摘要
提出一种标定非量测摄像机成像中心和畸变参数的新方法,解决了不进行摄像机线性内外参数求解即可对多种单参数畸变模型进行标定的问题.该方法由两步组成:近似标定,添加直线与近似呈圆弧的直线段畸变成像形成闭合曲线,得出其闭合面积及端点与畸变参数及中心的近似关系;参考模型逼近,依据该闭合曲线端点及当前标定值,由畸变模型生成闭合曲线的参考面积值和曲线端点的无畸变成像位置,以此修正该近似关系,进而逼近准确的畸变中心与参数.仿真和真实图像实验均表明,该方法利用四条成像直线段即可实现高精度标定且收敛迅速.
A new method which can calibrate the image center and distort parameters of the non-metric camera was proposed,independent of a known imaging center which is fit for several different mono-parameter distortion models.This method is made up of following two parts.First,approximate calibration： the curves distorted from straight lines are closed by adding straight-line segments,and the area of those closed curves is computed.The approximate relationships between these areas with theirends and the distortion parameter with the center are elicited.Second,approaching by model reference：the approximate relationship is iteratively modified by the reference values of the curve areas and the undistorted end points,according to the distortion calibration,based on the endpoints of these curves and the current calibration.Thus,the distortion parameter and the center are iteratively approaching the real values.Simulations and real image tests show that this method can calibrate distortion exactly using only four imaging lines without known any camera＇s linear parameters,and convergent fast.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2014年第3期404-413,共10页
Journal of Zhejiang University：Engineering Science
基金
航空科学基金资助项目(20121396008
20135896025)
关键词
机器视觉
摄像机标定
非量测摄像机
畸变校正
machine vision
camera calibration
non-metric camera
distortion calibration
