Fuzzy Image Segmentation Technology Research

Image segmentation is the first problem of image analysis and pattern recognition, is also one of the classic problems of image processing, it is an important part of the image analysis and pattern recognition system, and decide the final image analysis and pattern recognition of quality discrimination as a result, it divided the image into different typical areas and take out the target of interest. In recent years, many scholars have proposed many image segmentation methods, such as threshold segmentation, edge, region growing method, fuzzy clustering method and so on. Fuzzy c-means clustering algorithm as the most widely used image segmentation algorithms have been successfully applied to many fields in the society, such as atmospheric science, geography, medical image, target recognition and so on.Firstly, this paper proposes a FCM image segmentation method based on minimum opponent inhibition. The traditional FCM algorithm to ignore data affects the clustering center. The algorithm proposed inhibitory factor and suppress opponents, weakened the initial data of clustering center, can avoid the clustering problems of slow convergence speed. On the other hand, weaken the negative effect of times of maximum membership degree, improve the positive membership degree, and accelerate the convergence speed of the algorithm, the experimental results show that the algorithm segmentation more reasonable.Secondly, in order to improve the efficiency of image segmentation, this paper proposes a segmentation based on2d histogram weighted FCM clustering algorithm. The method by constructing a reasonable first2d histogram of noise suppression; Then clustering sample set to shrink by decomposition; Finally using weighted FCM clustering algorithm for classification. The simulation results show that the efficiency of the method is superior to the standard FCM algorithm.Finally, the paper is based on SAR image grayscale distribution model, this paper proposes a fuzzy domain image segmentation method combined with spatial information. This method not only can achieve the segmentation accuracy of pixel level, and better suppresses the effects of coherent spot noise, and the processing speed is better than that of MRF method.