Rankings
Publications
Search Publications
Cited-by Search
Sources
Publishers
Scholars
Scholars
Top Cited Scholars
Organizations
About
Scilit
About Us
Advisory Board
Data Sources
Career
Support / Contact
User Guide
Publisher User
Search Function
Rankings Function
API Resource
Scilit Knowledge
Metrics
Scilit Recognitions
Publications Classification
Media
News
Blog
Login
Register
Home
Publications
基于局部特征聚合与驱动点云采样的点云分类与分割
Home
Publications
基于局部特征聚合与驱动点云采样的点云分类与分割
基于局部特征聚合与驱动点云采样的点云分类与分割
Authors:
HS
Huang Shuqi
Huang Shuqi
ZZ
Zhou Zhifeng
Zhou Zhifeng
RP
Ren Pulin
Ren Pulin
ZC
Zhai Chuanrun
Zhai Chuanrun
Article Alert
Citation Alert
Add to Library
Cite
Download
Share
Download
Download PDF
Download
Publisher website
Google Scholar
1 January 2026
journal article
Published by
Journal of Applied Optics
in
Journal of Applied Optics
Vol. 47
(2)
,
340-348
DOI:
10.5768/jao202647.0202007
Abstract
针对目前大多数深度学习模型获取局部特征的能力比较欠缺的问题,提出了一种基于局部特征聚合与驱动点云采样的点云分类与分割网络——FCA-PointNet++。设计了局部特征聚合模块,在该模块中构建了增强相关局部特征并抑制不相关特征的FCA(feature channel attention)模块;设计了一种根据不同任务改变初始采样点分布的驱动点云采样网络(drive point cloud sampling,DPN);构造了一种能准确提取局部特征并且自适应细化特征的通道门(channel-wise gate)模块。在分类任务中,FCA-PointNet++在ModelNet10和ModelNet40上的总体准确率(overall accuracy,OA)分别达到了95.3%、93.4%。在分割任务中,FCA-PointNet++的平均交并比(mean intersection over union,MIoU)达到了86.1%,实验验证了FCA-PointNet++在分类和分割任务中的有效性。
Keywords
点云分类
点云分割
三维点云
深度学习
注意力机制
Related articles
Scroll to top