A scanning electron microscope (SEM) calibrating approach based on non-linear minimization procedure is presented in this article1. Both the intrinsic parameters and the extrinsic parameters estimations are achieved simultaneously by minimizing the registration error. The proposed approach considers multi-images of a multi-scale calibration pattern view from different positions and orientations. Since the projection geometry of the scanning electron microscope is different from that of a classical optical sensor, the perspective projection model and the parallel projection model are considered and compared with distortion models. Experiments are realized by varying the position and the orientation of a multi-scale chessboard calibration pattern from 300× to 10,000×. The experimental results show the efficiency and the accuracy of this approach.

本文介绍了一种基于非线性最小化过程的扫描电子显微镜（SEM）校准方法1。通过最小化配准误差，可以同时实现内在参数估计和外在参数估计。所提出的方法考虑了从不同位置和方向的多尺度校准图案视图的多图像。由于扫描电子显微镜的投影几何形状不同于传统的光学传感器，因此考虑了透视投影模型和平行投影模型，并将其与畸变模型进行比较。通过将多尺度棋盘校准图案的位置和方向从300倍更改为10,000倍来实现实验。实验结果表明了该方法的有效性和准确性。