3D Registration Of Medical Image

Registration is one of the basic tasks of image process, which is to transform image from different or same resources to the same space locality. Registration is the base of many kinds of image process (location of ROIs, 3D modeling and so on), and its quality will directly influence the effect of other processing.Neuroscience is an important topic in nowadays, and today, there are many kinds of brain atlas. For develop research, it's necessary to give them a precise description of the ROIs.This research comes from the Wuhan Institute of Physics & Mathematics the Chinese Academy of Sciences scientific research project"3D reconstruction of rat brain atlas and MR image analyzing system". The purpose is to further perfect the image analysis and visualization based on"3D MR image processing system".The thesis is focus on the registration between the MR images and the rat brain Atlas. Image registration is an often encountered problem in various fields including medical imaging, computer vision and image processing. Numerous algorithms for registering image data have been reported in these areas, including linear registration methods and non-linear registration.In Chapter 2, two kinds of registrations, PCA and GA, which are based affine modal, are introduced. PCA is used to find corresponding points in images, and realize coarse registration based affine model. GA, which is to achieve exact registration, is introduced after registration by PCA. We could find that the combine is very perfect through the results of experimentation.In Chapter 3, some non-linear registration methods are introduced, and a novel rigid method of registration is presented. It makes up the deficiencies of linear registration, and realizes the perfect registration between MR images and rat brain Atlas.