中文摘要：

三维扫描需要将不同视角下获取的点云数据进行配准，并统一到全局坐标系。针对旋转台扫描得到的点云数据提出了一种自动配准方法，其基本思路是先用NDT算法作粗配准，将待配准的两片点云旋转平移到较好的初始位置，接着用ICP算法作精配准。经实验证明：该方法配准速度快、精度高。

英文摘要：

In 3D scanning, it is often necessary to merge point cloud data measured in different views in a global coordinate system. According to the point clouds scanned on turntable, an automatic registration algorithm is presented, which firstly uses NDT algorithm in initialregistration stage to optimize the pose of two point clouds, then uses ICP algorithm in accurate registration stage to achieve a higher registration accuracy. The experiment results show that this algorithm is of high efficiency and high precision.