三维激光点云的降噪算法对比研究

Comparative research on denoising algorithms of 3D laser point cloud

点云降噪是数据预处理中的一项基本操作,并且噪声滤除的效果将直接影响后期点云分割、配准、识别等操作成果的质量。受扫描环境及仪器影响,扫描生成的点云不可避免地会产生噪声点,必须借助点云降噪算法方能剔除。统计滤波与半径滤波是点云去噪中的常用方法,可有效地滤除离散点。本文基于PCL实现两种滤波算法,并对试验点云数据进行降噪处理,以验证滤波算法的有效性,最终分析参数设定对滤波效果的影响及对比两种滤波结果质量。

Point cloud noise reduction is a basic operation in data preprocessing,and the effect of noise filtering will directly affect the quality of post-point cloud segmentation,registration,recognition and other operational results.Affected by the scanning environment and the instrument,the point cloud generated by the scan will inevitably generate noise points,which must be eliminated by means of the point cloud noise reduction algorithm.Statistical filtering and radius filtering are common methods in point cloud denoising,which can effectively filter out discrete points.Based on PCL,this paper implements two filtering algorithms and denoises the experimental point cloud data to verify the effectiveness of the filtering algorithm.Finally,the influence of parameter setting on the filtering effect is analyzed and the quality of the two filtering results is compared.