基于网格模板的相机标定方法

Camera calibration method based on grid pattern

为了给相机标定提供可靠的源数据,基于网格模板,提出了一种精确提取图像中网格线条交点的方法。该方法首先通过提取线条边界点来拟合线条边界曲线;然后根据线条两边界曲线得到线条中点坐标,并由中点坐标拟合线条中心线;最后对所有的中心线求交,进而得到网格线条交点,交点坐标定位精度优于0.1个像素。以所得的交点图像坐标及空间坐标为基础,以最小化空间点投影差为优化目标对相机进行标定实验,结果显示三组数据的解算过程均稳定收敛;解算所得参数中,畸变系数吻合很好,像主点和焦距偏差相对较大,表明此方法可以获取精确的相机镜头畸变系数及较准确的像主点和焦距。

In order to provide reliable source data for camera calibration, a method for precisely extracting the intersections of grid lines in the image was proposed based on grid pattern. First, the two edge curves of one line were fitted through extracting edge points of the line; subsequently, the middle points of the line were calculated based on two edge curves and the middle curve of the line was fitted by the middle points; lastly, the intersections, which have the locating precision of better than 0.1 pixel, were figured out by solving every two middle curve functions. Based on the image coordinate and spatial coordinate of the intersections, the calibration experiment was carried out with the optimization aim of minimizing the total projection error of spatial points. The results show that the solving process of three groups of data is stably convergent. The solved aberration coefficients are almost consistent, and the principle point and focal length have a distinct error, which indicate the method can get the accurate camera lens aberration coefficients. The principle point and focal length of relatively less precision can be got at the same time.