固连于可控旋转平台相机的外参标定
External Parameter Calibration of Camera Mounted on a Controllable Rotating Platform
Received:May 13, 2013  Revised:August 08, 2013
DOI:10.7520/1001-4888-13-074
中文关键词:  可控旋转平台  相机标定  一维标志物  奇异性分析
英文关键词:controllable rotating platform  camera calibration  one-dimensional marker  singularity analyze
基金项目:
Author NameAffiliation
YANG Zhen 1.College of Aerospace Science and Engineering, National University of Defense Technology, Changsha 410073, China
2.Hunan Key Laboratory of Videometrics and Vision Navigation, Changsha 410073, China 
SHANG Yang* 1.College of Aerospace Science and Engineering, National University of Defense Technology, Changsha 410073, China
2.Hunan Key Laboratory of Videometrics and Vision Navigation, Changsha 410073, China 
Hits: 1467
Download times: 1386
中文摘要:
      为了进行大视场角的测量任务,可以将相机安装在运动精确可控的平台上,以平台的姿态运动来扩展相机的视场角。相机相对于平台的位姿关系的精确标定是保证测量结果准确的前提。本文主要利用一维标志物体,根据成像关系,建立了外参标定的约束方程; 发展了线性求解外参的方法,以及非线性优化的标定方法。为了克服线性法正交性差,以及非线性优化法不能求解平移向量的缺点,总结出了组合求解法,该方法可以应用于仅能够作旋转运动的平台。数值模拟和实验结果均表明姿态角的标定精度较高,而平移向量则对噪声十分敏感。最后分析了算法偶尔出现奇异的原因。
英文摘要:
      In order to carry out large viewing angle measurement task, the camera is mounted on a platform whose movement is controlled precisely, thus the camera viewing angle can be enlarged by platform attitude movement. The accurate calibration of camera's position and orientation with respect to platform is a prerequisite to ensure accurate measurement result. Using one-dimensional marker, a constrain equation for external parameter calibration was established based on the imaging relation. A linear solving method of external parameter and a nonlinear optimization calibration method were developed. In order to overcome poor orthogonality of linear algorithm, and overcome the shortcoming of nonlinear optimization that can not solve translation vector, a combination solving method was summarized, which can only be applied to rotating platform. Numerical simulation and experimental results show that the attitude angle can be calibrated with good precision, while the translational vector is very sensitive to noise. Occasional singularity of the algorithm was analyzed at last.
View Full Text  View/Add Comment  Download reader
Close