Eight Directional Advanced Multistage Median Filter

Abstract

ABSTRACT newline newline newline newline newlineUnwanted information that corrupts an image is called noise. During image acquisition impulse noise is often introduced. Before processing the image in subsequent steps, noise in the corrupted pixel must be removed. A lot of techniques are available for the removal of impulse noise. Among these filtering algorithms, median filtering is having a prominent role in the suppression of noise. Intensity of the noised and corrupted pixel is relatively high or low. The main types of noise which are available are salt and pepper noise and random valued noise. The gray scale value is either 1 or 255 for the impulse noise. This will give rise to the appearance of noise as white and black superimposed dots on the image and hence the term salt and pepper noise. For extracting reliable and accurate details from digital image noise removal is critical. newlineNoise removal from images is one of the very important thing to be considered in digital image processing as different types of noise occur while image capturing, transmission or storage. The main information of the image should be preserved in the time of noise removal. There are so many algorithms available for obtaining the correct image after removing the impulse noise from the corrupted image. newlineAn algorithm is proposed in this work for the removal of impulse noise from a highly corrupted gray scale image. The proposed algorithm is having two steps and in the first step determination of whether a particular pixel is corrupted or not is carried out. In the next step using the proposed algorithm recreation of the corrupted pixel is done. newlineA variety of algorithms for image filtering are available that can make the image distortion free. During the acquisition and transmission of digital images impulse noise may be introduced. So, the noise must be eliminated before the processing of the image further. There are many impulse noise removal algorithms developed. This thesis proposes eight directional advanced multistage median filtering algorithm for impul