露天采场验收测量的SIFT-ICP点云配准方法

SIFT-ICP point cloud registration method for open pit acceptance measurement

露天采场验收测量是露天矿山测量中的重要工作,传统的散点式测量方式如GPS、全站仪等逐渐向密集的点云测量方式转变,然而多期数据的坐标关联与统一需要点云数据的精确配准。为此,文中在分析以往ICP配准算法的存在不足的基础上,提出了一种改进的SIFT-ICP配准方法。该算法首先对预处理后的点云进行Delaunay三角剖分建模及栅格化,然后对栅格化后的图像实施SIFT算法的粗配准,最后根据粗配准的结果进行ICP算法的精确配准。实验结果表明,该方法能够加快ICP算法的迭代收敛,并提高了配准的精度。

Open pit acceptance measurement was an important task in open pit mine measurement.Traditional scatter measurement methods such as GPS and total station were gradually shifting to dense point cloud measurement. However,the coordinate association and uniformity of multi-period data required accurate registration of point cloud data. To this end,in the paper,based on the analysis of the existing ICP registration algorithm,an improved SIFT-ICP registration method was proposed. Firstly,Delaunay triangulation modeling and rasterization were carried out on the pre-processed point cloud,and then the coarse registration of the SIFT algorithm was carried out on the rasterized image. Finally,the ICP algorithm was carried out precise registration according to the result of coarse registration. Experimental results showed that the iterative convergence of ICP algorithm was accelerated and the accuracy of registration was improved.