基于单经纬仪的视觉测量三维数据拼接方法

本文针对大型自由曲面三维视觉测量中的数据拼接问题，建立了经纬仪透视投影模型，提出了以平面靶标为中介坐标系，以经纬仪坐标系为全局坐标系的三维数据拼接方法。在大型自由曲面的若干测量子区域附近放置一个平面靶标，通过视觉传感器拍摄平面靶标上的特征点，求得视觉传感器坐标系到靶标坐标系的变换矩阵；通过经纬仪观测靶标上的特征点，求得靶标坐标系到经纬仪坐标系的变换矩阵。因此，可求得视觉传感器坐标系到经纬仪坐标系的变换矩阵。将视觉传感器所测得的各子区域的三维数据统一到了经纬仪坐标系下，即完成了大型自由曲面的全局测量。经实验验证，数据拼接的RMS误差小于0．487mm。

3D data registration method based on single theodolite in Vision measurement

The perspective projection model of theodolite is established, and a 3D data registration method which takes a planar target as intermediary coordinate frame and single theodolite as global coordinate frame is proposed for the problem of data registration in 3D vision measurement for large free surface. The planar target is placed near sub-areas of a large free-form surface, and the transformation matrix from the sensor coordinate frame to the target coordinate frame is estimated by taking image of feature points on the planar target using the vision sensor; the transformation matrix from the target coordinate frame to the theodolite coordinate frame is estimated by observing the feature points using the theodolite. The transformation matrix from the sensor coordinate frame to the theodolite coordinate frame is then found. All the 3D data of sub-areas obtained by the vision sensor are unified in the theodolite coordinate frame, and the global measurement for a large object is accomplished. The experimental results show that the RMS error of 3D data registration is less than 0.487 mm.