福建电脑2025，Vol.41Issue(3)：30-35,6.DOI:10.16707/j.cnki.fjpc.2025.03.007
混合池化在城市遥感图像语义分割中的应用
The Application of Hybrid Pooling in Semantic Segmentation of Urban Remote Sensing Images
摘要
Abstract
To improve the efficiency of semantic segmentation of urban remote sensing images,this paper studies the use of pooling technology to solve the problems of multi-scale and remote dependency establishment.Firstly,advanced feature maps are obtained through deep residual networks.Then,pyramid pooling mechanism and stripe pooling mechanism are used in parallel to obtain multi-scale information and establish remote dependencies.This hybrid pooling mechanism is used to optimize the segmentation effect.The experimental results showed that the pixel accuracy on the datasets ISPRS Vaihingen and ISPRS Potsdam reached 89.7%and 91.8%,and the average intersection to union ratios reached 78.7%and 79.1%.关键词
深度学习/语义分割/混合池化/遥感影像Key words
Deep Learning/Semantic Segmentation/Hybrid Pooling/Remote Sensing分类
计算机与自动化引用本文复制引用
张垚杰..混合池化在城市遥感图像语义分割中的应用[J].福建电脑,2025,41(3):30-35,6.基金项目
本文得到长治学院校级科研项目(No.020/XN0580)资助. （No.020/XN0580）
