三维激光扫描数据处理与曲面重建方法研究

发布时间：2018-01-20 07:48

本文关键词： 三维激光扫描技术点云数据点云去噪点云简化曲面重建　出处：《东华理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：三维激光扫描技术是二十世纪九十年代新兴的一门测量技术，它采用非接触式高速激光测量，以获取研究目标的三维坐标和数码照片的方式，快速高效的得到目标的三维立体信息。因此该技术被广泛应用于逆向工程、医学、考古、建筑业、船舶、电力、市政建设等众多领域。但是三维激光扫描获取的数据存在各种误差和噪声，而且一般得到的原始扫描数据密度较大，在利用这些数据进行后续的处理、曲面重建时，为了保证精度和速度，需要对原始数据进行预处理。另一方面，进行曲面重建时，涉及几何计算、拓扑关系建立以及三维模型的绘制问题。因此对三维激光扫描数据处理与曲面重建方法进行研究具有重大理论意义与现实意义。针对以上问题，本文从以下三个方面进行了深入研究。 1.点云去噪。在分析已有去噪算法的基础上，针对散乱点云，，给出了基于最小二乘法的分步去噪方法。通过对噪声点进行分类，先利用求取包围盒最大连通域的方法去除离群点，再用最小二乘法拟合出K邻域内点的最佳逼近平面，通过判定邻域各点到该平面的距离与设定的阈值的大小来去除振幅较小噪声，达到了预定效果。 2.点云简化。针对基于曲率简化的方法能较好的保留重建曲面的细节特征，但由于需要计算每个点的曲率并进行比较，效率较低，本文给出了基于轮廓点提取的简化方法，通过先提取轮廓点作为简化点云需要保留的特征点，然后再以轮廓点中曲率最小值为分界，把点云数据按曲率大小分为两个部分。对于大于等于最小值的部分，进行轮廓点保留，其他点删除处理，对于小于最小值的部分，根据曲率精简原则进行简化处理。该方法既继承了基于曲率简化的方法能较好的保留重建曲面的细节特征的优点，又在一定程度上减少了计算量，提高了效率。 3.曲面重建。介绍了常用的曲面重建方法：参数曲面重建、隐式曲面重建、分片线性曲面重建、细分曲面重建与变形曲面重建，并分析了各自优缺点。详细介绍了利用Geomagic Studio进行曲面重建的过程，最后用该软件对三个点云数据进行了曲面重建。
[Abstract]:In 1990s, 3D laser scanning technology is a new measurement technology. It uses non-contact high-speed laser measurement to obtain 3D coordinates and digital photos of the research object. Therefore, this technology is widely used in reverse engineering, medicine, archaeology, construction, ship, electricity. Municipal construction and many other fields. But 3D laser scanning data obtained by a variety of errors and noise, and generally the original scanning data density is high, in the use of these data for subsequent processing. In order to ensure the accuracy and speed of surface reconstruction, it is necessary to preprocess the original data. On the other hand, geometric calculation is involved in surface reconstruction. Therefore, it is of great theoretical and practical significance to study the methods of 3D laser scanning data processing and surface reconstruction. This article has carried on the thorough research from the following three aspects. 1. Point cloud denoising. Based on the analysis of existing de-noising algorithms, a step de-noising method based on least square method is presented for scattered point clouds. The noise points are classified. The outlier is removed by the method of finding the largest connected domain of the bounding box, and then the best approximation plane of the point in the K-neighborhood is fitted by the least square method. By determining the distance from each point to the plane and the value of the threshold value, the amplitude of the noise is reduced, and the predetermined effect is achieved. 2. Point cloud simplification. The method based on curvature simplification can better preserve the detailed features of surface reconstruction, but because of the need to calculate and compare the curvature of each point, the efficiency is low. In this paper, a simplified method based on contour point extraction is presented. Firstly, the contour point is extracted as the feature point to be retained by the simplified point cloud, and then the minimum curvature in the contour point is taken as the boundary. The point cloud data is divided into two parts according to the curvature. For the parts greater than or equal to the minimum, the contour points are reserved, the other points are deleted, and the parts less than the minimum value are processed. According to the principle of curvature reduction, the method not only inherits the advantages of curvature simplification method, but also reduces the computational complexity to a certain extent. Improved efficiency. 3. Surface reconstruction: parametric surface reconstruction, implicit surface reconstruction, piecewise linear surface reconstruction, subdivision surface reconstruction and deformation surface reconstruction. The process of surface reconstruction using Geomagic Studio is introduced in detail. Finally, three point cloud data are reconstructed with the software.
【学位授予单位】：东华理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P225.2

【参考文献】