基于光流场模型的医学图像配准

基于最大化互信息的配准方法因其复杂性高而不能满足临床的实时性要求;基于光流场模型的图像配准方法计算简单快速,但已有的采用光流场进行配准的方法都直接采用原始的Horn模型,该模型简单地采用速度场的模作为光滑性约束,而导致配准过程不能保持图像的不连续性,使得图像严重模糊以致无法应用,该模型还存在不能处理大位移的问题.为了能在配准过程中尽可能地保持图像特征,采用了Lucas等人提出的窗口化光流场模型,通过选用小的窗口减轻对图像造成的模糊;对于大位移问题,可以采用金字塔结构来有效处理.实验结果证明,Lucas光流场模型应用于医学图像配准能够得到较准确的配准结果,具有很高的运行效率.

Optical Flow Model Based on Medical Image Registration

Medical image registration plays a very important role in computer aided diagnosis as well as in computer surgery navigation.The popular registration method based on mutual information maximization can get accurate result,but the algorithms are too complex to apply in practice.The registration methods based on optical flow can be conveniently implemented and the registration process is efficient.But all of the appeared registration methods based on optical flow use the initial optical flow model proposed by Horn,which could not preserve the discontinuity of image and fuzz the image seriously.In order to keep the image feature,the optical flow model based on window manipulation proposed by Lucas is used instead of the Horn model.Small window is used to reduce the blurring and the coarse-to-fine strategy is applied to deal with the large displacement problem.The experiment results show that accurate results and high efficiency can be obtained using the image registration method based on Lucas optical flow model.