Medical Image Registration Based On Edge And Corner

Medical image fusion, a subject involving information science and medical imaging, is a new study topic in the field of medical image processing. As the issue should be first resolved in information fusion technology, medical image registration is a key part of integration. By processing two different images or two sets of image information, medical image registration can establish corresponding relations in space between two images, and the matching result makes the corresponding features of the two images reach agreement in space.Firstly, the background, application and development of medical image registration are introduced in the thesis. Then the present thesis analyzes in detail the principle and the steps to finish the medical image registration, including the geometry transformation, the interpolation method, the optimized algorithm and assessment of the principle of registration. Subsequently, according to whether extracting the feature of images or not, the present author divides methods into two parts, that is, feature extraction-based method and voxel similarity-based method. Then, the paper introduces and makes comparison through experiments the different images edge detection operator and presents several commonly used image feature extraction algorithm Corner. On this basis, the thesis studies the characteristics of medical image registration method, first proposed by Canny operator edge extraction, and then extracts the corner points which are on the edge by using MIC corner detection operator. In this thesis, the present author uses a method of the edge oriented of a corner to find the relationship between different corners to match the feature point, using rigid transformation and finally through the optimization algorithm PPSO to match the image registration. Theories and experiments indicated that this method has the advantages of high precision and high speed, thus is a kind of effective medical image registration.