Research On Image Segmentation Algorithm Based On Spatial Information Adaptive Fuzzy Clusterin

As the first step in image analysis,image segmentation is the basis of computer vision in image processing and attracts much attention.For traditional fuzzy clustering algorithm,Traditional fuzzy clustering algorithm uses spatial information to resist noise and reduce the influence of noise on image segmentation.However,as the complexity of the spatial structure of the image,how to choose spatial information is still a challenging problem.In order to solve this problem,a new improved algorithm is proposed,Fuzzy C-means with Adaptive Spatial Information Method(FCM-AS for short).The new model aims to integrate the spatial information of different images,and use local space and non-local space information as examples to create the model.The non-local information is used to improve the smoothness of the image region to obtain anti-noise.Meanwhile,the retained local information can also help to achieve the division of image block edge details.An adaptive weight coefficient is set between local information and non-local information to realize the adaptive adjustment of spatial information.In addition,we propose an iterative updating algorithm for this model,which makes the new model achieve better accuracy of image segmentation and robustness against noise.In this paper,two types of image databases,synthetic and real MR brain medical images,were used to compare with FCM,FCM＿S,K-means and FCM-NLS image segmentation algorithms based on fuzzy clustering.some indexes are used to evaluate the image segmentation.We can see from results: compared with the traditional fuzzy clustering method,the new model has higher segmentation accuracy when processing composite images.In medical image part,the result is closed to the original image and performs well under the index without correct labels,it has better robustness to defeat noise.