Research and Implementation of Non-rigid Medical Image Registration Method Based on Improved Optical Flow Model
|Course||Signal and Information Processing|
|Keywords||Medical image registration Optical flow model Non-rigid registration ITK VTK OpenCV|
Nowadays, as the development of research in the field of the digital image processing, the medical imaging technology has been widely used in clinical work. Image registration is one of the most important technologies in this field. The current clinical practice requires the technology of the image registration applications to improve better. This paper researches non-rigid medical image registration algorithm deeply and purposes a novel non-rigid medical image registration method based on the improved optical flow modal..The paper focus on the non-rigid medical image registration after discuss the development history of the medical image registration and introduce the latest progress and the technology shortages at home and abroad. Discussed the standard of Digital Imaging and Communications in Medicine and the standard is widely used in medical equipment.Research of the important points of optical flow filed calculate method and the research hot spots of optical flow model. Compare the difference of optical flow field with the motion field. Through the model of the original optical flow image distortion cause by the reasons for registration of a careful analysis, this paper put forward a final optical flow field model to improve the registration method. Improved the regularization term of optical flow model and imported the motion-blurred image restoration algorism to solve the blurring question.The image processing platforms at home and abroad has been studied deeply. An image registration platform base on the ITK, VTK and VC++ has been developed, which apply the united programming style and open source. Moreover, many kinds of medical image algorithm will be used though this platform.A software program base on the SDK of the OpenCV has been designed, and the non-rigid medical image registration base on the Hausdorff measure and the improved optical flow modal and mutual information registration method has been achieved. Compare the algorism in this paper with the popular non-rigid medical image registration method and the original optical flow modal. The experiment result shows that we can achieve more accurate registration result by applying the registration method base on improved optical flow model to medical image registration. Integrated various algorithms to OpenCV platform and provided the basis for the future development of registration software platform.