Removal of Salt and Pepper noise: A Game Theoretic Approach

Abstract

Digital images are often corrupted by the salt and pepper noises which eventually deteriorate the quality of images and affect the subsequent processing steps. This paper presents a game theory based approach to remove the salt and pepper noises. A two-player, non-cooperative, non-zero sum 2-in-1 game is designed and implemented at the pixel level to extract features or local information about the image pixels. This game's payoff values are used to detect whether a pixel is noisy or not. Then the noisy pixel is replaced with a proper value, keeping in mind the pattern followed by the non-noisy pixels. Proposed method efficiently removes the noises of density up to 40%, which is commonly observed in digital images. We have applied our method on some standard images by incorporating varied amount of noises and it has been observed that the proposed method outperforms some state-of-the-art methods.