基于特征影像的室内静态激光点云高效粗配准方法研究

Research on Efficient Coarse Registration Method of Indoor Static Laser Point Cloud Based on Feature Image

点云配准是室内静态激光点云数据处理中必不可少的一步,其配准效率直接影响配准算法的可用性。在处理大量密集的室内点云数据时,需要自动、高效的粗配准方法。为此,提出了一种室内静态三维激光点云自动、高效的粗配准方法。该方法充分利用静态激光扫描仪在数据获取中使用内置倾斜传感器补偿倾斜角的先验知识,将粗配准的自由度从六自由度降至四自由度,并通过三步法实现高效四自由度粗配准。结果表明,所提方法不仅能达到厘米级的配准精度,满足精配准精度的需求,而且配准的效率要优于其他对比方法。

Point cloud registration is an essential step in indoor static laser point cloud data processing,and its registration efficiency directly affects the usability of the registration algorithm.When processing a large amount of dense indoor point cloud data,an automatic and efficient coarse registration method is required.Therefore,this paper proposes an automatic and efficient coarse registration method for indoor static 3D laser point clouds.The method used the tilt angle in data acquisition as the prior knowledge to reduce the degree of freedom(DOF)of coarse registration from 6 DOF to 4 DOF,and through the three-step method to achieve efficient 4-DOF coarse registration.The results show that the proposed method can not only achieve centimeter-level registration accuracy and meet the requirements of fine registration accuracy,but also the registration efficiency is better than other comparison methods.