地面三维激光扫描形变监测关键技术研究

发布时间：2018-06-15 14:25

本文选题：地面三维激光扫描技术 + 点云滤波　；参考：《长安大学》2017年硕士论文

【摘要】：近年来,地面三维激光扫描(Terrestrial Laser Scanning,TLS)技术以其快速、高密度、高精度等特点被广泛应用于变形监测领域。这项技术在极大地方便了测绘外业的同时,将主要工作放在了内业数据处理上,其关键步骤有点云拼接、点云滤波、建立模型、多期模型配准和形变量解算,其中,点云滤波和多期模型配准问题一直是研究的难点和热点。本文对三维激光扫描技术的特点及分类、地面三维激光扫描技术的原理、数据处理和应用做了系统介绍,并在详细介绍了地面三维激光扫描数据滤波和DEM(Digital Elevation Model)模型配准现状的基础上,对点云数据滤波和无控制DEM配准进行了研究。首先,在二维聚类算法的基础上进行研究改进,提出了适用于点云数据处理的三维点云聚类滤波算法,有效避免了二维聚类算法使用中进行点云投影时不可避免的点云空间结构信息的损失,充分利用了TLS点云数据的高维度和高密度特点,并通过实验,将其与点云滤波处理常用的曲率平滑滤波算法、移动最小二乘趋势面法进行比较,验证了该算法在处理地形复杂的茂密植被区点云数据的优势;其次,针对无控制DEM配准问题,在现有迭代最近点(Iterations closest point,ICP)算法、最小高差(Least Z-Difference,LZD)算法和最小二乘3D表面匹配(Least squares 3D surface matching,LS3D)算法的基础上,用不同地形特征的仿真数据对进行实验,比较这三种算法在配准精度、配准效率和拉入范围方面的配准性能,为多时相无控制DEM数据的配准提供参考。然后使用上述方法对鸡冠岭地面三维激光扫描数据进行处理,解算其在10个月内的形变量并进行了相应的分析。
[Abstract]:In recent years, Terrestrial Laser scanning TLSs (TLSs) technology has been widely used in deformation monitoring due to its rapid, high density and high accuracy. This technology greatly facilitates the field of surveying and mapping at the same time, the main work is on the internal data processing, its key steps are cloud splicing, point cloud filtering, modeling, multi-phase model registration and shape variable calculation, in which, Point cloud filtering and multi-phase model registration are always the difficulties and hot spots. In this paper, the characteristics and classification of 3D laser scanning technology, the principle, data processing and application of 3D laser scanning technology are systematically introduced. Based on the detailed introduction of 3D laser scanning data filtering and digital elevation model registration, the point cloud data filtering and uncontrolled Dem registration are studied. Firstly, based on the research and improvement of two-dimensional clustering algorithm, a 3D point cloud clustering filtering algorithm is proposed, which is suitable for point cloud data processing. The loss of point cloud spatial structure information which is unavoidable in the point cloud projection in the use of two-dimensional clustering algorithm is effectively avoided, and the characteristics of high dimension and high density of TLS point cloud data are fully utilized, and the experiments are carried out. Compared with point cloud filtering and moving least square trend surface method, this algorithm has the advantages in dealing with point cloud data in dense vegetation areas with complex terrain. Secondly, aiming at the problem of uncontrolled Dem registration, it is proved that this algorithm has advantages in dealing with point cloud data in dense vegetation areas with complex terrain. On the basis of the existing iterations closest points algorithm, the least elevation difference algorithm and the least square 3D surface matching algorithm, the simulation data of different terrain features are used to compare the registration accuracy of the three algorithms. The registration efficiency and pull-in range provide a reference for the registration of multi-phase uncontrolled Dem data. Then, the above method is used to process the 3D laser scanning data of Jiguanling ground, and the shape variables within 10 months are calculated and the corresponding analysis is carried out.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P225.2

【参考文献】