云计算与深度学习协同的深圳市1986-2020年城市扩张分析  

Cooperating Cloud Computing Platform with Deep Learning for Urban Expanding Analysis in Shenzhen During 1986-2020

作　　者：罗新 周翔 胡忠文 杨超 邬国锋 LUO Xin;ZHOU Xiang;HU Zhongwen;YANG Chao;WU Guofeng(MNR Key Laboratory for Geo-environmental Monitoring of Great Bay Area,Shenzhen 518060,China;College of Life Sciences and Oceanography,Shenzhen University,Shenzhen 518060,China;College of Civil and Transportation Engineering,Shenzhen University,Shenzhen 518060,China;College of Architecture and Urban Planning,Shenzhen University,Shenzhen 518060,China)

机构地区：[1]自然资源部大湾区地理环境监测重点实验室,深圳518060 [2]深圳大学生命科学与海洋学院,深圳518060 [3]深圳大学土木与交通工程学院,深圳518060 [4]深圳大学建筑与城市规划学院,深圳518060

出　　处：《遥感信息》2022年第3期21-27,共7页Remote Sensing Information

基　　金：国家自然科学基金项目(42001351);中国博士后科学基金项目(2020M672804);深圳市科技创新委员会基础研究项目(JCYJ20180507182022554)。

摘　　要：深度学习技术的发展使得遥感信息的智能化提取水平有了极大的提高。然而,对海量遥感数据(如大尺度、长时间序列遥感数据)进行处理时智能化水平仍受到限制。文章探索了一种协同云计算和深度学习技术的遥感信息智能提取方法框架,通过对所构建深度学习模型进行云计算平台应用部署,实现遥感数据资源与计算资源的高效利用并以此提升遥感信息的智能化提取水平。基于Landsat 5 TM、Landsat 7 ETM+以及Landsat 8 OLI遥感影像,开展了深圳市1986-2020年逐年不透水面自动提取及城市扩张分析。结果表明,所提出方法能够实现高达94.93%平均总体精度的不透水面提取;通过对深圳城市不透水面扩张进行分析,表明近35年来,由于受不同时期经济社会发展状况和宏观土地政策影响,深圳市在时间上呈现“快速扩张-扩张放缓-加速扩张-趋于饱和”的阶段性特征。The development of deep learning technique significantly increasing the intelligence and refinement level in remote sensing information extraction.However,some limitations still exist when a large amount remote sensing data(e.g.,large scale,long time series)processing is required.In this study,we explored a new framework for remote sensing information extraction based on the cooperation of deep learning model and cloud computing platform.Generally,we deployed the deep learning model to the cloud computing platform and therefore facilitating highly efficient usage of the data resource and computing resource.Based on Landsat 5 TM,Landsat 7 ETM+and Landsat 8 OLI images,we further carried out a year-by-year impervious extraction in Shenzhen during 1986-2020 which is followed by an urban expansion analysis.As demonstrated by results,the obtained impervious surface information reaches to high overall accuracy of 94.93%.Through comprehensive analysis of the urban expansion in Shenzhen,in the last 35 years,due to society and economic development and land management policy in different stage,Shenzhen shows“fast expansion-slowing expansion-increasing expansion-saturation”characteristic in stages.

关 键 词：深度学习 云计算 智能化 不透水层 城市扩张 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]