Color image denoising by means of three-dimensional discrete fuzzy numbers

The quality of color image is degraded by the random value impulse noise, which not only affects the overall visual effect of color image, but also causes errors in the application of color image, such as image recognition and image understanding. In this article, a new color image denoising method that uses the three-dimensional discrete fuzzy numbers (3DFN) is presented. Firstly, the definition of three-dimensional discrete fuzzy numbers is introduced based on a special non-empty finite subset of three-dimensional real number space. Secondly, the 3DFN is employed to interpret each pixel value of a color image. In addition, the mass centers and the ambiguity degree are defined to describe the properties of 3DFN. Then, these properties are selected to determine corrupted pixels and replace the noisy pixels; meanwhile, the non-noisy pixel is not changed. At last, some experiments are included to demonstrate that this algorithm can effectively remove the random value impulse noise and preserve the texture and edge of the color images. At the same time, the algorithm also brings forward a new idea in the application of discrete fuzzy numbers.