Please wait a minute...
img

官方微信

遥感技术与应用  2020, Vol. 35 Issue (4): 767-774    DOI: 10.11873/j.issn.1004-0323.2020.4.0767
甘肃遥感学会专栏     
深度学习U-Net方法及其在高分辨卫星影像分类中的应用
杨瑞1,2(),祁元1(),苏阳1,2
1.中国科学院西北生态环境资源研究院, 甘肃 兰州 730000
2.中国科学院大学, 北京 100049
U-Net Neural Networks and Its Application in High Resolution Satellite Image Classification
Rui Yang1,2(),Yuan Qi1(),Yang Su1,2
1.Key Laboratory of Remote Sensing of Gansu Province, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China
2.University of Chinese Academy of Sciences, Beijing 100049, China
 全文: PDF(4191 KB)   HTML
摘要:

高分辨率遥感影像有精确的几何结构和空间布局,但是光谱信息有限,增大了对光谱特征相似地物的分类难度。针对高分辨率遥感影像分类的问题,采用深度学习U-Net模型分类方法。基于黑河下游额济纳绿洲高分二号遥感影像,通过U-Net模型提取胡杨、柽柳、耕地、草地和裸地五种地物覆被类型,分类总体精度和Kappa系数分别为85.024%和0.795 6,并与传统的支持向量机(SVM, Support Vector Machine)和面向对象的分类方法比较,结果表明:相对于SVM和面向对象,基于U-Net模型的高分辨率卫星影像地物覆被分类,能够更好地对地物本质特征进行提取,分类效果较好,满足精度要求。

关键词: 深度学习U-Net模型高分二号遥感影像SVM分类    
Abstract:

High-resolution remote sensing images have precise geometric structure and spatial layout, but the spectral information is limited, which increases the difficulty of classifying similar features of spectral features. Aiming at the problem of high resolution remote sensing image classification, a U-Net convolutional neural network classification method based on deep learning is proposed. Based on the remote sensing image of the Ejina Oasis GF-2 in the lower reaches of the Heihe River, the U-Net model was used to extract the five types of land cover types of Populus euphratica, Tamarix chinensis, cultivated land, grassland and bare land. The overall classification accuracy and Kappa coefficient were 85.024% and 0.795 6 respectively. Compared with the traditional Support Vector Machine(SVM) and object-oriented method, the results show that compared with SVM and object-oriented method, the U-Net model is used to classify the high-resolution remote sensing, and the classification effect is better. The ground extracts the essential features of the features to meet the accuracy requirements.

Key words: Deep learning    U-Net model    Gaofen-2 remote sensing image    SVM    Classification
收稿日期: 2019-01-29 出版日期: 2020-09-15
ZTFLH:  TP75  
基金资助: 中国科学院A类战略性先导科技专项(XDA20100101)
通讯作者: 祁元     E-mail: yangrui@lzb.ac.cn;qiyuan@lzb.ac.cn
作者简介: 杨 瑞(1993-),女,甘肃平凉人,硕士研究生,主要从事生态遥感研究。E?mail:yangrui@lzb.ac.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
杨瑞
祁元
苏阳

引用本文:

杨瑞,祁元,苏阳. 深度学习U-Net方法及其在高分辨卫星影像分类中的应用[J]. 遥感技术与应用, 2020, 35(4): 767-774.

Rui Yang,Yuan Qi,Yang Su. U-Net Neural Networks and Its Application in High Resolution Satellite Image Classification. Remote Sensing Technology and Application, 2020, 35(4): 767-774.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2020.4.0767        http://www.rsta.ac.cn/CN/Y2020/V35/I4/767

图1  U-Net神经网络总体结构
图2  二维卷积过程
图3  数据集样本选取方式
图4  U-Net网络模型分类技术流程图
图5  GF-2标准假彩色与分类结果图
图6  研究区域的典型区域(A、B、C)分类结果对比
图7  数据集损失函数和精度关于迭代次数的变化曲线

