摘要
点云滤波处理作为点云数据处理的初始步骤,其去噪效果直接影响到后续点云分割、地物分类、三维模型重建等操作结果。不同滤波算法由于其原理不同,运行效率及适用性也存在很大差异,通过选取直通滤波、体素滤波、统计滤波、条件滤波、半径滤波这五种经典滤波算法,对其原理作详细描述。在PCL库的基础上,设置不同的阈值对每种算法进行了大量的实验及分析,验证参数设定对每种算法的运行效率,适用范围及滤波结果质量的影响。
As the initial step of point cloud data processing,and the denoising effect directly affects the quality of subsequent operation results such as segmentation and classification of point cloud,reconstruction of independent feature extraction 3d model,and other operation results.different filtering algorithms due to its principle,operation efficiency and applicability is also very big difference,By selecting five classical filtering algorithms,namely pass-through filtering,voxel filtering,filtering,conditional filtering and radius filtering,the principle of these algorithms is described in detail.Cloud Library,a lot of experiments and analyses were carried out for each algorithm by setting different thresholds to verify the influence of parameter setting on the operation efficiency,application scope and filter result quality of each algorithm.
作者
李瑞雪
邹纪伟
Li Ruixue;Zou Jiwei(Shanghai Jieshi Information Technology Co.,Ltd.;Jiangxi University of Science and Technology)
出处
《卫星电视与宽带多媒体》
2020年第13期82-84,共3页
Satellite TV & IP Multimedia
关键词
点云数据
预处理
滤波算法
阈值
PCL库
Point cloud data
Pretreatment
Filtering algorithm
The threshold value
Point Cloud Library
