基于点线特征的建筑物点云配准方法

Building Point Cloud Registration Method Based on Point Line Feature

针对建筑物表面具有明显的几何特征这一特点,提出基于点线特征的建筑物点云配准方法。首先基于点云配准基本理论,利用提取的建筑物点特征使用对偶四元数实现建筑物不同测站点云的粗配准,获取初始配准参数以及配准后的建筑物点云数据,然后将粗配准获取的参数作为待求精确参数值的初始值,利用建筑物中的线特征将共线方程作为精配准的数学模型,最后通过平差迭代获取参数的精确值,实现不同测站建筑物点云的高精度配准。实验结果表明,获取的配准后同名特征距离中误差为2.1×10^(-3)m,证明了将点线特征相结合可以有效改善建筑物点云配准质量,提高建筑物点云配准的精度,对建筑物三维重建具有重要意义。

A point cloud registration method based on point line features is proposed for building surface with obvious geometric features.Firstly,based on the basic theory of point cloud registration,the extracted building point features are used to realize the coarse registration of different building site clouds by dual quaternion,and the initial registration parameters and the registered building point cloud data are obtained.Then,the parameters obtained by coarse registration are used as the initial value of the accurate parameter values to be calculated,and the collinear equation is used as the mathematical model of fine registration by using the line characteristics in the building.Finally,the accurate values of parameters are obtained by adjustment iteration,and the high-precision registration of point clouds of buildings at different stations is realized.The experimental results show that the error of the same name feature distance obtained by this method is 2.1 X 10^(-3)m,which proves that the combination of point and line features can effectively improve the quality of building point cloud registration and improve the accuracy of building point cloud registration,which is of great significance to three-dimensional reconstruction of buildings.