MBC-CA: Multithreshold binary conversion based salt-and-pepper noise removal using cellular automata

Abstract

The salt-and-pepper noise, one of the forms of impulse noise, is one of the important problems that needs to be taken care of. Salt-and-pepper noise in the images are introduced during their acquisition, recording and transmitting. Cellular Automata (CA) is an emerging concept in the field of image processing due to its neighborhood dependence. Various methods have been proposed using CA for noise removal, simply due to the high complexity of CA, most of them are proven to be inefficient. However, CA can be used efficiently with some modifications that result in a reduction in its complexity, for the large number of image processing techniques. In this paper, we overcome the problem of CA by Multithreshold Binary Conversion (MBC) in which we convert the grayscale images to binary images based upon a chosen set of threshold values, reducing the state from 256 to 2 for every pixel. The resulting images are then fed to the CA. The result obtained is a set of binary images and these binary images need to be recombined to obtain a noise free grayscale image. We have used a method similar to a binary search that reduce the complexity of recombining the images from N2K to N2logK making our recombination algorithm an efficient algorithm, in terms of complexity, to recombine binary images to a single grayscale image. This reduction in the complexity of noise removal has no effect on the quality of a grayscale image. © Springer Nature Singapore Pte Ltd 2020.