面向手持光刀法三维测量的点云滤波算法

Research on Point Cloud Filtering Algorithms for 3-D Measurement with Handheld Light Knife Method

手持光刀三维测量具有速度快、精度高、使用便携的优点,同时三维测量模型的点云数据存在碎片化、密度大、背景冗余等问题。针对该问题,在研究现有点云滤波算法基础上,提出基于包围盒的点云平滑、数据压缩方法,分割点云空间为若干网格,匹配点云数据与网格关系,滤除噪声点,分别对网格内点云进行中值滤波,精简点云数据。针对冗余背景提出了交互式滤波方法,将目标点云与背景点云进行了有效分割。实验结果表明,所提方法鲁棒性好,为三维测量模型重建提供了准确的数据预处理。

Three-dimensional measurement with hand-held light knife has the advantages of fast speed, high accuracy and portable use. Based on the analysis of three-dimensional measurement model, it is concluded that the point cloud data have the problems of fragmentation, high density and background redundancy. Based on the existing point cloud filtering algorithms, a point cloud smoothing and data compression method based on bounding box was proposed, which divides the point cloud space into several grids and matches the point cloud data with the grid relationship to filter out the noise points. Meanwhile, the point cloud in the grid was filtered by median filtering to simplify the point cloud data. Finally, an interactive filtering method was proposed for redundant background. The point cloud was segmented effectively. Experiments show that the proposed method has good robustness and provides accurate data prepossessing for three-dimensional measurement model reconstruction.