Research On Image Filtering Based On Fuzzy Set Theory

Image filtering is an important part of image processing field, which is the fundamental basis of image analysis such as image segmentation, extraction of texture and shape feature, etc. A large number of papers on this subject focus on image filtering based on the mathematical models. However, many of the filtering algorithms based on mathematical models can only be useful to specific kinds of noises. For example, median filtering is used to remove impulse noise, but mean filtering is used to Gaussian noise reduction. The focus of image filtering is the distinction between image noise and signal points, the process of noise detection is, in fact, a problem of uncertainty. Fuzzy logic has been shown to be very well suitable to address this uncertainty. Meanwhile, the parameters in some of the mathematical models are difficult to determine when little information about the image is known. Since some human knowledge can be expressed in terms of linguistic rules, the use of fuzzy logic is appropriate. The idea of applying fuzzy logic to experts systems, where If…Then rules represent the knowledge of human being, a fuzzy approach to image filtering is a very promising research area.The purpose of this paper is to explore new methods of image filtering, and propose a new image filtering method based on fuzzy set theory. The dissertation firstly elaborates the basic theory of fuzzy logic, and then introduces some classic image filtering algorithms. Based on this, the purpose of this paper is to research the correlation algorithm, which applies fuzzy logic to image filtering. The emphasis of this paper is as follows: Firstly, the noise detection algorithm based on the physical model is studied. The physical model, the improved spring-mass model is described in detail, and we use balance conditions of the plane system of forces as the standard, which is used to determine whether or not the pixel is noise point, at last, through the iterative method, only to modify the value of the pixel of the noise points using signal points. Secondly, in the base of the improved spring-mass model, combine this physical model and fuzzy sets theory and give full consideration to the color information of images, at last, new rules to detect image noises are established, and a new filtering method is proposed. The article will be not only common salt and pepper impulse noise for the study, as well as the random impulse noise with uniform distribution is considered. Finally, to every algorithm, there are both theory analysis and relative programs, which are designed by Matlab to verify the properties of these algorithms. A lot of experimental results show that the method proposed in this paper is effective, feasible and more super,especially in terms of the ability of removing noises and the ability of preserving the partial details of images in comparison with some recent methods when the noise level is higher.