旋转台多视场扫描点云的自动配准方法  

An Automatic Registration Algorithm for Multiple Point Clouds Scanned on Turntable

作　　者：王伟[1] 高健[1] 林辉[1,2] 

机构地区：[1]广东工业大学机械装备制造与控制技术教育部重点实验室,广东广州510006 [2]广东韶关学院物理与机电工程学院,广东韶关512005

出　　处：《机械工程与自动化》2017年第5期47-49,共3页Mechanical Engineering & Automation

基　　金：国家自然科学基金资助项目(51675106);广东省科技计划项目(2015B010104008;2015A030312008)

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。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.

关 键 词：点云配准 NDT算法 ICP算法 三维扫描 

分 类 号：TP391.76[自动化与计算机技术—计算机应用技术]