Fuzzy logic approach in salt and pepper noise

Abstract

This paper describes a new method based on fuzzy logic to denoising images corrupted by the salt and pepper noise. In this approach, noisy pixels are detected via fuzzy IF-THEN rules and restored with a multi-level fuzzy filter. To reduce the amount of noise corruption, the filtering process begins with a window of size 3 × 3 centered on the noisy pixels. Then, a floating window in the form of fuzzy compositions is used for the remaining unrestored noisy pixels. A fuzzy logic controller is employed to improve the restoration results. By comparing this method with many well-established denoising methods, we have shown that the mentioned technique has high accuracy and reasonably low complexity at very high noise densities while preserving the structure and edge information of the images.