Abstract: In view of camera calibration’s preferable robustness based on conic，a new method for camera calibration based on coplanar circles is proposed.The technique only requires three coplanar circles（or more） which are arbitrarily distributed.There is no demand of orientation of conics，the matching between model plane and image，and solution of non-linear system of equations.In the paper algorithm is described in view of geometry，and it’s validity is proved in view of algebra.Experiment results have shown that the method proposed is not only accurate and robust，but also has the better practicality because the method proposed is not image matching and information about geometric information of conic such as size and position.

Key words: conic, camera calibration, circular points, coplanar circles, projective invariant

摘要： 考虑到基于二次曲线这种几何基元的摄像机标定方法比基于点或直线的方法具有更好的鲁棒性，给出了一种新的基于共面圆的摄像机标定方法。该方法的主要特点是模板形式简单、易于制作，仅需任意分布的三个或三个以上共面圆，且不需要进行圆环定位；不需要模板与图像之间的匹配；也无需求解任何非线性方程组。从几何角度对算法进行形象描述，并从代数的角度给出了严格论证。模拟和真实图像实验表明，该算法精确度高，鲁棒性强，表现出了十分良好的实用性。

关键词: 二次曲线, 摄像机标定, 圆环点, 共面圆, 射影不变量