深度学习

网络模型

SVM/%面向对象/%U-Net/%
Kappa0.726 40.753 00.795 6
裸地74.7084.3490.09
胡杨75.3381.3386.50
耕地76.3683.0682.56
柽柳82.3573.9971.01
草地85.1587.3189.59
总体精度79.02882.05285.024
表1  分类结果精度评价表
1 Zhao Yingshi. Remote Sensing Application Analysis Principle and Methods[M]. Beijing: Science Press, 2003.
1 赵英时. 遥感应用分析原理与方法[M]. 北京: 科学出版社, 2003.
2 Yuan Chen. Research for Classification of High Spatial Resolution Remotely Sensed Imagery[D].Xi’an:Chang’an University, 2016.
2 元晨. 高空间分辨率遥感影像分类研究[D].西安:长安大学, 2016.
3 Lobo A, Chic O, Casterad A. Classification of Mediterranean Crops with Multisensor Data: Per-pixel Versus Per-object Statistics and Image Segmentation[J]. International Journal of Remote Sensing, 1996, 17(12): 2385-2400.
4 Gualtieri J A, Cromp R F. Support Vector Machines for Hyperspectral Remote Sensing Classification[C]//27th AIPR Workshop: Advances in Computer-assisted Recognition. International Society for Optics and Photonics, 1999, 3584: 221-233.
5 Foody G M, Mathur A. A Relative Evaluation of Multiclass Image Classification by Support Vector Machines[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(6): 1335-1343.
6 Li Ying, Li Yaohui, Wang Jinxin, et al. A Comparative Study of SVM and ANN in Multispectral Image Classification[J].Hydrographic Surveying and Charting,2016,36(5):19-22.
6 李颖, 李耀辉, 王金鑫, 等. SVM和ANN在多光谱遥感影像分类中的比较研究[J]. 海洋测绘, 2016, 36(5): 19-22.
7 Bottou L,Chapelle O, DeCoste D,et al. Large-scale Kernel Machines[M]. London:MIT Press, 2007.
8 Hinton G E, Salakhutdinov R R. Reducing the Dimensionality of Data with Neural Networks[J]. Science, 2006, 313(5786): 504-507.
9 Hannun A, Case C, Casper J, et al. Deep Speech: Scaling up End-to-end Speech Recognition[J].Preprint arXiv:1412.5567, 2014.
10 Shen Y, He X, Gao J, et al. A Latent Semantic Model with Convolutional-pooling Structure for Information Retrieval[C]//Proceedings of the 23rd ACM International Conference on Conference on Information and Knowledge Management. ACM, 2014: 101-110.
11 LeCun Y, Boser B, Denker J S, et al. Backpropagation Applied to Handwritten Zip Code Recognition[J]. Neural computation, 1989, 1(4): 541-551.
12 Cao Linlin, Li Haitao, Han Yanshun, et al. Application of Convolution Neural Network in Classification of High Resolution Remote Sensing Image[J]. Science of Surveying and Mapping, 2016, 41(9):170-175.
12 曹林林, 李海涛, 韩颜顺,等. 卷积神经网络在高分遥感影像分类中的应用[J]. 测绘科学, 2016, 41(9):170-175.
13 Collobert R, Weston J, Karlen M, et al. Natural Language Processing (Almost) from Scratch[J]. Journal of Machine Learning Research, 2011, 12(1): 2493-2537.
14 Collobert R, Weston J. A Unified Architecture for Natural Language Processing: Deep Neural Networks with Multitask Learning[C]//Proceedings of the 25th international conference on Machine learning. ACM, 2008: 160-167.
15 LeCun Y. Generalization and Network Design Strategies[C]//Connectionism in Perspective,Elsevier,1989: 143-155.
16 Hu W, Huang Y, Wei L, et al. Deep Convolutional Neural Networks for Hyperspectral Image Classification[J]. Journal of Sensors, 2015, 2015:258619. doi:10.11551 2015/258619.
doi: 10.11551 2015/258619
17 Liu Dawei, Han Ling, Han Xiaoyong. Research on Classification of High Resolution Remote Sensing Image based on Deep Learning[J]. Acta Optica Sinica, 2016, 36(4):298-306.
17 刘大伟, 韩玲, 韩晓勇. 基于深度学习的高分辨率遥感影像分类研究[J]. 光学学报, 2016, 36(4):298-306.
18 Maggiori E, Tarabalka Y, Charpiat G, et al. Fully Convolutional Neural Networks for Remote Sensing Image Classification[C]//2016 IEEE International.Geoscience and Remote Sensing Symposium (IGARSS), 2016: 5071-5074.
