Research Of Image Segmentation Algorithms Based On FCM Clustering

Image segmentation is the process of dividing an image into several meaningful areas.It is one of the key technologies in the field of image processing.The fuzzy Cmeans clustering algorithm(FCM algorithm)is an image segmentation algorithm based on fuzzy clustering.It combines fuzzy theory and image segmentation organically,which is more suitable for describing the uncertain information in images,and so that it have been widely concerned by scholars.In this paper,the image segmentation algorithm based on FCM clustering is deeply studied.Firstly,the basic theory and main methods of image segmentation are introduced.Then,the fundamental theory and research status of fuzzy C-means clustering algorithm are analyzed in detail.On the basis of this,the image segmentation algorithm based on FCM clustering is researched deeply in view of some main problems existing in the current research.Firstly,in order to improve the segmentation effect and stability of the existing fuzzy C-means clustering algorithm for noise image,a novel image segmentation algorithm based on FCM clustering is proposed.Firstly,the algorithm construct a sum image by using non-local spatial information.Then,the initial clustering center is automatically selected according to the histogram of the image.Finally,the image segmentation is completed by finding the minimum value of the objective function.The theoretical analysis and experimental results show that the proposed algorithm has better segmentation performance,the segmentation result is stable and the robustness to the noise is stronger.It successfully solved the problem that existing FCM algorithm can not effectively detect the small class in the histogram.Secondly,an improved FCM algorithm based on region merging is proposed to correct the defect that existing FCM algorithm based on the global characteristics to segment the image,prone to over-segmentation.Firstly,the partitioning result of FCM algorithm is taken as the initial segmentation.Secondly,the spatial relation of each initial segmentation region is detected and the adjacency matrix is established.Then,according to the comprehensive distance which combines the color and edge information of each region,adjacency matrix with weight is obtained.Finally,the area of the initial segmentation is merged according to the relationship between areas,and the final segmentation result is obtained.Experiments show that the improved algorithm can effectively suppress the over-segmentation of existing FCM algorithms.segmentation performance has a not small improved.