基于Tsai法的摄像机标定反推修正模型  

Revised Model of Camera Calibration Based on Tsai's Method

作　　者：田原嫄[1] 李驷洋[1] 潘敏凯[1] 

机构地区：[1]东北电力大学 机械工程学院,吉林吉林132012

出　　处：《东北电力大学学报》2013年第3期19-22,共4页Journal of Northeast Electric Power University

基　　金：博士科研启动基金(BSJXM-200907);吉林省教育厅科研项目(2010290);吉林市华隆有限公司横向科研项目(2012)

摘　　要：摄像机标定数学模型是通过双向推导建立世界坐标系与图像像素坐标系之间的关系,由图像坐标推导到世界坐标是一个尺寸放大的过程,如果用图像像素坐标点拟合出的标定系数求解世界坐标的时候就会把误差相应的扩大。提出基于Tsai法的摄像机标定数学模型进行单方向的反推修正模型,用图像像素坐标点表示相机坐标系下对应点坐标,然后再用相机坐标表示世界坐标,最后用世界坐标拟合曲线,从而得到精度比较高的标定系数。实验结果证明,反推法模拟出的标定系数精度为0.128 47,比Tsai法标定系数精度0.129 43要高。Generally, camera calibration model is made up of relationship between world and image pixel coordinates by bidirectional in reverse. The size is enlarged in the process that image coordinate induces to world coordinate. The coefficients calibrated by simulation with points in image pixel coordinate enlarge errors accordingly during resolving world coordinate. The paper has worked out to build mathematics model of calibration in reverse based on Tsai＇ s method. Firstly, points in image pixel coordinate express those in camera coordinate accordingly. Secondly, points in camera coordinate express those in world coordinate. Finally, points in world coordinate simulate curve. Above all, it can be obtained much accurate calibration coefficients. The experiment shows that the coefficients precise obtained by method proposed in the paper which is 0. 128 47 is higher than Tsai＇ s which is 0. 129 43.

关 键 词：机器视觉 摄像机针孔模型 标定 数学模型 畸变 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]