最小二乘法的摄像机标定  

Camera Calibration Method of Least Squares

作　　者：邰晶 翟为刚[1] 谭力宁[2] 

机构地区：[1]第二炮兵工程大学907教研室,西安710025 [2]第二炮兵工程大学701教研室,西安710025

出　　处：《现代计算机》2013年第15期25-28,41,共5页Modern Computer

摘　　要：为了获取空间三维点与图像二维像点之间的对应关系,摄像机标定必不可少。标定的实质是求取摄像机的内外部参数,确定物点的准确位置,而求解单应性矩阵是求解参数的前提和基础。利用最小二乘法来求解单应性矩阵,既提高单应性矩阵的精度,又为摄像机内外参数和畸变系数的精确测量提供理论依据。在此基础上借助OpenCV函数库获取图形中角点高精度坐标的功能,在Visual C++环境下实现对摄像机的标定。数值实验和实际标定实验均表明,提出的标定方法具有更高的精度和抗噪声能力。In order to obtain the 3D space and 2D image point correspondence, camera calibration is es- sential. Calibration is the essence of the internal and external parameters of the camera＇s re quest, determine the exact location of object points, and the homography matrix is the prerequi site and basis for solving parameters.The method can not only improve the accuracy of homog raphy matrix, but also provide the theory basis for the accurate measurement of the camera pa rameters and the distortion coefficient. With the help of high precision corner point coordinate obtained by OpenCV Library, the video camera can be calibrated by Visual C＋＋. Numerical ex periments and practical calibration experiments indicates that the calibration methods based on total least squares has more precision.

关 键 词：最小二乘法 摄像机标定 噪声 OPENCV 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]