摘要
3D视觉的点云数据处理中,滤波和配准等算法十分重要,传统的滤波算法,是通过求取距离半径,采样或利用统计学方法来对数据去除噪声,不能满足处理特定位置处数据的需求。提出了一种设置“基准面”的方法对点云数据进行预处理,使其满足后续的滤波处理。首先对给定的点云数据进行平面拟合并估计和平面质心的估计;其次求取某一已知平面的法线及基准点,求出两个点云面之间的变换关系,使两者重合;最后根据实际问题,将指定区域中的噪声去除。实验证明,该方法大大降低了滤波的复杂度且提高了滤波效果。
In point cloud data processing of 3D vision,filtering and registration algorithms are very important.Traditional filtering algorithms remove noise from data by distance radius,sampling or statistical methods which can not meet the needs for a specific location of the data clearly.In this paper,a method was proposed to pre-process point cloud data by setting“datum level”,so that it can satisfy the subsequent filtering processing.Firstly,the given point cloud data were initially fitted and estimated by normals and centroids.Secondly,the normals and datum points of zero-one plane were obtained,and the transformation relationship between the two point clouds was obtained,so that the two point clouds were coincided.Lastly,according to the actual problem,the noise in the specific area was removed.Experimental results show that this method greatly simplifies the complexity of filtering and improve the effects of filtering..
作者
李朋超
王金涛
宋吉来
LI Pengchao;WANG Jintao;SONG Jilai(State Key Laboratory of Robotics(Shenyang Institute of Automation,Chinese Academy of Sciences),Shenyang Liaoning 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang Liaoning 110169,China;University of Chinese Academy of Sciences,Beijing 100049,China;Shenyang SIASUN Robot&Automation Company Limited,Shenyang Liaoning 110168,China)
出处
《计算机应用》
CSCD
北大核心
2019年第S02期227-230,共4页
journal of Computer Applications
