基于正交的两组平行直线形成的消隐点的几何特性,提出了一种摄像机内外参数的自标定方法。该方法利用连接光心与消隐点向量的正交性质,建立关于相机内参数的约束方程,并给出了约束方程的线性解法;针对现有自标定方法未能标定畸变系数的现状,提出了一种考虑畸变的非线性最优化算法,该算法以线性求解得到的内参数为初值,利用非线性单纯型法寻优标定畸变;建立消隐点坐标系,给出了摄像机外参数的求解算法。所给出的自标定方法不需要知道空间点的精确坐标,具有易于实现、精度高和稳健性好的优点。

Based on the geometrical property of the vanish points determined by two perpendicular groups of parallel lines,a self-calibration approach was proposed to calibrate the camera′s intrinsic and extrinsic parameters. Some constraints on the intrinsic parameters were established via the perpendicular property of the vectors,which connect the optical center and the vanish points. These constraints can be solved linearly by using variable substitute. Aiming at determination of the lens distortion coefficient,a nonlinear optimization approach,based on nelder-mead simplex algorithm,was proposed when the center to center distance between adjacent sensor elements is known. The intrinsic parameters obtained by the linear approach can be used as the initial value of the nonlinear optimization algorithm. The extrinsic parameters were obtained by establishing a transitional coordinate system. The approaches proposed in this paper do not need the actual coordinate of the calibration points. Extensive simulations and experimental results with real images show that these proposed approaches are both accurate and robust.