基于深度学习的卫星图像道路分割算法  

Road Segmentation Algorithm for Satellite Images Based on Deep Learning

作　　者：张新华 黄梦醒[2] 张雨[1] 李玉春 单怡晴 冯思玲[2] ZHANG Xinhua;HUANG Mengxing;ZHANG Yu;LI Yuchun;SHAN Yiqing;FENG Siling(School of Computer Science and Cyberspace Security,Hainan University,Haikou 570228,China;School of Information and Communication Engineering,Hainan University,Haikou 570228,China)

机构地区：[1]海南大学计算机与网络空间安全学院,海口570228 [2]海南大学信息与通信工程学院,海口570228

出　　处：《计算机工程》2021年第10期306-313,共8页Computer Engineering

基　　金：国家重点研发计划(2018YFB1703403,2018YFB1404400);海南省重点研发计划(ZDYF2019020);海南省高等学校科学研究项目(Hnky2019-22)。

摘　　要：针对道路分割时存在的梯度消失问题,构建基于U-Net的卫星道路图像语义分割模型。通过密集连接模块减少梯度消失,并引入空间空洞金字塔结构保留更多的图像特征,在学习深层次特征信息时采用注意力监督机制,提取道路要素的特征信息。在卫星图像道路数据集上的测试结果表明,与FCN、SegNet、U_Net算法相比,该算法模型的准确率、召回率和精确率指标分别达到96.3%、96.9%和96.6%,能够有效地对道路元素进行准确分割。The existing road segmentation methods generally suffer from gradient disappearance,and are limited in feature utilization and semantic segmentation accuracy.To address the problem,a U-Net-based semantic segmentation model is proposed for satellite road images.The dense connection module is used to reduce gradient disappearance.Then the Atrous Spatial Pyramid Pooling(ASPP)structure is introduced to retain more image features.Finally,the attention monitoring mechanism is used when learning deep-level feature information,so as to more effectively extract the feature information of road elements.The test results on a road data set from satellite images show that compared with FCN,SegNet and U_Net algorithms,the proposed algorithm improves the accuracy to 96.3%,recall rate to 96.9%and precision to 96.6%.This algorithm can segment the road elements accurately.

关 键 词：深度学习 道路分割 密集连接模块 空间空洞金字塔结构 注意力监督机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]