摘要
在3D点云测量行业,点云数据容易受到背景噪声、随机噪声、传感器自身电子噪声影响,使测量值在静态、动态重复测量中鲁棒性较差,不利于精确测量。文章深入研究点云滤波技术,包含直通滤波、统计滤波、半径滤波等,结合点云本身几何形态以及法线特征,提出了一种基于几何形态特征与点云法线相结合的滤波方法,并引入去噪干净度来衡量不同方法滤波效果,最终得出形态特征与点云法线相结合的去噪方法更具有稳定性,更能保持数据完整性,且能够将粘黏在一起的非本形态状噪声有效去除,为后期尺寸测量提供稳定数据和精度的保障。
In 3D point cloud measurement,point cloud data are susceptible to background noise,random noise,and the electronic noise of the sensor itself,which makes the measured values less robust in static and dynamic repeated measurements,and is not conducive to accurate measurement.After an in-depth study of point cloud filtering techniques such as straight-pass filtering,statistical filtering,and radius filtering,combined with the geometric morphology and normal features of the point cloud itself,a filtering method based on the combination of geometric morphological features and point cloud normals is proposed.The denoising cleanliness is introduced to measure the filtering effect of different methods.It is concluded that the proposed method is more stable and more effective in maintaining data integrity,which can effectively remove the non-native morphological noise that sticks together.It provides stable data and accuracy for later dimensional measurements.
作者
周旭廷
刘剑琴
Zhou Xuting;Liu Jianqing(Anhui Institute of Information Engineering,Wuhu,Anhui 241000,China)
出处
《计算机时代》
2022年第12期70-72,77,共4页
Computer Era
基金
安徽信息工程学院特色示范软件学院(2021cyxy040)
安徽省教育厅高校自然科学重点项目(KJ2019A1291)。
关键词
点云
噪声
滤波
去噪干净度
point cloud
noise
filter
denoising cleanliness
