Abstract: Camera calibartion technology is the key step of recovering 3D shape from 2D images，and the precision of camera calibartion plays a crucial role to ensure the accuracy of the measurement.This paper raises an automatic camera calibration method based on the traditional calibration method.The improved calibration template，having a marked point of region，is used to obtain the sequence of calibration images.The edge detection technology is presented，which might help to find the region of feature point detection.According to the camera perspective projection model，it carries on the linear calibration to the camera，and uses the nonlinear to revise，then obtains the parameters of camera.Experimental results demonstrate that the improved calibration method has a good performance of speed，accuracy and robustness，and makes the traditional camera calibration automated.

Key words: computer vision, camera calibration, Tsai two-step calibration, feature point detection

摘要： 摄像机标定技术是从二维平面图像中获得三维信息的关键步骤，摄像机的标定精度对保证系统测量精度有着至关重要的作用。提出了一种基于传统相机标定方式的自动标定方法。采用带有区域标记点的改进标定模板获取标定图像序列。利用边缘检测技术，确定特征点检测区域。根据摄像机透视投影模型，对摄相机进行线性标定，再进行非线性修正，得出摄像机各参数。实验结果表明，改进的标定方法具有更快的检测速度，稳定性好，鲁棒性高，使传统相机标定过程实现了自动化。

关键词: 计算机视觉, 摄像机标定, Tsai两步标定法, 特征点检测