基于FPFH特征的三维点云配准方法研究

Research on 3D point cloud registration based on FPFH features

随着三维激光扫描技术的发展,利用点云数据的三维重建技术在逆向工程、虚拟现实等领域有着广泛应用。在实际扫描测量中,由于环境条件不好或者目标物体被遮挡,为了获得被测物体完整的表面信息,通常需要在多视角下采集数据,点云配准将不同视角下的点云坐标变换到统一的坐标系,便于后续应用。目前点云配准有很多算法,本文主要研究基于FPFH特征描述子对点云进行粗配准,然后使用ICP算法配准,实现三维点云的精确配准。实验结果表明,本文使用的配准方法具有可行性与有效性,为点云配准提供了一种借鉴方法。

With the development of 3 D laser scanning technology, 3 D reconstruction technology using point cloud data has been widely used in reverse engineering, virtual reality, and other fields. In the actual scanning measurement, due to poor environmental conditions or the target object is blocked, in order to obtain the complete surface information of the object to be measured, it is usually necessary to collect data from multiple perspectives. Point cloud registration transforms the point cloud coordinates from different perspectives into a unified coordinate system, which is convenient for subsequent applications. At present, there are many algorithms for point cloud registration. This paper mainly studies the coarse registration of point cloud based on FPFH feature descriptor, and then uses ICP algorithm to achieve accurate registration. The experimental results show that the registration method used in this paper is feasible and effective, which provides a reference for the point cloud registration.