Research And Realization Of Segmentation Algorithms For Medical Image

Recently, with the development of computer techniques, digital image processing and application increasingly play an important role in various fields. As an important branch of image segmentation, medical image segmentation has attracted extensive attention from all over the international academic community. Medical image segmentation refers to the prpcess in which medical images are divided into a series of homogeneous regions which don’t overlap mutually.And it is a basic technique of medical image processing which is an indispensable procedure to analyze the structures of human body, quantitatively or qualitatively. Image segmentation is one of the most crucial procedures before implementing many applications. Therefore, image segmentation has become a key subject in the digital image processing.The content of this paper is:(1) Applying the automatic seed placement to carry out the segmentation of colon image, comparing with the manual one, it reduces the human operation, rises the efficiency and accuracy of segmentation. The images of colon vary a lot and have been a difficulty in the image segmentation area. This thesis begins with the anatomic structure of colon, it introduces some very classical methods of segmentation, like thresholding, edge detecting and region growing. For the segmentation of colon, we have put forward a segmentation method of colon by combining several methods, to provide better image segmentation technique for colon assisted diagnostic system, which is applied to automatic analysis of colon images and prepares for the3D restruction of colon images.(2) Applying the level set method to medical image processing. The level set method which is based on deformable models converts the evolution of a2D closed curve (3D) closed surface) into an evolution of level set function which is a dimension higher implicit expression of the curve (surface).It has an adaptability of topology and high accuracy of calculation, thus the level set method has been widely used. However, level set method also has weak points, such as the large amount of calculation. This paper expounds the Narrow Band Method in details, and applies the algorithm to processing images of cells. For the images with topology change, the narrow band method gets nice results.