Based On The Mutual Information Of Multi-resolution 3d-mr Image Registration Method

The need for new approaches to image analysis has become increasingly important and pressing as advances in imaging technology enable more complex objects and processes to be imaged and simulated .The medical image registration technique is an important discipline of medical image processing which has been developing rapidly since 1990's It is widely used in modern clinical application and medical research.Firstly, this thesis introduces the present situation of the research and methods of medical image registration. The major classes of algorithms and relative techniques are reviewed. Then, this thesis give focus on the popular algorithm in the image registration—Mutual Information (MI) method. Compared with other algorithms, MI algorithm take advantage of high precision and robust and widely used is the clinical field. This thesis analyzes the reasons of the local extremum in Mi-based image regitstration, then implemented an improved method, which is integrated previous information to get a more smoothly object function and overcome interpolation artifacts effectively. Experimental results show that this method's registration accuracy is improved and can be applied.Secondly, we applied this method in 3D-MR image registration. By analyzing the pixels' grey level distributing character in MR, this paper proposes an automated method to extract the edge, then use chamfer distance transformation in pre-registration processing, this method can give a pixel accuracy start parameters of the Mi-based image registration method. To increase the registration speed of 3D MR , a multi-resolution optimization strategy is used for the registration algorithm, which can guarantee the accuracy and high speed.Finally, we present some suggestions of future work.