Automated registration of dense terrestrial laser-scanning point clouds using curves
Automated registration of dense terrestrial laser-scanning point clouds using curves
复制标题
使用曲线自动配准密集地面激光扫描点云
DOI：
10.1016/j.isprsjprs.2014.05.012
复制
发表时间：
2014-09-01
影响因子：
12.7
通讯作者：
Zang, Yufu
中科院分区：
文献类型：
--
作者：
Yang, Bisheng;Zang, Yufu
本文提出了一种自动配准地面激光扫描的鲁棒性和准确性的方法。该方法使用空间曲线作为匹配基元，以克服基于点、线或面片作为基元的配准方法的局限性。这些方法通常难以找到自由曲面的扫描点云之间的对应关系（例如，雕像、文化遗产）。该方法首先根据点云的几何曲率选取视觉上显著的点进行聚类，提取描述点云形状特征的脊线。其次，提出了一个变形能模型来衡量这些波峰线的形状相似性，以选择正确的匹配曲线对。基于这些对，可以获得良好的初始取向参数，从而导致精细配准。实验结果表明，该方法具有较好的鲁棒性和准确性，能够在不需要良好初始对准的情况下准确配准复杂场景。(C)2014年国际摄影测量与遥感学会（摄影测量和遥感学会）。由Elsevier B.V.出版，版权所有。
This paper proposes an automatic method for registering terrestrial laser scans in terms of robustness and accuracy. The proposed method uses spatial curves as matching primitives to overcome the limitations of registration methods based on points, lines, or patches as primitives. These methods often have difficulty finding correspondences between the scanned point clouds of freeform surfaces (e.g., statues, cultural heritage). The proposed method first clusters visually prominent points selected according to their associated geometric curvatures to extract crest lines which describe the shape characteristics of point clouds. Second, a deformation energy model is proposed to measure the shape similarity of these crest lines to select the correct matching-curve pairs. Based on these pairs, good initial orientation parameters can be obtained, resulting in fine registration. Experiments were undertaken to evaluate the robustness and accuracy of the proposed method, demonstrating a reliable and stable solution for accurately registering complex scenes without good initial alignment. (C) 2014 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved.