter incorporated with an iterative clustering based noise detection algorithm that effectively restores images corrupted with impulse noise is proposed to improve the performance of switching based median filters. The proposed filter mechanism involves a noise detection algorithm for image restoration that detects corrupted pixels by classifying the pixels into three pixel clusters iteratively. If the pixel in consideration falls under the middle cluster in the last iteration, the pixel is classified as uncorrupted otherwise it is detected as corrupted. Corrupted pixels go through median filtering leading to restoration of the noisy image effectively while uncorrupted pixels remain unperturbed. Experimental results show that the proposed filter provides appreciable quality restored images preserving important details of the image.

Keywords: Impulse noise, noise model, noise detection, image restoration, switching median filters