Please wait a minute...
img

官方微信

遥感技术与应用  2020, Vol. 35 Issue (4): 741-748    DOI: 10.11873/j.issn.1004-0323.2020.4.0741
甘肃遥感学会专栏     
一种改进U-Net的高分辨率遥感影像道路提取方法
王卓1,2,3(),闫浩文1,2,3(),禄小敏1,2,3,冯天文4,5,李亚珍4,5
1.兰州交通大学测绘与地理信息学院, 甘肃 兰州 730070
2.地理国情监测技术应用国家地方联合工程研究中心, 甘肃 兰州 730070
3.甘肃省地理国情监测工程实验室, 甘肃 兰州 730070
4.中国科学院西北生态环境资源研究院,甘肃 兰州 730000
5.中国科学院大学,北京 100049
High-resolution Remote Sensing Image Road Extraction Method for Improving U-Net
Zhuo Wang1,2,3(),Haowen Yan1,2,3(),Xiaomin Lu1,2,3,Tianwen Feng4,5,Yazhen Li4,5
1.Faculty of Geomatics, Lanzhou Jiaotong University, Lanzhou 730070, China
2.National-Local Joint Engineering Research Center of Technologies and Application for National Geographic State Monitoring, Lanzhou 730070, China
3.Gansu Provincial Engineering Laboratory for National Geographic State Monitoring, Lanzhou 730070, China
4.Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China
5.University of Chinese Academy of Sciences, Beijing 100049, China
 全文: PDF(5119 KB)   HTML
摘要:

从遥感影像中准确高效地提取道路信息,对基础地理数据库的建立与维护具有重大意义。高分辨率遥感影像背景信息复杂,导致现有算法无法较好地从中提取道路信息。U-Net网络在图像分割方面有较好的实验效果,但道路分割结果准确性不佳,因此,提出了一种改进U-Net网络的高分辨率影像道路提取方法。首先,设计基于U-Net的网络结构,将VGG16作为网络编码结构,可更好地提取特征语义信息;其次,利用Batch Normalization与Dropout解决网络训练过程中出现的过拟合;最后,对训练数据利用旋转与镜像变换进行扩充,采用ELU激活函数,提升了网络训练速度。实验结果表明:该方法可以较为准确高效地提取道路信息。

关键词: 高分辨率遥感影像道路提取U-Net网络    
Abstract:

Accurate and efficient extraction of road information based on remote sensing image is a great significance for the establishment and maintenance of basic geographic databases. Due to the complex background information of high-resolution remote sensing images, existing algorithms cannot extract road information very well. U-Net network has good experimental results in image segmentation, but the accuracy of road segmentation results is not good. For this reason, this paper proposes a high-resolution image road extraction method based on improved U-Net network. Firstly, the U-Net-based network structure is designed and implemented. The network uses VGG16 as the network coding structure, which can extract feature semantic information better. Secondly, the use of Batch Normalization and Dropout solves the phenomenon of over-fitting that occurs during the network training process. Finally, the training data is expanded by rotation and mirror transformation, and the ELU activation function is used to improve the network training speed. The experimental results show that the method can extract road information more accurately and efficiently.

Key words: High resolution remote sensing image    Road extraction    U-Net
收稿日期: 2019-10-17 出版日期: 2020-09-15
ZTFLH:  TP75  
基金资助: 国家重点研发计划项目(2017YFB0504203);国家自然科学基金项目(41671447);国家青年基金项目(41801395);中国博士后科学基金(2019M653795);兰州交通大学优秀平台(201806)
通讯作者: 闫浩文     E-mail: 1078095001@qq.com;haowen2010@gmail.com
作者简介: 王卓(1994-),男,甘肃庆阳人,硕士研究生,主要从事遥感影像解译。E?mail:1078095001@qq.com
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  
王卓
闫浩文
禄小敏
冯天文
李亚珍

引用本文:

王卓,闫浩文,禄小敏,冯天文,李亚珍. 一种改进U-Net的高分辨率遥感影像道路提取方法[J]. 遥感技术与应用, 2020, 35(4): 741-748.

