基于空间感知和特征增强的三维点云分类与分割研究
Research on classification and segmentation of 3D point cloud based on spatial awareness and feature enhancement
方银 1张惊雷 2文彪1
作者信息
- 1. 天津理工大学电气工程与自动化学院,天津 300384
- 2. 天津理工大学电气工程与自动化学院,天津 300384;天津市复杂系统控制理论及应用重点实验室,天津 300384
- 折叠
摘要
针对直接处理点云数据的深度神经网络PointNet++无法充分学习点云形状信息的问题,提出一种融合空间感知模块和特征增强模块(spatial awareness and feature enhancement,SAFE)的三维点云分类与分割方法(SAFE-PointNet++).首先,设计了空间感知(spatial awareness,SA)模块,使特征提取网络在特征升维时融合了包含空间结构的权重信息,增强了特征在空间上的表现力.其次,设计了特征增强(feature enhancement,FE)模块,通过把增强后的几何信息和附加信息拆分并分别进行编码,达到充分利用点云附加信息的目的.实验结果表明,在ModelNet40和S3DIS数据集上,SAFE-PointNet++与其他10种经典网络相比具有更高的分类和分割精度.
Abstract
To solve the problem that PointNet++,a direct point cloud data processing deep neural net-work,cannot thoroughly learn the shape information of point cloud,and SAFE-PointNet++(spatial awareness and feature enhancement PointNet++),a 3D point cloud classification and segmentation method is proposed,which combines both spatial awareness module and feature enhancement module(SAFE).Firstly,the spatial awareness(SA)module is designed to help the feature extraction network integrate the weight information of spatial structure when the feature dimension is raised,thus enhancing the expression function of the feature in space.Secondly,the feature enhancement(FE)module is de-signed so that the additional information of the point cloud can be fully used by respectively splitting and encoding the enhanced geometric information and additional information.The experiment results show that SAFE-PointNet++achieves higher classification and segmentation accuracy than the other ten classical networks on ModelNet40 and S3DIS datasets.
关键词
三维点云/空间感知(SA)/特征增强(FE)/分类与分割Key words
3D point cloud/spatial awareness(SA)/feature enhancement(FE)/classification and seg-mentation引用本文复制引用
基金项目
天津市研究生科研创新项目(2021YJSO2S27)
出版年
2024
NETL
NSTL
万方数据