东南大学学报（自然科学版）2011，Vol.41Issue(3)：543-548,6.DOI:10.3969/j.issn.1001-0505.2011.03.022
基于立体靶标的摄像机标定方法
Camera calibration method based on 3D board
摘要
Abstract
A method of camera calibration based on 3D calibration board, which is used for vision measuring system, is proposed in this paper. Transform matrix is worked out with least square method on an ideal pin-hole camera model and then optimized by Levenberg-Marquardt(LM) algorithm.According to several transform matrices, initial camera intrinsic parameter matrix and corresponding extrinsic parameter matrices are solved. Then, the distortion coefficient is estimated according to the relationship between the real image and the detected image coordinate with initial camera parameter fixed. At last, LM algorithm is applied to the global optimization and the more accurate intrinsic and extrinsic parameters are obtained. Experimental results indicate that error of equivalent focal length is less than 0. 1％ and error of principal point is less than 0. 5 pixel when Gaussian noise with less than 0. 5 pixel standard deviation is added to the simulation image data. Under the same Gaussian noise level the more the pictures used, the smaller the standard deviation of camera parameters. Compared with other calibration methods 3D calibration board proposed in this paper is more accurate. More stable camera parameters can be obtained by using more pictures of the calibration board.关键词
摄像机标定/立体靶标/透镜畸变/投影矩阵/LM优化Key words
camera calibration/ 3D calibration board/ lens distortion/ transform matrix/ Levenberg-Marquardt (LM) ptimization分类
信息技术与安全科学引用本文复制引用
张捷,李新德,戴先中..基于立体靶标的摄像机标定方法[J].东南大学学报（自然科学版）,2011,41(3):543-548,6.基金项目
国家自然科学青年基金资助项目(60804063)、江苏省自然科学基金资助项目(BK201022557)、江苏省科技成果转化专项资金资助项目(BA2007058). （60804063）
