Camera calibration method based on circular ring

Abstract

A novel camera calibration method based on circular ring is proposed in this paper. It has been proven that the first two columns of calculated point transfer homography mapping from the circular ring plane to the image plane have one isometric ambiguity. But in that case the restriction of homography on the IAC (the image of absolute conic) is still tenable, so the restriction could be applied to the calibration of the internal camera parameters. The ambiguity of the first two columns of homography directly results in the isometric ambiguity of the rotation matrix which can be explained in geometry as the isotropy of circular ring. But the third column of homography has no ambiguity, so the unique of which could not lead to the ambiguity of the translation vector. The external camera parameters can be calibrated using circular ring while there is a discrepancy of isometric transformation of rotation matrix within the model plane, which is most distinguished from the principle of the other plane-based calibration method such as points, lines and multiple conics. The proposed method has two distinctly superiority over the calibration based on coplanar points or lines: Better noise immunity because of the global property of the circular ring feature, and automatic calibration because the image matching of the circular ring feature is much easier compared with the one of points or lines. Both simulation and real data are used to prove the correctness, high accuracy and robustness of our calibration method.