摘要
针对SLAM技术获取的室内点云数据,提出一种先基于离散点云提取墙面点,再基于图像处理提取墙线的方法。该方法首先采用法向量粗提取墙面点,将粗提取的墙面点降维后,依据局部空间特征精确提取墙面点,再将点云数据转为二值图像,采用边界提取、膨胀、腐蚀、骨架提取和骨架特征点提取的方法处理图像,最后结合凸包算法实现室内墙线的提取。通过对某地下车库的SLAM点云数据分析进行验证,结果表明本文算法可准确有效提取室内墙线。
Aiming at indoor point cloud data obtained by SLAM technology, a method is proposed to extract wall points first and then extract wall lines based on image processing. This method first uses normal vectors to extract wall points and some local space features, then transform point cloud data into a binary image, the boundary extraction, expansion, corrosion, skeleton extraction and skeleton feature point extraction are taken by methods of image processing. Finally, combining with the convex hull algorithm, the wall lines are extracted. The proposed algorithm is verified by analyzing SLAM point cloud data of an underground garage, and the results show that the proposed algorithm can extract indoor wall lines accurately and effectively.
作者
严科文
彭海驹
林松
霍志红
赖浩源
Yan Kewen;Peng Haiju;Lin Song;Huo Zhihong;Lai Haoyuan(Huizhou Planning Survey Research Institute,Huizhou 516000,China)
出处
《工程勘察》
2023年第1期68-73,共6页
Geotechnical Investigation & Surveying
关键词
SLAM点云
墙线提取
凸包算法
图像膨胀腐蚀
骨架提取
SLAM point cloud
wall line extraction
convex hull algorithm
image expansion corrosion
skeleton extraction
