Research On Robust Segmentation Algorithm Baded On Picture Fuzzy Clustering

Image segmentation is an important basis for image processing and analysis,and has been widely used in the field of pattern recognition and machine vision.The essence of image segmentation is to classify different targets in the image according to the requirements,which is used to solve the classification problem of pixels with similar characteristics,so that similar pixels are more likely to be classified into one class.Because of the loss of information in the imaging process,there is uncertainty about the cognition of the adjacent grayscale level in the human visual system,which leads to the fuzziness of grayscale,texture and edge.Fuzzy clustering can deal with such uncertain information of images well.Therefore,many scholars at home and abroad have put forward a series of fuzzy clustering segmentation algorithms,which has an important influence on image segmentation research.Fuzzy C-mean Clustering(FCM)is used as the basis of fuzzy clustering algorithm,and its sample clustering membership degree can only express the certainty that the sample belongs to a certain class,but it can not express the uncertainty of the sample in the clustering process.By introducing the picture fuzzy set into fuzzy clustering,a kind of picture fuzzy clustering segmentation algorithm is proposed to solve the clustering problem of the algorithm sample,which not only considers the definite information that the sample belongs to a certain class,but also considers the negative information that the sample does not belong to a certain class,as well as the unknown information when the sample is classified,It has aroused the wide research and application of many scholars.This paper mainly improves the existing graph fuzzy clustering segmentation algorithm,and introduces the neighborhood information of sample pixels,Hilbert regenerative kernel space and fuzzy local information into the objective function,and proposes a series of new algorithms.The main research contents are as follows:1.Because of the problem of improper initialization in the clustering process,aming at this shortcoming,based on the picture fuzzy clustering algorithm,this paper constructs the idempotent expression into the objective function,the three parameters of membership degree,neutrality,and rejection degree in the picture fuzzy clustering algorithm are regularized,which ensures that the algorithm can converge in the finite iteration number,and the neighborhood mean pixel grayscale information is embedded in the improved picture fuzzy clustering objective function,and a high performance robust picture fuzzy clustering segmentation algorithm is proposed to further enhance the validity of the algorithm clustering and the robustness to the noise.2.Aiming at the problem that robust picture fuzzy clustering segmentation algorithm is difficult to effectively cluster non-convex irregular data,to solve this problem,the kernel function is introduced into the objective function of the algorithm to construct a robust fuzzyclustering based on kernel space.The segmentation algorithm uses the new linear weighting and image instead of the original image,which has a good effect on reducing the influence of noise on the image,In order to further enhance the robustness and adaptability of the algorithm,the neighborhood pixel variance of the current pixel is used as the adaptive factor,and an adaptive robust kernel space picture fuzzy clustering segmentation algorithm is proposed in order to automatically adjust the ability of the algorithm to suppress the noise and weakness.3.Aiming at the picture fuzzy clustering segmentation being not suitable to segment high noised image with large size,a fast robust kernel space picture fuzzy clustering segmentation algorithm is proposed in this paper.Firstly,the samples in European Space is mapped to the high dimensional feature space through the kernel function.Then the linear weighted filtering image is obtained by combining the current pixel with its neighborhood pixels through the space information in the segmenting image.In the end,the two-dimensional histogram between the clustered pixels and its neighborhood mean is introduced into the robust kernel space picture fuzzy clustering,In this way,the real-time performance of the algorithm is improved.4.In order to enhance the robustness and effectiveness of picture fuzzy clustering for image segmentation with noise interference,a picture fuzzy clustering segmentation algorithm based on local information is proposed in this paper.The algorithm embeds the improved fuzzy local information factor to affect the clustering objective function value,That the spatial distance information with membership degree and the regular factor of the control neighborhood pixel cluster compactness information introduced can better meet the anti-noise requirements of medical image segmentation,and better preserve the edge and detail information of the image.