Serial Spect Images Quasi-automatic Segmentation Of The Study

As the effect of tomography used in industry and medicine fields becoming more markedly, image segmentation becomes a challengeable research subject in these fields. Image segmentation is a process that divides an image into several different regions according to the similarity or difference between regions. Separating the interested region from an image is the chief problem to be solved for the next image analysis and reorganization, and also is a choke point to the development and applications of relative tectonics in image processing field.On the basis of systemic research about recent image segmentation arithmetic, we class the arithmetic into several sorts and did comparison and analysis on them, then find the merits and demerits of the arithmetic. But practically, it's hard to get perfect effect using current automatic segmentation methods, and we have to add some interactive operations to get good segmentation effect.To sequential tomography images, a quasi automatic segmentation method is presented. First choose a fiducial image, and interactively choose some points forming a primary control polygon around the region to be segmented. Then each control point gets to the contour of the region according to certain rules and forms final control polygon. The next step is using the final control polygon approach the region contour by the midpoint method which put forward as a new segmentation method. For the latter sequential images, the primary control polygon of current image comes from the final control polygon of former image. And then get the contour by the methods mentioned above. The experiments show that, by using the midpoint method, can get continuous contour which still keep the original shape of the region to be segmented, and also the method can avoid lots of noise in the segmentation process. And the method transforms the segmentation into one dimensional problem that can improve efficiency and flexibility. So the segmentation method put forward has much high practical value.At last, conclude the paper work, point out the insufficient and further develop direction.