<title>Removal of mixed noise on color image processing by using fuzzy rules</title>

Abstract

Abstract We have proposed fuzzy filters in order to remove additive non-impulsive noise (e.g., Gussian noise)while preserving signal details. In this paper, we propose a novel fuzzy filter for removing mixed noise(i.e., Gaussian noise and impulse noise are mixed). Furthermore, we apply the proposed method to colorimage processing. In order to remove mixed noise efficiently, we set fuzzy rules by using multiple dii-ference values between arbitrary two pixels in a filter window. We show tuning result of the proposedfuzzy filter and present some simulation results. 1. Introduction Since the first introducing of Fuzzy Set Theory [1] fuzzy techniques for image processing applicationshave mainly detail with high-level computer vision and pattern recognition [2]. Only recently, however,fuzzy techniques have successfully entered the area of low-level computer vision for general purposeapplications becoming competitive with classical methods in some very important pre-processmg tasks.In particular, focusing on the area ofnonlinear filtering of noisy image data, many original approacheshave been proposed in the last few years[3-lOJ.Taguchi and Takashima propose a fuzzy filter which is one ofnonlinear filter for noisy image data, in[8]. This fuzzy filter is an adaptive weighted average filter whose weight derived by using fuzzy rules.The antecedents offuzzy rules are constructed by using two important local characteristics: the differ-ence value from centre pixel's value and the distance from centre pixel. Since the fuzzy filter is able tochange its property according to local characteristics, it can remove non-impulse noise (e.g., Gaussiannoise) while preserving details. However, the fuzzy filter isn't able to remove impulse noise, becausethe difference value from centre pixel's value which one of local characteristics, is very sensitive toimpulse noise.In this paper, we propose a novel fuzzy filter which is able to remove not only non-impulse noise butalso impulse noise, since we introduce a new local characteristics to the fuzzy filter. A new local char-acteristic is calculated by using multiple difference values between arbitrary pixels in the filter window.The proposed filter can be applied to color image processing. Thus, we also propose the fuzzy filter forcolor image processing.In Section 2, we introduce a new local characteristic for filtering. Furthermore we also show thesame concept local characteristic for color image processing. In Section 3, we propose a novel fuzzyfilter for monochrome and color images, which is called a modified fuzzy (MF) filter, by using these