辽宁石油化工大学学报

辽宁石油化工大学学报 ›› 2023, Vol. 43 ›› Issue (6): 89-96.DOI: 10.12422/j.issn.1672-6952.2023.06.014

• 信息与控制工程 • 上一篇    

基于位置自适应的三维点云处理模型

侯健1(), 刘恒1, 刘琳珂1, 潘斌2(), 张玉萍2   

  1. 1.辽宁石油化工大学 人工智能与软件学院,辽宁 抚顺 113001
    2.辽宁石油化工大学 理学院,辽宁 抚顺 113001
  • 收稿日期:2022-04-27 修回日期:2022-05-18 出版日期:2023-12-25 发布日期:2023-12-30
  • 通讯作者: 潘斌
  • 作者简介:侯健(1997⁃),男,硕士研究生,从事计算机视觉方面的研究;E⁃mail:1650616360@qq.com
  • 基金资助:
    国家自然科学基金资助项目(61602228);辽宁省教育厅一般项目(L2020018);辽宁省“兴辽英才计划”青年拔尖人才项目(XLYC1807266);辽宁省自然科学基金项目(201502041);山东省自然科学基金项目(ZR2018MF006)

3D Point Cloud Processing Model Based on Local Position Adaptation

Jian HOU1(), Heng LIU1, Linke LIU1, Bin PAN2(), Yuping ZHANG2   

  1. 1.School of Artificial Inteligence and Software,Liaoning Petrochemical University,Fushun Liaoning 113001,China
    2.College of Sciences,Liaoning Petrochemical University,Fushun Liaoning 113001,China
  • Received:2022-04-27 Revised:2022-05-18 Published:2023-12-25 Online:2023-12-30
  • Contact: Bin PAN

摘要:

在点云处理领域中深度学习是一种主流的方法,但是现有方法对三维点云的局部结构信息利用不够充分,对局部形状感知较差。为此,提出了一种基于改进PoinetNet的三维点云处理模型,本模型将位置自适应卷积引入到PointNet中。位置自适应卷积采用动态的方式组合权重库中的权重矩阵来构造核函数,其中权重矩阵的系数是通过位置相对系数网络从点与点相对位置自适应学习得到的。通过此方式构建的核函数,可以更好地解决点云数据的不规则性和无序性问题。位置自适应网络在三维物体分类实验上分类准确率相较于PointNet提升3.60%,在三维物体零件分割实验上平均交并比相较于PointNet提升2.20%,在三维场景语义分割实验上平均交并比相较于PointNet提升9.14%。

关键词: 点云, 深度学习, 局部位置自适应, 分类, 零件分割, 语义分割

Abstract:

In the field of point cloud processing, deep learning is a mainstream method, but the existing methods do not fully utilize the local structure information of 3D point clouds, and have less local shape perception. We proposes a 3D point cloud processing model based on improved PoinetNet. Network model introduces position adaptive convolution into PointNet. The position?adaptive convolution constructs the kernel function by combining the weight matrices in the weight bank in a dynamic way, in which the coefficients of the weight matrix are adaptively learned from the relative positions of the points through the position?relative coefficient network. The kernel function constructed in this way can better solve the problem of irregularity and disorder of point cloud data. The classification accuracy of the position?adaptive network in the 3D object classification experiment is 3.60% higher than that of PointNet, and the average intersection ratio in the 3D object part segmentation experiment is 2.20% higher than that of PointNet. In the 3D scene semantics In the segmentation experiment, the average intersection and union ratio is improved by 9.14% compared with PointNet.

Key words: Point cloud, Deep learning, Local position adaptation, Classification, Part segmentation, Scene segmentation

中图分类号: 

引用本文

侯健, 刘恒, 刘琳珂, 潘斌, 张玉萍. 基于位置自适应的三维点云处理模型[J]. 辽宁石油化工大学学报, 2023, 43(6): 89-96.

Jian HOU, Heng LIU, Linke LIU, Bin PAN, Yuping ZHANG. 3D Point Cloud Processing Model Based on Local Position Adaptation[J]. Journal of Liaoning Petrochemical University, 2023, 43(6): 89-96.

使用本文

0
    /   /   推荐

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://journal.lnpu.edu.cn/CN/10.12422/j.issn.1672-6952.2023.06.014

               https://journal.lnpu.edu.cn/CN/Y2023/V43/I6/89