Gaussian Noise Removal of Image on the Local Feature

Abstract

The traditional removing algorithm of Gaussian noise can only reduce the effect of noise rather than remove it. Furthermore, the noise points in the image will diffuse after removing. According to the effect of the Gaussian noise on the visual images, this paper introduces an algorithm based on the local feature of the image to eliminate Gaussian noise, and this method overcomes the defects of traditional methods. Firstly, we categorize the location of the pixels into three classes-on the noise point, on the edge, and in the local texture, based on the local continuous smoothing in the image. Secondly, we can extract the edge information and texture of the image by morphology according to the local continuity of the image edge and texture property, then we can accurately locate the noise points of the image. Lastly, we use adaptive neighborhood to eliminate the other noise points. Comparing to the traditional methods, this algorithm can remove the noise better and have satisfying image visual impression.