Research And Application Of Image Segmentation Algorithm Based On Fuzzy Clustering

Image segmentation is an important research topic in the field of image processing and image analysis. In recent years, domestic and foreign scholars have carried on a great deal of research, but there is no general segmentation algorithm. Because the insensitive of human vision to image gray information, and the image itself is ambiguous and inhomogeneity, it is difficult for the traditional segmentation methods to obtain ideal segmentation results. Clustering algorithm based on fuzzy theory is suitable to deal with the fuzzy of image, it can more objectively reflect the reality of generic relationship. Fuzzy C means clustering algorithm as the most classical form of fuzzy clustering algorithm, it has been widely used in the field of image segmentation. This paper introduces the methods of image segmentation, and emphatically conducts the deep research on the fuzzy C means algorithm, aiming at the shortcomings of fuzzy C means algorithm made some improvements. The main work of this paper are as follows:(1) This paper sums up the research situation of image segmentation algorithm. It describes the research status of fuzzy C means algorithm in image segmentation field. It discusses the fuzzy theory and fuzzy clustering algorithm.(2) Because the fuzzy C means algorithm is sensitive to noise, and easily affected by the initial clustering center, this paper proposes a fuzzy C means algorithm which combined with local information. Through adding the neighborhood constraints into the objective function, each pixel is influenced by its neighborhood pixels, the anti noise of the algorithm is improved. In order to improve the convergence speed and avoid the algorithm to local extremum, using the peak maximum density algorithm to select the initial cluster centers,. The segmentation results of synthetic and natural images demonstrate that the proposed algorithm, compared with other fuzzy C means algorithm has stronger anti noise ability and higher accuracy of segmentation.(3) Because the traditional fuzzy C means algorithm only consider the gray information of image, and ignore the spatial position information of pixel, and the algorithm can not classify different objects with similar gray value in the segmentation process, so this paper proposes a fuzzy C means algorithm which combined with minimum distance classifier. Firstly, the algorithm uses FCM algorithm to achieve the clustering according to the gray information of the image, then automatically extracts the clustering information of region of interest, and uses the minimum distance classifier to classify them according to their distance. By using this method to clinical trials, this paper achieve the classification of breast magnetic resonance imaging in the situation of different tissues with gray similarity. The experimental results show that this algorithm has higher accuracy in breast magnetic resonance image segmentation, and improves the accuracy of tissue classification.