基于局部特征聚合与驱动点云采样的点云分类与分割

Read full text Summarise with VIRAT
Authors: Shuqi Huang, 上海工程技术大学 机械与汽车工程学院,上海 201620, Zhifeng ZHOU, Pulin Ren, Chuanrun Zhai, 空装驻常州地区军事代表室,江苏 常州 213100, 上海司南卫星导航技术股份有限公司 产业研究院,上海 201800, Empty Military Representative Office in Changzhou, Changzhou 213100, China
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

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。

Fields of Study & Subjects

3D Shape Modeling and Analysis Advanced Neural Network Applications Domain Adaptation and Few-Shot Learning

Publication details

Journal / source
Journal of Applied Optics
Volume
47
Issue
2
Pages
340–348
Publication date
1 January 2026
Language
ZH