摄象机内征参数的标定

Intrinsic Parameter Calibration of a Camera

从目标的视图定量地计算它的三维视觉信息必须首先标定摄象机的内征参数[1]。本文介绍一种基于平面板的多视图标定这些内征参数的技术，该技术适合于标定无非线性失真或非线性失真可以忽略的摄象机，尤其适合于标定水下摄象机，而且无须任何高精度装置，因此成本很低。我们根据一平面标定板的特征点集合在实体视图的计算机坐标系中的坐标与它在目标的模型坐标系中的坐标之间的矢量距离和内积是几何不变的原理，得到了关于每个视图的基准特征点在三维空间中的深度值及摄象机内征参数之间的非线性方程组;并使用简单有效的非线性选代计算标定出所需的内征参数。实验表明所开发的标定技术具有经济有效和快速准确的优点。

It is necessary to calibrate the lntrinsic parameters of a camera applied to visually measuring the location and attitude of an interested target in the 3-D space. A method of the camera calibration on the basis of the multi viewing images of a plane-shaped plate is introduced in this paper. The method is suitable for the cameras with no or negligible nonlinear distortion,especially for the underwater cameras. One of lts very attractive advantages is that any accurate equipment such as the accurate calibration stand is unnecessary in the calibration process. Thus,its calibration cost is very low.In this method,the nonlinear equations on the intrinsic parameters of a camera and the depths of the cardinal grids in each viewing are obtalned from the geometric invariance principle of the representation relatlonship,the dlstances and the vector inner-productions of the grid points on a plane-shaped plate under the transform from the model coordInate system to the computer coordinate system. And,then,the required parameters are calibrated from solvlng the equations by employing an efficient and effective nonlinear iteration. lt is experimentally demonstrated that the developed technique is efficient,effective,fast and accurate.