基于平行直线的相机标定和相对定向
2018-07-26
作者：黄汝田，孙鹏
单位：北京信息科技大学现代测控技术教育部重点实验室
摘要：摄像机标定是实现机器人视觉系统的必要辅助手段，目前的摄像机标定方法大都基于点特征，当点特征被遮挡或者在纹理贫乏区域同名点稀少时，都会导致无法标定和定向，适用性不强。为克服这些障碍，本文提出一种新的相机标定和定向方法，仅需检测出空间四组共面的平行直线，并利用其进行相机标定和定向，无需提前给出特定的物体形状，这两组平行直线可以组成平行四边形或者矩形。
关键词：机器视觉；自标定；相机标定；相对定向
中图分类号：TG706；TH162文献标志码：A
Camera Calibration and Relative Orientation Based on Parallel Lines
Huang Rutian，Sun Peng
Abstract:Camera calibration is the necessary auxiliary means to realize the robot vision system，and the camera calibration methods are mostly based on point feature.When some characteristics or obscured in a poor texture，few corresponding point of the region will lead to can't calibration and orientation，the applicability is not strong.In order to overcome these obstacles，this paper proposes a new method for camera calibration and orientation，just detect space four groups of coplanar parallel straight lines，and use the camera calibration and orientation，without giving a specific object shape，these two sets of parallel lines can form a parallelogram，or rectangle.
Keywords:machine vision；calibration；camera calibration；relative orientation