深度学习方法用于遥感图像处理的研究进展  

Researches on Remote Sensing Images Processing Using Deep Learning Methods

作　　者：罗仙仙[1,2,3] 曾蔚 陈小瑜[4] 张东水 庄世芳[1,2,3] 

机构地区：[1]泉州师范学院数学与计算机科学学院,福建泉州362000 [2]福建省大数据管理新技术与知识工程重点实验室,福建泉州362000 [3]智能计算与信息处理福建省高等学校重点实验室,福建泉州362000 [4]泉州师范学院资源与环境科学学院,福建泉州362000 [5]湖南科技大学资源环境与安全工程学院,湖南湘潭411201

出　　处：《泉州师范学院学报》2017年第6期35-41,共7页Journal of Quanzhou Normal University

基　　金：泉州市科技计划项目(2016N057);国家级和各部委项目预研基金(2016YYKJ14);福建省教育厅科研A类项目(JA13270);王宽诚德意志学术交流中心博士后奖学金(91551268)

摘　　要：深度学习是当前机器学习与人工智能研究热点,深度学习方法用于遥感图像处理取得快速发展.首先简要介绍现有遥感数据源及其非监督与监督分类方法.在总结深度学习典型方法及其最新演化模型基础上,分析了深度信念网络、卷积神经网络、自动编码器在遥感图像处理中的国内外研究现状.针对当前应用现状与存在问题,指出今后研究方向:一方面要适应智能化遥感图像处理的发展趋势,加强算法理论研究,尤其人机协同工作、典型方法应用与修正、新模型拓展与应用;另一方面针对遥感大数据的应用需求,应加强遥感数据集建设、构建行业统一遥感大数据监测平台.Deep learning is the state-of-the-art of machine learning and artificial intelligence research,deep learning methods for remote sensing image processing have achieved rapid development.Firstly,the existing remote sensing data sources and their unsupervised and supervised classification methods are introduced.On the basis of summarizing the typical methods of deep learning and its latest evolution model,the research status of deep belief network,convolutional neural network and automatic encoder in remote sensing image processing at home and abroad are analyzed.According to the current application situation and associated issues,the development trend of deep learning＇s application in remote sensing image processing is discussed.On the one hand,it is necessary to strengthen theoretical research in order to adapt to the development trend of remote sensing image processing and intelligent algorithm,especially strengthening the man-machine collaboration,the application and revision of typical methods,the development and application of new model;on the other hand,according to the application requirements of remote sensing big data,it is also necessary to strengthen the construction of remote sensing data sets,and build a standard monitoring platform of remote sensing big data industry.

关 键 词：遥感 深度学习 图像处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]