旋转平台点云数据的配准方法

A Registration Algorithm for Point Clouds Obtained by Scanning Objects on Turntable

提出一种旋转平台点云数据的配准方法。该方法首先利用3片以上不同角度的位于圆柱面上的点云数据标定出旋转平台中心轴的位置,从而获取扫描设备和旋转平台之间的相对关系。然后根据旋转平台的角度关系计算出多片点云之间的旋转矩阵,最后将任意角度获取的点云配准到统一的坐标系下。该方法不需要多片点云之间具有公共特征关系,也不需粘贴和提取标志点,仅利用旋转平台转动的角度构建旋转矩阵实现自动配准。利用C++语言实现该方法。试验证明,该方法算法稳定,效率较高,拼接精度与ICP配准方法和标志点配准方法的精度相当,自动化程度高,实现成本较低,可配合多种激光扫描仪或结构光扫描仪使用。

A registration algorithm for point clouds obtained by scanning objects on turntable with a stepper-motor is presented.Firstly three or more point clouds were used to calibrate the position of central axis of the rotating platform.So the relationship between the scanner and the rotating platform were defined.Then the rotate matrix was calculated with the angle of the rotating platform to rotate point clouds into the same coordinate system.Only the angle of the rotating platform was needed in this method.The method implemented with C＋＋ language had proved stable,efficient,high degree of automation,good registration performance and high practical value,which could be used in many 3D scanning system such as laser finder system and structure light scanning system.