多站激光点云数据全自动高精度拼接方法研究

目的 针对目前多站点云数据拼接存在的效率低和自动化程度低等问题,提出了基于地面激光点云强度信息的2D-3D点云数据高精度全自动拼接方法。首先,将强度信息通过三次样条插值算法生成二维影像,采用基于图形处理器(GPU)的加速尺度不变特征变换(SIFT)算子匹配得到二维同名特征点,剔除粗差;然后,反算得到特征点在三维点云中的坐标,并通过三维空间法向量对三维同名特征点进行精炼。利用精炼的三维特征点进行多站点云数据拼接,可提高多站点云海量数据拼接的精度和效率。

Automatic and Accurate Mosaicking of Point Clouds fromMulti-station Laser Scanning

Objective The mosaicking of point clouds is a key step in point cloud processing.We propose a pointcloud mosaicking technology for multi-station laser scanning based on 2Dimage matching and 3Dcor-responding feature point refinement to solve the problems in existing point cloud mosaicking method-ologies for multi-station laser scanning,such as low efficiency,poor accuracy,and low automation.Firstly,the 2Dimages are generated from the derivative information from laser scanning data using in-terpolation algorithms.Secondly,2Dcorresponding feature points are obtained using GPU accelera-tion SIFT image matching,eliminating gross errors.Finally,3Dcorresponding feature points are ac-quired using an inversion algorithm;identifying whether they are same corresponding feature points inthe 3Dpoint cloud.Experiments demonstrate the feasibility and effectiveness of the proposed method.