一种3D相机标定方法的研究

Research on a Calibration Method of 3D Camera

针对以参数化投影模型为代表的相机标定所求参数多、效率低,且大多数针对特定相机等问题,提出一种以广义成像模型为代表的相机标定方法,直接建立物方平面标定板与像方成像面投影之间的关系。对于Z向标定,通过光条纹中心的直线特性将投影到像面的中心点坐标与实际物方的中心点坐标建立非线性优化模型,利用Levenberg-Marquardt算法与多项式畸变模型迭代算出一阶径向畸变系数,从而完成畸变矫正;将矫正完成的像方坐标与物方坐标通过最小二乘法建立二次多项式关系,完成Z向标定。对于X向标定,提取棋盘格角点像方坐标,再与物方坐标建立多项式关系,完成标定。试验结果表明,对Z轴标定与X轴标定结果均满足精度要求。

To address the problems that the camera calibration represented by the parametric projection model has many parameters,low efficiency,and most of them are camera-specific,a camera calibration method represented by the generalized imaging model is proposed to directly establish the relationship between the object-side plane calibration plate and the projection of the image-side imaging plane.For Z-directional calibration,a nonlinear optimization model is established between the coordinates of the center point projected onto the image plane and the coordinates of the center point of the actual object plane through the linear property of the optical stripe center,and then the first-order radial aberration coefficient is calculated by using the Levenberg-Marquardt algorithm and the polynomial aberration model iteratively to complete the aberration correction;then the image plane coordinates and the object plane coordinates are corrected by the least squares method Then the corrected image-square coordinates and object-square coordinates will be established by the least-squares method to complete the Z-directional calibration.For the X-axis calibration,the image square coordinates of the checkerboard corner points are extracted,and then a polynomial relationship is established with the object square coordinates to complete the calibration.The test results show that both the Z-axis calibration and X-axis calibration meet the accuracy requirements.