19 Zhang C, Pan X, Li H, et al. A Hybrid MLP-CNN Classifier for very Fine Resolution Remotely Sensed Image Classification[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 140: 133-144.
20 Zhang C, Sargent I, Pan X, et al. VPRS-based Fegional Decision Fusion of CNN and MRF Classifications for very Fine Resolution Remotely Sensed Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(8): 4507-4521.
21 Ronneberger O, Fischer P, Brox T. U-net: Convolutional Networks for Biomedical Image Segmentation[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, Cham, 2015: 234-241.
22 Li R, Liu W, Yang L, et al. Deepunet: A Deep Fully Convolutional Network for Pixel-level Sea-land Segmentation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018 (99): 1-9.
23 Zhang Z, Liu Q, Wang Y. Road Extraction by Deep Residual U-Net[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(5): 749-753.
24 Zhou Y T, Chellappa R. Computation of Optical Flow Using a Neural Network[C]//IEEE International Conference on Neural Networks. 1988, 1998: 71-78.
25 Golik P, Doetsch P, Ney H. vs Cross-entropy. Squared Error Training: a Theoretical and Experimental Comparison[C]//Interspeech, 2013(13): 1756-1760.
26 Li Xin, Liu Shaomin, Ma Mingguo, et al. HiWATER: An Integrated Remote Sensing Experiment on Hydrological and Ecological Processes in the Heihe River Basin[J]. Advances in Earth Science, 2012, 27(5): 481-498.
26 李新, 刘绍民, 马明国, 等. 黑河流域生态—水文过程综合遥感观测联合试验总体设计[J]. 地球科学进展, 2012, 27(5):481-498.
27 Zhang Xiaoyou, Gong Jiadong, Zhao Xue, et al. The Change of Land Cover Land Use in Ejina Oasis over 20 Years[J]. Advances in Earth Science, 2005, 20(12): 1300-1305.
27 张小由, 龚家栋, 赵雪, 等. 额济纳绿洲近20年来土地覆被变化[J]. 地球科学进展, 2005, 20(12):1300-1305.
28 Forestry Ejinaqi. Populus Euphratica in Ejina National Nature Reserve[J].Journal of Inner Mongolia Forestry, 2007(3): 19.
28 额济纳旗林业局. 阿拉善戈壁上的绿色明珠——额济纳胡杨林国家级自然保护区[J].内蒙古林业, 2007(3):19.
29 Gong Jiadong, Cheng Guodong, Zhang Xiaoyou, et al. Enviromental Changes of Ejina Rrgion in the Lower Reaches of Heihe River[J]. Advances in Earth Science, 2002, 17(4):491-496.
29 龚家栋, 程国栋, 张小由,等. 黑河下游额济纳地区的环境演变[J]. 地球科学进展, 2002, 17(4):491-496.
30 Fu Aihong, Chen Yaning, Li Weihong. Water Use Strategies of The Desert Riparian Forest Plant Community in the Lower Reaches of Heihe River Basin[J]. China Science: Earth Sciences, 2014(4): 693-705.
30 付爱红, 陈亚宁, 李卫红. 中国黑河下游荒漠河岸林植物群落水分利用策略研究[J]. 中国科学: 地球科学, 2014(4): 693-705.
[1] 李宏达,高小红,汤敏. 基于CNN的不同空间分辨率影像土地覆被分类研究[J]. 遥感技术与应用, 2020, 35(4): 749-758.
[2] 李净,温松楠. 基于3种机器学习法的太阳辐射模拟研究[J]. 遥感技术与应用, 2020, 35(3): 615-622.
[3] 虞瑶,苏红军,姚文静. 基于Boosting的高光谱遥感切空间协同表示集成学习方法[J]. 遥感技术与应用, 2020, 35(3): 634-644.
[4] 张峰极,吴艳兰,姚雪东,梁泽毓. 基于改进DenseNet网络的多源遥感影像露天开采区智能提取方法[J]. 遥感技术与应用, 2020, 35(3): 673-684.
[5] 谷祥辉,张英,桑会勇,翟亮,李少军. 基于哨兵2时间序列组合植被指数的作物分类研究[J]. 遥感技术与应用, 2020, 35(3): 702-711.
[6] 程娟,肖青,闻建光,唐勇,游冬琴,卞尊健,吴胜标,郝大磊,钟守熠. 地物波谱数据库应用方法及遥感应用现状[J]. 遥感技术与应用, 2020, 35(2): 267-286.
[7] 柴旭荣,李明,周义,王金风,田庆春. 影像的土地覆被快速分类[J]. 遥感技术与应用, 2020, 35(2): 315-325.
[8] 李强,冯德俊,瑚敏君,伍燚垚,杨历辉. 集成特征分量的高分二号影像阴影检测[J]. 遥感技术与应用, 2019, 34(6): 1252-1260.
[9] 刘培,余志远,马威,韩瑞梅,陈正超,王涵,杨磊库. 基于地形信息的Landsat与Radarsat-2遥感数据协同分类研究[J]. 遥感技术与应用, 2019, 34(6): 1269-1275.
[10] 李越帅,郑宏伟,罗格平,杨辽,王伟胜,桂东伟. 集成U-Net方法的无人机影像胡杨树冠提取和计数[J]. 遥感技术与应用, 2019, 34(5): 939-949.
[11] 李哲,张沁雨,彭道黎. 基于高分二号遥感影像的树种分类方法[J]. 遥感技术与应用, 2019, 34(5): 970-982.
[12] 刘天福,陈学泓,董琪,曹鑫,陈晋. 深度学习在GlobeLand30-2010产品分类精度优化中应用研究[J]. 遥感技术与应用, 2019, 34(4): 685-693.
[13] 周壮,李盛阳,张康,邵雨阳. 基于CNN和农作物光谱纹理特征进行作物分布制图[J]. 遥感技术与应用, 2019, 34(4): 694-703.
[14] 林志玮,丁启禄,黄嘉航,涂伟豪,胡典,刘金福. 基于DenseNet的无人机光学图像树种分类研究[J]. 遥感技术与应用, 2019, 34(4): 704-711.
[15] 崔先亮,陈立福,邢学敏,袁志辉. 基于频带特征融合的GL-CNN遥感图像场景分类[J]. 遥感技术与应用, 2019, 34(4): 712-719.