堆叠工件的点云采集与去噪算法研究Research on Point Cloud Collection and Denoising Methods for Stacked Workpieces
李伟尧,黄海龙,宋忠祥,曹衍琛
摘要(Abstract)：
激光三角法采集堆叠工件的点云数据时，会引入与目标点云无关的噪声点云，传统的点稀疏度去噪算法又会产生位姿估计偏差大，进而导致识别准确率低。针对以上问题，这里对堆叠工件的点云采集和去噪算法进行了研究。首先采用正射入式激光三角法获取堆叠工件的深度图像，然后计算相机参数矩阵和图像坐标矩阵，把深度图像中的数据转换为空间中的三维点坐标，得到堆叠工件的原始点云，接着提出了一种改进的点稀疏度去噪算法，对原始点云进行去噪处理，最后根据点云的点特征分离出目标点云。实验结果表明，与传统的点稀疏度算法相比，这里算法能够有效地去除噪声点云，提高点云三维重建精度，为后续的工件识别和位姿估计提供良好的基础。
关键词(KeyWords)： 激光三角法;噪声点云;点稀疏度算法;点特征;三维重建
基金项目(Foundation): 辽宁省“揭榜挂帅”科技计划重点项目（2021JH1/10400074）
作者(Author): 李伟尧,黄海龙,宋忠祥,曹衍琛
DOI: 10.19356/j.cnki.1001-3997.20260402.017
参考文献(References)：
- [1]王楷文.基于立体视觉的焊缝三维重建和机器人焊接轨迹自动规划[D].哈尔滨：哈尔滨工业大学，2021.(Wang Kaiwen.3D reconstruction of weld seam based on stereo vision and automatic planning of robot welding trajectory[D].Harbin:Harbin Institute of Technology,2021.)
- [2]王春香，孟宏，张勇.散乱点云的孔洞识别和边界提取算法研究[J].机械设计与制造，2019,337(3):74-76+81.(Wang Chunxiang,Meng Hong,Zhang Yong.Research on hole recognition and boundary extraction algorithms for scattered point clouds[J].Mechanical Design and Manufacturing,2019,337(3):74-76+81.)
- [3]韩先锋.三维点云去噪处理及特征描述的研究[D].天津：天津大学，2019.(Han Xianfeng.Research on 3D point cloud denoising and feature description[D].Tianjin:Tianjin University,2019.)
- [4]He Jianfeng,Deng Jiacheng,Zhang Tianzhu,et al.Hierarchical shapeconsistent transformer for unsupervised point cloud shape correspondence[J].IEEE Transactions on Image Processing:A Publication of the IEEE Signal Processing Society,2023.
- [5]钟文彬，肖振远，刘光帅.一种高效的点云去噪聚类方法[J].机械设计与制造，2022,378(8):233-237.(Zhong Wenbin,Xiao Zhenyuan,Liu Guangshuai. An efficient point cloud denoising clustering method[J].Mechanical Design and Manufacturing,2022,378(8):233-237.)
- [6]徐小薇.基于深度学习的三维点云去噪算法研究[D].长春：吉林大学，2022.(Xu Xiaowei.Research on 3D point cloud denoising algorithm based on deep learning[D].Changchun:Jilin University,2022.)
- [7]Cattai T,Delfino A,Scarano G,et al. VIPDA:A Visually Driven Point Cloud Denoising Algorithm Based on Anisotropic Point Cloud Filtering[J].2022.
- [8]田昊.基于机器视觉的生产线智能分拣系统的研究与设计[D].锦州：辽宁工业大学，2021.(Tian Hao.Research and design of an intelligent sorting system for production lines based on machine vision[D].Jinzhou:Liaoning University of Technology,2021.)
- [9]孙有春，庞亚军，白振旭，等.激光三角测量法应用技术[J].激光杂志，2021,42(4):1-8.(Sun Youchun,Pang Yajun,Bai Zhenxu,et al.Application technology of laser triangulation[J].Laser Journal,2021,42(4):1-8.)
- [10]杨昭辉.高精度视觉三维测量中双目相机标定及多视拼接技术研究[D].合肥：合肥工业大学，2022.(Yang Zhaohui. Research on binocular camera calibration and multi view stitching technology in high-precision visual 3D measurement[D].Hefei:Hefei University of Technology,2022.)
- [11]焦莉娟，王文剑.一种基于差异系数的稀疏度自适应图像去噪算法[J].计算机科学，2018,45(2):94-97+134.(Jiao Lijuan,Wang Wenjian.A sparsity adaptive image denoising algorithm based on difference coefficients[J].Computer Science,2018,45(2):94-97+134.)
- [12]林乐彬，周军，皇攀凌，等.基于图像检测的三维激光点云聚类方法研究与应用[J].控制工程，2023,5(13):1-8.(Lin Lebin,Zhou Jun,Huang Panling,et al.Research and application of3D laser point cloud clustering method based on image detection[J].Control Engineering,2023,5(13):1-8.)