基于锥形特征的二维横断面锚杆噪点剔除方法研究

Research on Noise Removal Method of 2D Cross-sectional Anchor Based on Conical Features

本次研究针对地下洞室扫描采集的激光点云中存在支护锚杆噪点的问题,提出了一种基于锥形特征的二维横断面锚杆噪点剔除方法,设计了洞室二维横断面点云获取和锚杆点云精剔除的过程。首先提取洞室点云的二维中轴线,根据不同位置局部点云密度来自适应截取横断面点云切片,降低单次数据处理量;然后通过构建移动检测窗口,结合支护锚杆的结构参数,初步剔除锚杆点云;再采用基于锚杆锥形特征的噪点精剔除方法,对初步剔除的锚杆点云进行重采样。实验结果表明:构建的移动检测窗口初步识别锚杆噪点数量略多于目视判读的噪点数量,无锚杆漏识别现象;采用基于锚杆锥形特征的噪点精剔除方法,总体上,点云数据精确剔除噪点前后的平均误判值由37.32%降至18.19%,能有效地为地下洞室点云后续进行变形分析提供高质量的点云数据。

To solve the problem of bolt noise in laser point cloud collected by underground cavern scanning,a method of removing bolt noise in 2D cross section based on conical features is proposed.The following processes are designed to obtain 2D cross section cavern point cloud and extract bolt point cloud essence.Firstly,the two-dimensional central axis of the cavity point cloud is extracted,and the cross-sectional point cloud slices are adaptedly intercepted according to the local point cloud density at different locations to reduce the single data processing capacity.Then,by constructing the mobile detection window and combining with the structural parameters of the supporting bolt,the bolt point cloud is preliminarily eliminated.Then,the noise precision elimination method based on the conical characteristics of bolt is adopted to resample the initially eliminated bolt point cloud.The experimental results show that the number of bolt noise identified by the constructed mobile detection window is slightly more than that of visual interpretation,and there is no bolt leakage recognition phenomenon.In general,the average misjudgment value of point cloud data before and after accurate removal of noise is reduced from 37.32%to 18.19%,which can effectively provide high-quality point cloud data for subsequent deformation analysis of underground cavity point cloud.