【摘要】 在数字减影血管造影中,常要用配准技术来消除因病人运动而形成的减影图像中的运动伪影。过去许多配准算法都忽视了图像特征,在这里介绍一种配准方法,它利用图像特征,在图像的边缘上选取控制点,采用差分图像局部熵作为相似性测度.该算法用C++编程实现,并和手工象素移动及网格法作比较,结果表明优于后两种方法。更多还原

【Abstract】 In digital subtraction angiography, registration techniques were frequently used to reduce motion artifact in subtracted image because of patient motion. However image feature was ignored in most registration algorithms. In this paper, a registration method is described, image feature was involved, control points are selected on edge, entropy measure is used as similarity measure. The algorithm was implemented in C++ programming language, compared with methods of manual pixel shifting and grid; results demonstrate the advantage of the former.更多还原