Segmentation Based On Markov Random Field And Fuzzy Clustering Of Brain Images Ds

According to different purpose, the brain's MRI segmentation is mainly sorted: extracting brain organization from the brain's MRI; Brain organization classification, namely the brain's MRI is marked as three different brain organization regions:white matter (WM), gray matter (GM), cerebrospinal fluid (CSF). And extracting the parts of brain pathological changes.Because of brain image segmentation is relatively complicated and exquisite, and then signal method is difficult to obtain better segmentation effect, therefore on comprehensive considerations, this article adopt some different methods that are adopted advantages and eliminated disadvantages to segment image. This article main thought is as follows:(1)Markov random fields is favorable to considerate image spatial information, therefore in this article, it is considerable to apply in extracting brain nuclear organization and brain nuclear organization segmentation. First of all, when extracting brain nuclear organization, Markov random fields potential energy superiority of Markov restore technology effectively make use of spatial information, that makes images that are rebuilt become more smoothness and exquisite, this is good to extract brain nuclear organization. Secondly, in brain nuclear organization segmentation, Markov segmentation technology also adequately considerates spatial potential energy, so images segmentation become more exquisite, but if it is excessive, images will be not continuous, this is bad for diagnosis, therefore it need to combine Dempster-Shafer(D-S) theory.(2) Fuzzy C-means Algorithm can rapidly find the center of set, and segment, it has advantage of segmentation speed while disadvantage of quality, to be concrete, segmentation is not enough exquisite, at this moment, it need to combine Markov segmentation result and use Dempster-Shafer (D-S) theory to overcome. But Fuzzy C-means Algorithm in this article, the most important apply is that it rapidly and accurately finds the center of set for Markov and Dempster-Shafer (D-S) theory, then it uses different methods to find basic set classification for Markov and Dempster-Shafer (D-S) segmentation. (3)Dempster-Shafer (D-S) theory is more exquisite for images, therefore Dempster-Shafer(D-S) theory is the best close to humanity vision in all existing algorithm, but excessive morbidezza will lead to images spread, thus this can not gain assistant diagnosis' ends. The purpose of this article is making morbidezza in proper image regions, by contraries, giving entirety outline so as to medicine diagnosis. So the solution is that dealing with Markov and Fuzzy C-means Algorithm segmentation result, marking same classification between them, making differ classification too. The same classification marks are probably images outline, differ marks are images details, then we can get perfect result by using Dempster-Shafer (D-S) for accurate details classification.This article adopts different methods fussions and finally achieves relatively perfect segmentation result.