结构体整体变形量提取算法.pdfVIP

2015 2015 年 测绘科学技术学报 第 32 卷第4 期 Journal of Geomatics Science and Technology Vo l. 32 No .4 文章编号: 1673-6338(2015)04-0336-ω 结构体整体变形量提取算法 马国正 (华东交通大学土木建筑学院，江西南昌 330013 ) 摘要:以往利用三维激光扫描提取变形量都是基于 DEM 模型进行变形量的提取。该方法提取的是一维变 形量，没有考虑不同方向的变形量。针对此，对基于 K-最邻近点搜索的 ICP 算法进行改进:并基于局部匹 配的方法，提取结构体的整体变形量，包括变形的方向和大小。首先，对 ICP 算法中最邻近点搜索容易出 现重复的地方进行了修改，构建了改进的 K-最邻近点搜索的 ICP 算法。其次，利用该算法对采集的两期点 云进行全局配准，并将配准后的结构体点云进行归一化处理。最后，利用该算法对结构体点云进行局部匹 配，从而可以得到结构体的整体变形量，包括 3 个旋转向量和 3 个平移向量。通过实例，验证了该方法在 结构体变形提取中的有效性。 关键词:三维激光扫描;结构体变形;变形提取;匹配;旋转;平移 中图分类号:P258 文献标识码:A DOI 编码: 1O.3969/j.issn.1673-6338.2015.04.002 The Extraction of Structure Deformation MA Guozheng (School of Civil Engineer and Architecture , East China Jiao Tong Vniversity, Nanchang 33∞13， China) Abstract: The extraction of deforrnation was based on DEM by using the point cloud of terrestrial laser scanning. This method extracts the 1 D deforrnation without considering the deforrnation of different direction. This paper im- proved the k-Nearest-Neighbor search ICP algorithm and extracted the structure deforrnation based on the local reg- istration , which include the direction of deforrnation and the size of the deformation. Firstly , the repeated point cloud of Nearest-Neighbor of ICP algorithm was improved. The improved k-Nearest-Neighbor search ICP algorithm was constructedly. Secondly , the two point cloud was transformed by the global registration using the proposed method , the point cloud of structure