Design modeling and certain investigations of median filter

Abstract

This paper proposes a detection and an estimation of pixels corrupted by impulse noise. A simple and efficient impulse detector known as Min-Max impulse detector (MMID) is utilized for impulse detection. The detector classifies the pixels into noisy and noise free based on the dynamic range of an image. A pixel is identified as noisy if its intensity value coincides with the minimum or maximum value of the dynamic range. The noisy pixels are restored using the median of noise free pixels and noise free pixels are retained as it is. The experimental analysis declares that MMID is good at identifying the pixels corrupted by fixed valued impulses and enhances the performance of a median filter.