Open Computing Medical Image Segmentation, Based On The Threshold Of Kfcm And Gray And Gray Matter

With the development of digital image processing technology and application requirements, it does not require the output is a complete image, but after pretreatment of the image, and then extracted through the segmentation and description of the characteristics of effective, then to be judged categories. In artificial intelligence, computer vision focused on developing and analyzing algorithm for image content. In the image analysis and processing, image segmentation is a hot and difficult research problem. Although there are many image segmentation methods, it does not exist a widely used optimal method, sometimes combine a variety of segmentation algorithms to obtain better segmentation results.To extract the white matter in magnetic resonance brain image(MRI), the first : this paper uses kernel fuzzy C means clustering algorithm (KFCM) on the distribution of the class was like reunion information and no interference with the first MRI brain image segmentation, when the KFCM algorithm parameters m=3,will to achieve better segmentation by the imulation results, then it discusses the parameters of elected too large, the segmentation result is not satisfactory, and the m's value is too large, the use of KFCM algorithm for image segmentation of the time will be longer. Second As the uneven image brightness or image brightness changes in the individual itself, using KFCM algorithm on MRI brain image segmentation would result in misclassification that are not well extracted white matter in MRI brain images. Therefore,this paper will effective combination KFCM algorithm, gray level threshold and gray matter of opening operation of the three algorithms. First, using KFCM algorithm extract a category contains images of white matter,using gray level threshold algorithm for selecting the appropriate threshold for segmentation (according to this type of image histogram selected threshold), the white matter surrounded by pseudo-Characteristics of edges removed, and then use mathematical morphology opening operation for further segmentation of gray matter, first remove the white matter segmentation by erosion around the remaining pseudo-edge, small particle noise, and then fill the image segmentation algorithm expansion is left after the corrosion hole under the segmentation results achieved good results, with the final area of overlap and misclassification rates of both degree of segmentation evaluation algorithm simulation results show that the segmentation algorithm designed to achieve the desired effect.