论文部分内容阅读
通过三维激光扫描技术可快速获取地质对象高精度空间信息,有效弥补了传统地质编录方法存在工作强度大、作业危险性高等问题,应用前景广阔。目前,对三维激光扫描地质编录技术的研究主要集中于岩体结构面提取方法方面,但现有方法存在提取结果不完整、准确度低等问题。本文提出了一种融合3D Hough变换和区域生长的点云分割方法,针对随机选取区域生长种子点稳健性不足的问题,通过结合法向量的3D Hough变换提高种子点选取稳健性,实现岩体结构面的自动提取。实验证明,本方法能有效区分产状差距微小或大小不同的结构面,对主要结构面提取的准确率较高。
The 3D laser scanning technology can quickly obtain the high-precision spatial information of geological objects, which effectively makes up the problems of large working intensity and high operational risk of traditional geological cataloging methods, and has wide application prospect. At present, the research on 3D laser scanning geological cataloging technology mainly focuses on the method of extracting rock structure plane, but the existing methods have some problems such as incomplete extraction results and low accuracy. In this paper, a method of point cloud segmentation based on 3D Hough transform and region growing is proposed. In order to solve the problem of insufficient robustness of seed points in randomly selected regions, 3D robust Hough transform based on normal vectors is used to improve the robustness of seed points selection and to realize the rock mass structure Automatic extraction of noodles. Experiments show that the proposed method can effectively distinguish between structural surfaces with small or small differences in production, and has a high accuracy in extracting the major structural surfaces.