论文部分内容阅读
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。
To quantify its 3D visual information from the target’s view, the intrinsic parameters of the camera must be calibrated first [1]. This article presents a technique based on flat panel multi-view calibration of these intrinsic parameters that is suitable for calibrating cameras that have no non-linear distortion or negligible nonlinear distortion and are especially suitable for calibrating underwater cameras without the need for any High-precision device, so the cost is very low. Based on the principle that a set of feature points of a planar calibration plate sets the vector distances and the inner products between the coordinate in the computer view of the physical view and the coordinate in the model coordinate system of the object, The nonlinear equations between the depth values of the reference feature points in three-dimensional space and the camera internal parameters are calculated and the required intrinsic parameters are calibrated using simple and effective nonlinear selection. Experiments show that the developed calibration technique has the advantages of economy, efficiency and accuracy.