| With the vigorous development and progress of information and technology,people obtain information in a variety of ways.Image information is one of the most important forms for human beings to understand the world and receive information.In order to further process the acquired image information and meet different needs,image segmentation technology is needed.Image segmentation is the main problem in the field of computer vision.It is very important for post-processing and analysis of image data,and is of great significance in the field of image processing.Fuzzy C-Means Clustering Algorithm(FCM)is a typical fuzzy clustering algorithm,which is developed from K-means algorithm and widely used.Yanhui Guo et al.combined the concepts of Neutrosophic Theory and proposed Neutrosophic C-means Clustering Algorithm(NCM)to overcome the shortcomings of FCM algorithm.This paper mainly studies the Fuzzy C-Mean Clustering(FCM)and the Neutrosophic C-means Clustering algorithm(NCM),and optimizes and improves the shortcomings of the above algorithms.Firstly,the method needs to select the number of clusters manually,which is easy to fall into local minimum and be easily disturbed by noise.In this paper,wavelet transform,particle swarm optimization,anisotropic filtering and validity function are introduced into the FCM algorithm,and an adaptive FCM method based on wavelet transform and particle swarm optimization(IWPSOFCM)is proposed.The initial denoising image is obtained by wavelet decomposition and anisotropic filtering;the optimal solution is found by particle swarm optimization to improve the convergence rate;the number of clusters is judged by the validity function to achieve adaptive segmentation.The comparison experiment proves that the algorithm has better segmentation effect on noise image.Secondly,combining the IWPSOFCM algorithm mentioned above with the FGFCM algorithm,an IWPSOFGFCM algorithm is proposed,which improves the robustness of the algorithm by introducing spatial neighborhood information.The comparison experiments show that the algorithm has stronger robustness to noise.Finally,in view of the shortcomings and limitations of NCM algorithm,this paper introduces the concepts of fuzzy local information,noise distance and hierarchical clustering in NCM algorithm,so that the algorithm does not need to select parameters artificially and has a certain robustness to noise.Through a series of contrast experiments,it is proved that the algorithm has better segmentation results for noisy images. |
