张正友法的摄像机标定试验

Experimental research of camera calibration based on ZHANG's method

针对机器视觉领域的摄像机标定问题,采用介于自标定与传统标定之间张正友法,对Sunway-130D数字摄像机进行标定。本研究首先阐述张正友法标定原理,即利用针孔模型匹配模板平面与其所成图像中的角点,计算出图像和模板之间的单应性矩阵,利用该单应矩阵线性解出摄像机内部参数,并由单应矩阵求出理想成像模型下的摄像机外部参数。然后考虑非线性畸变因素,求出畸变系数的初始值,最后对所有的标定参数进行迭代修正,通过非线性优化得到所有标定参数的最优解。制作模板并采集模板图像,利用MATLAB提取模板图像角点坐标并进行标定。结果表明:该方法可以有效地对摄像机进行标定;绝大多数偏差在(-1.5,1.5)像素之间,个别偏差超过1.5个像素,达到像素级精度,有较高标定精度。研究结果为进一步研究农业机械机器人奠定基础。

For the camera calibration in the field of machine vision, ZHANG＇s method was used to calibrate the Sunway 130D digital camera. The principle of ZHANG＇s method was elaborated. Though matching the template plane and its image corners by using the pin-hole model, the homography between the model plane and its image was calculated. The intrinsic parameters of the camera were worked out with the homography linear, and the extrinsic parameters were also computed with the homography in the ideal imaging model. The nonlinear radial distortion was taken into account and the initial value of the distortion coefficient was acquired. All the extrinsic and intrinsic parameters were modified with iteration, and the optimal solutions were obtained by nonlinear optimization. The model plane was manufactured and its images were acquired, and the coordinates of corners on the images were extracted and calibrated by MATLAB. This method is suitable for calibrating the camera, and most deviations are between （-1.5, 1.5） pixels, which is of high accuracy. It also provides a basis for the further study of the agriculture robot.