摘要
在分析Opencv摄像机模型和现有的双目视觉系统标定方法的基础上,提出了一种新的标定算法。该算法包括3个步骤,首先利用最小二乘法求解图像的像面中心坐标,在不考虑畸变的情况下,选取中心处一小块作为摄像机标定的初始空间,采用线性优化方法求出除畸变系数以外的其他摄像机内外参数;然后将求得的参数作为初值,利用线性优化方法求解畸变系数;最后利用非线性优化方法对所有内外参数全局优化。实验结果表明:该方法能获得更高精度的标定参数,可以满足双目三维重建及其他应用的要求。
On the basis of analyzing Opencv camera distortion model and the existing calibration method of binocular vision system, a new calibration algorithm consisting of three stages was proposed. Firstly, the center coordinates of the image plane were estimated by least square algorithm. Selecting the area near the image center as the initial space of camera calibration, the internal and external parameters except for distortion parameters were estimated by linear optimization without regard to the distortion. Secondly, the obtained parameters were set as initial values to solve distortion parameters with linear optimization. Finally, the parameters estimated in the front two stages were improved iteratively through a nonlinear optimization, to make all the parameters globalization. The experimental results indicate that the algorithm can get more precise calibration parameters, which meets the requirements of binocular three-dimensional reconstruction and other application.
出处
《红外与激光工程》
EI
CSCD
北大核心
2011年第1期133-137,共5页
Infrared and Laser Engineering
基金
江苏省高校自然科学基金资助项目(06kja14003)
关键词
摄像机标定
计算机视觉
透镜畸变
非线性优化
camera calibration
computer vision
lens distortion
nonlinear optimization
作者简介
牛海涛（1985-），男，硕士生，研究方向为三维重建。Email：niuhaitaofeilong@126．com
导师简介：赵勋杰（1960-），女，教授，硕士生导师，博士，主要从事计算机视觉、图像信息处理等方面的研究。
