A Quartile Based Novel Approach For The Reduction Of Salt And Pepper Noise

Abstract

In the modern era, people prefer short and informative data like images over the large texts. In text messages, we are focusing on security of data in a similar way in images along with the security of data the noise in an image is a big issue. Image noise is basically a pixel affected by some false value which may be the cause of misleading the result especially in an object and edge detection in an image. Hence for performing the operation of an object and edge detection in the noisy environment some pre-processing must be required. Noise reduction from an image is active research problem in Digital Image Processing. Gaussian noise, Speckle noise, Poisson noise, Salt and Pepper noise are the most common types of noises in an image. In this paper the authors proposed a quartile based novel noise reduction approach for salt and pepper noise which perform better than the median filter in noisy images having a high density of salt and pepper noise. For experimental result analysis, a qualitative and quantitative comparison is done with the traditionally widely accepted approach for salt and pepper noise i.e median filter, which is considered best for the salt and pepper noise. All experimental analysis is done with the randomly selected images from the MSRA dat-aset (Salient Object images data-set). Noisy images are prepared by mixing the artificial noise using MATLAB with different level of noise density.