基于曲面拟合的双向配准医学图像层间插值（研究生论坛）

针对传统的基于对象的层间插值算法对异常数据点不敏感，同时对形变场没有充分利用的缺点。为了更好解决在医学3D模型建立和分析的问题，提出一种新的层间插值方法。首先引入图像配准方法得到相邻序列的形变场，考虑到两张相邻序列配准方向不同会得到不同的形变场的特性，将两个形变场进行融合，用以得到信息更加丰富的形变场。其次，通过形变场得到配准图像之间的位移矢量集，并利用三次样条插值得到位移矢量集两端的灰度值。然后，利用线性关系得到位移矢量和插值层交点的灰度值。最后，利用薄板样条函数进行曲面拟合，得到位于图像坐标系上的灰度值。实验结果表明，本算法与其他算法相比，图像质量主观和客观上均有所提高。

Slice Interpolation for MRI Using Surface Fitting Method

The traditional slice interpolation algorithm is not sensitive to unusual data points, while the deformation vector field based underutilized shortcomings. In order to solve the problem in the medical establishment and analysis of 3D models, we proposed a new slice interpolation method .First introduced in image registration methods to get deformation field between adjacent sequences. Firstly, taking into account the registration in different directions will get different deformation field, we will fuse the two fields to get the information richer deformation field. Secondly, we will get the displacement vector set registration between images by deformation field, and use cubic spline interpolation to get the gray value set gray value at displacement vector ends. Thirdly, using the linear relationship between the displacement vectors and interpolation layer to get the gray in intersection. Finally, using the thin plate spline surface fitting to give the gray value of the image coordinate system. Experimental results show that the algorithm proposed by this paper compared with other algorithms, image quality is better.