Abstract: Towards to the problems of inconvenient to automatic camera calibration by the traditional circular target, and low precision of the circle extracting and the influence of circle points sorting result by the rotation angle of the target, a new circular target suitable for the automatic camera calibration is designed, a circle fitting method by the least squares based on the random sample consensus, and a sorting method based on the vector angle are proposed. The experimental results show that compared with the traditional least squares method, the mean fitting error is 0.006?1 pixels by the proposed arithmetic, and the sorting arithmetic can sort the circle points fast and accurately for random rotation angle of the target, and the mean sorting time is 0.009?6 seconds. The proposed arithmetic is easy to realize the automatic calibration online.

Key words: circle fitting, circle points sorting, Random Sample Consensus（RANSAC）, least squares, vector angle

摘要： 针对传统圆形靶不便于相机全自动标定、特征圆圆心提取精度不高以及圆心点阵排序受圆形靶旋转角度影响较大的问题，设计了适用于相机全自动标定的圆形靶，并针对该靶，提出了基于随机抽样一致的最小二乘圆心拟合法和基于向量夹角的圆心点阵顺序排序法。实验结果表明，采用该圆心提取算法，平均拟合误差为0.006?1个像素，相比传统的最小二乘拟合法，精度较高；对于标定靶0°~360°角度的旋转，均能够快速且准确地完成圆心点阵排序，平均排序时间为0.009?6?s，易于实现在线的相机全自动标定。

关键词: 圆心拟合, 圆心点阵排序, 随机抽样一致, 最小二乘, 向量夹角