An Alternative Fuzzy Set Usage for Reducing Fixed Impulse Noise

Abstract

Image filtering is an important process frequently used in image processing applications in order to eliminate the noise of noisy images caused by disturbing effects. As digital images are often distorted by noise during shooting, and with the effects of other factors may in some cases make the original characteristics of the image unobservable. For this reason, noise reduction is an important step for the analysis of noisy images produced by digital camera flaws and during transmission of data. Noise reduction improves perception of images and usually results in better performance for different image analysis and computer vision algorithms. In this study, an alternative fuzzy set and 5x5, 3x5, 2x3 size filtering windows and mean - median filters were used to reduce salt-pepper noise. The performances of the proposed methods are then compared with some methods in the literature according to Peak Signal-to-Noise Ratio (PSNR) and Mean Square Error (MSE) on 10 test images.