亚像素断层图像配准数据集的建立

目的针对数字化冰冻铣切断层图像的特点,探讨一种实用的高精度图像配准方法,建立基于数字化断层图像的亚像素级配准数据集。方法采用冰冻铣切技术获取成年男性头颈标本的冰冻连续断层图像,在M atlab软件中自动提取定标点图像特征,采用基于2点的刚体变换算法实现图像的自动配准。结果配准后图像定标点与基准配准点的误差小于1个像素,达到亚像素水平。结论采用外定标的图像配准算法可建立亚像素级的配准数据集,定位标记物的准确识别是获得亚像素级配准数据集的保证。

Construction of sub-pixel level registration data set for sectional images

Objective To further investigate the method of image registration and to construct a sub-pixel level registration data set for sectional images.Methods A middle-aged male frozen specimen sectional images of the head and neck were acquired using cryosectiome method.With the software of Matlab,the character of the ex-calibrator was acquired automatically,and the data set were auto-registrated based on two point projective transformation algorithm.Results After registration,the error between the ex-calibrator and the benchmark was less than 1 pixel,which has reached sub-pixel level.Conclusion Registration algorithm with ex-calibrator can construct a satisfying sub-pixel level registration data set for sectional images.