基于B样条插值函数的人脑MR图像非刚体配准方法

提出了一种B样条插值函数结合图像特征标记的人脑MR图像非刚体配准方法。图像的特征标记选取图像的内外部轮廓来描述,目标和源特征之间的对应关系通过距离图来自动获得。形变过程采用B样条曲面函数插值来完成。通过多级B样条插值方法,即先全局计算后局部细化的方法来逐步优化形变结果。实验表明,该算法在不失准确度的前提下,具备较快的计算速度。

Non-rigid Method for Brain MR Image Registration Based on B-Splines Interpolation Function

A hybrid non-rigid method is presented based on B-splines interpolation function and landmarks of the image for human brain MR images registration.Landmarks are selected by the inner and the outer contours of the image,and the correspondence between them is automatically determined by the distance-map.The deformation is accomplished through the B-splines interpolation procedure.By the multilevel B-splines method the global registration is performed and then the result is refined using the local B-splines mesh thus obtaining a better result.Experiments on the registration of human brain MR images show that the proposed method offers efficient computation speed without loss of accuracy.