一种基于3D点云的规则化模型自动构建方法

A Method of Building a Regularized Model Automatically Based on a 3D Point Cloud

摘要: 近年来人们对三维模型的需求与日俱增，三维激光点云技术的兴起为此提供了一个新的方向。然而，目前基于3D点云的建模工作大都通过人工交互的方式进行，因此如何利用3D点云数据实现三维模型的自动构建是目前研究的热点。本文提出一种基于3D点云的规则化模型自动构建方法：首先使用优化后的RANSAC算法对数据进行平面分割，然后提取出每个平面的轮廓线并进行分段；最后用线段中的关键点进行三角化，完成对三维模型的构建。实验结果表明，该方法可以完成独立平面和规则物体的三维模型自动构建，并可以灵活地与其他方法相结合以帮助构建物体中的规则部分，对规则物体的有较好的还原效果且所需三角面少。

Abstract: In recent years, people’s demand for 3D models is increasing day by day, and the rise of 3D laser point cloud technology provides a new direction for this. However, most of the current modeling based on 3D point cloud is carried out through human interaction, so how to use 3D point cloud data to realize the automatic construction of 3D model is the focus of current research. This paper proposes a method of automatic construction of a regularized model based on 3D point cloud: first, the data is split flat using the improved RANSAC algorithm, then the outline of each plane is ex-tracted and segmented, and finally the three-dimensional model is constructed by triangulation by the key points in the segment. Experimental results show that the method can complete the auto-matic construction of three-dimensional models of independent planes and regular objects, and can be flexibly combined with other methods to help build the rule part of the object, which has a good reduction effect on the regular object and requires fewer triangular faces.