Research And Application Of Image Segmentation Algorithm Based On Fuzzy C-Means

Image segmentation is an important part of image analysis,and it is also a difficult part of image processing technology.People increasingly introduce new methods and technology to the image segmentation,having achieved good results,while it still remained the problem of initialization of clusters number and it is sensitive to noise.Due to the advantages of fuzzy theory in image processing,more and more scholars have introduced the fuzzy theory into image processing,in which the method of fuzzy clustering has achieved good results in image segmentation.We studied the application of Fuzzy C-means(FCM)clustering algorithm in the field of image segmentation,and intended to solve the problems by improving the original method.On the one hand,in order to solve the problem of human intervention of initialization of clustering center and cluster number in FCM,this paper proposed an improved FCM algorithm based on information of image gray histogram to initialize cluster center.Through the information of gray distribution histogram,the algorithm divides the different peak regions into different clusters.Then we can determine the number of clusters,and get the value of the initial clustering center by computing the average gray of each part.This result is then used as the initial input of the FCM algorithm to complete the clustering segmentation.This method avoids the artificial participation which is necessary in the traditional FCM algorithm.Experiments show that the proposed algorithm can reduce the number of iterations and improve the efficiency of the algorithm compared with the original algorithm.On the other hand,the FCM image segmentation algorithm considers gray feature as the only reference feature,so it is very sensitive to the noise.On the basis of the original FCM algorithm,the character of the spatial position of the image was introduced to decrease the noise interference.We added a penalty term containing the spatial information to the original objective function,and adopted a method of averaging after eliminating outliers to filter the neighborhood,effectively reducing the noise effect on image segmentation.The experimental results show that the proposed algorithm can achieve good segmentation results for many types of images.