A New Optimum Method of Mean Difference Fuzzy Filter for Noise Reduction in Color Images

Abstract

Image Reduction is one of the most important image processing tasks. In this article, a new Mean Difference fuzzy filter (MDFF) is presented for the removal of both Gaussian noise & impulse noise in color images. This new filter consists of two sub filters which are effectively employed to denoise both the Gaussian noise & impulse noise. This new fuzzy filter MDFF is applied only for the faulty pixel, without considering other original good pixels for which two thresholds are used to decide whether the pixel is corrupted or not. The first sub filter detects the noisy fixel along with the amount of noise with the help of membership functions using fuzzy logic. Fuzzy rules are constructed for fuzzy filters to realize the amount of noise [1]. Then the corrupted fixels are rectified with the help of noise free pixels. The second sub filter removes the noises in color images by using the relation between the color components of all corrupted pixels. This proposed fuzzy filter is tested with standard image under different noisy conditions using 3x3 windows. The proposed filter shows the better performance over all other existing standard some denoising filters in color images.