在基于圆点阵二维平面靶标的相机标定方法中,直接在拍摄的标靶图像中提取的椭圆中心并非真实的圆心投影像点,该圆心成像投影偏差必然会降低相机参数标定精度。基于此,提出将标靶图像逆向投影至空间虚拟矩阵以获得真实圆心像点的迭代标定算法。首先,使用椭圆中心提取方法进行平面相机标定;其次,由标定参数和拍摄图像进行逆向投影获得虚拟物理标靶图像,在近似圆虚拟图像上完成圆心坐标提取;再次,将虚拟圆心坐标转化为物理坐标值并投影到图像上,将其作为圆心真实像点坐标值来进行相机标定;最后,经多次迭代投影和标定完成高精度相机标定。模拟和实验结果表明,所提方法将相机标定精度提高约一倍,可为三维重建和视觉测量提供高精度的相机参数。

The camera calibration method based on a two-dimensional dot array calibration plate with its ellipse center directly extracted from the target image is not the real projection image point of the circle center. Thus, the accuracy of calibration parameters is reduced by the projection deviation of the circle center. Therefore, this study proposes an iterative calibration method for detecting the real circle center image point by re-projecting the target image back to the spatial virtual matrix. First, planar camera calibration was achieved using the ellipse extraction method. Second, the virtual physical target image was obtained using calibration parameters and captured images for re-projection; moreover, the feature points of the dot center were extracted on the virtual image of the approximate circle. Third, the virtual circle center coordinate values were converted into physical coordinate values and projected onto the image that was used as the coordinate value of the real circle center image point for camera calibration. Finally, high-precision camera calibration was completed via iterative re-projection and calibration. Simulation and experimental results show that the proposed method approximately doubles the camera calibration accuracy and provides high-accuracy camera calibration parameters for three-dimensional reconstruction and visual measurement.