International Journal of Scientific & Engineering Research Volume 3, Issue 4, April-2012 1 ISSN 2229-5518 IJSER © 2012 http://www.ijser.org An approach for image noise identification using minimum distance classifier Raina , Shamik Tiwari ,Deepa Kumari ,Deepika Gupta Abstract :- This paper deals with the problem of identifying the nature of noise in order to apply the most appropriate algorithm for de-noising. The key idea involves isolation of some representative noise samples and extraction of their features for noise identification. The isolation of the noise samples is achieved through application of filters. Statistical features are extracted and the minimum distance classifier is applied for identification of the noise type present. Keywords :- Noise, noise identification, statistical features, minimum distance classifier. ——————————  —————————— 1. INTRODUCTION The identification of the nature of the noise affecting an image is an important stage in all information interpretation systems by vision when the nature of the degradation is unknown. The majority of filtering algorithms (Lee, Kuan, ...) [l] [2] and certain algorithms of contour detection (Canny, Deriche, ...) [3] [4] found in literature, assume that the nature of the noise and its statistical parameters are known. Whereas in most practical cases we have not a priori knowledge on these data [5]. For this reason the statistical parameters of the noise must be estimated as they condition the quality of the filtering or the analysis of the images [6].In [7], we proved that it is possible to identify the nature of the noise by recording variations of local statistics (the standard deviation as a function of the average) computed in the homogeneous regions of the observed image. If the recording is parallel to the average axis, then the noise is declared as an additive one and its standard deviation is equal to the sampling average of the different values of the local standard deviation. If the recording can be assimilated by a line passing through zero, then the noise is declared as a multiplicative one and its standard deviation is given by the slope of the line. And finally, if the recording can not be viewed as a line passing through zero, then the noise is declared as an impulsive one. The previous methods presented in [7] [8] [9] are based on the criterion of maximum likelihood for the selection of the most homogeneous masks (Lee, Nagao etc.), from which the value of the local standard deviations are calculated. However, the disadvantage with this approach is the estimation of parameters from pixels belonging to masks a priori decked. This means that the estimates of standard deviations are sometimes necessarily biased and the final identification rates inevitably decreased in the case of images degraded either by a weak multiplicative or an impulsive noise. The search for efficient image de-noising methods is still a valid challenge at the crossing of functional analysis and statistics. In spite of the sophistication of the recently proposed methods, most algorithms have not yet attained a desirable level of applicability. All show an outstanding performance when the image model corresponds to the algorithm assumptions but fail in general and create artifacts or remove image fine structures In order to increase the rate of identification and to improve the estimation of statistical noise parameters, we propose a new method. The statistical parameters kurtosis and skewness are calculated and the Minimum Distance Classifier is applied. Classification includes a broad range of decision-theoretic approaches to the identification of images. All classification algorithms are based on the assumption that the image in question depicts one or more feature and that each of these features belongs to one of several distinct and exclusive classes. Classification analyzes the numerical properties of various image features and organizes data into categories. 2. THE PROPOSED METHOD In principle, the noise identification method proposed here consists of three key steps: Step 1. Extract some representative noise samples from the given noisy image, Step 2. Estimate some of their statistical features, and Step 3. Use a simple pattern classifier to identify the type of noise. In this paper, we consider four different types of commonly occurring image noise, namely,