Zhuo Wang,Haowen Yan,Xiaomin Lu,Tianwen Feng,Yazhen Li. High-resolution Remote Sensing Image Road Extraction Method for Improving U-Net. Remote Sensing Technology and Application, 2020, 35(4): 741-748.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2020.4.0741        http://www.rsta.ac.cn/CN/Y2020/V35/I4/741

图1  U-Net网络结构
图2  VGG16网络结构
图3  ELU函数
图4  实验网络结构
参数名称参数值
学习率0.000 1
优化器Adam
损失函数binary_crossentropy
Batchsize2
Epochs35
Dropout(keep_prob)0.5
表1  实验参数
图5  数据扩充前后实验结果
区域扩充后未扩充
总体分类精度/%Kappa系数总体分类精度/%Kappa系数
A99.130.9498.400.89
B95.320.8393.590.76
表2  数据扩充前后实验
图6  道路对比实验提取结果
区域本文文献[13]U-Net[21]
总体分类精度/%Kappa系数总体分类精度/%Kappa系数总体分类精度/%Kappa系数
a98.010.9197.390.8997.160.88
b98.420.9397.840.9096.910.87
c97.970.9097.890.9097.390.88
d97.680.9197.200.9097.530.91
表3  不同算法对比实验
1 Cheng G, Wang Y, Xu S, et al. Automatic Road Detection and Centerline Extraction via Cascaded End-to-End Convolutional Neural Network[J]. IEEE Transactions on Geoscience & Remote Sensing, 2017, 55(6):3322-3337. doi: 10.1109/TGRS.2017.2669341.
doi: 10.1109/TGRS.2017.2669341
2 Cheng Jianghua, Gao Gui, Xishu Ku, et al. Summary of Road Network Extraction Methods for SAR Images [J]. Chinese Journal of Image Graphics, 2013, 18(1): 11-23.
2 程江华, 高贵, 库锡树,等. SAR图像道路网提取方法综述[J]. 中国图象图形学报, 2013, 18(1):11-23.
3 Feng Peng, Gao Feng. Summary of Road Information Extraction Methods in High-Score Remote Sensing Images [J]. Modern Electronic Technology, 2015, 38 (17): 53-57.
3 冯鹏, 高峰. 高分遥感影像中道路信息提取方法综述[J]. 现代电子技术, 2015, 38(17):53-57.
4 Zhang Yonghong, He Jing, Bingxi, et al. A Review of Road Extraction Methods from Remote Sensing Images [J]. Computer Engineering and Applications, 2018,54(13):1-10,51.
4 张永宏, 何静, 阚希,等. 遥感图像道路提取方法综述[J]. 计算机工程与应用, 2018,54(13):1-10,51.
5 Gomez R B. Hyperspectral Imaging: A Useful Technology for Transportation Analysis[J]. Optical Engineering, 2002, 41(41):2137-2143. doi: 10.1109/TGRS.2017.2669341.
doi: 10.1109/TGRS.2017.2669341
6 Luo Qingzhou, Yin Qiu, Kuang Dingbo. Research on Road Extraction Method Combining Spectrum and Shape Characteristics [J]. Remote Sensing Technology and Applications, 2007, 22 (3): 339-344.
6 罗庆洲, 尹球, 匡定波. 光谱与形状特征相结合的道路提取方法研究[J]. 遥感技术与应用, 2007, 22(3):339-344.
7 Cai Hongyue, Yao Guoqing. Optimized Method for Road Extraction from High Resolution Remote Sensing Image based on Watershed Algorithm [J]. Land and Resources Remote Sensing, 2013, 25 (3): 25-29.
7 蔡红玥, 姚国清. 基于分水岭算法的高分遥感图像道路提取优化方法[J]. 国土资源遥感, 2013, 25(3):25-29.
8 Yu Changhui, Yi Yaohua. Object-based Road Extraction in Remote Sensing Image Using Markov Random Field [J]. Journal of Wuhan University (Information Science Edition), 2011, 36 (5): 544-547.
8 余长慧, 易尧华. 利用MRF方法的高分辨率影像道路提取[J]. 武汉大学学报(信息科学版), 2011, 36(5):544-547.
9 Li M, Stein A, Bijker W, et al. Region-based Urban Road Extraction from VHR Satellite Images Using Binary Partition Tree[J]. International Journal of Applied Earth Observation & Geoinformation, 2016, 44:217-225. doi: j.jag. 2015. 09.005.
doi: j.jag. 2015. 09.005
10 Cao Yungang, Wang Zhipan, Shen Li, et al. Fusion of Pixel-based and Object-based Features for Road Centerline Extraction from High-resolution Satellite Imagery[J]. Journal of Surveying and Mapping, 2016, 45 (10): 1231-1240.
10 曹云刚, 王志盼, 慎利,等. 像元与对象特征融合的高分辨率遥感影像道路中心线提取[J]. 测绘学报, 2016, 45(10):1231-1240.
11 Das S, Mirnalinee T T, Varghese K. Use of Salient Features for the Design of a Multistage Framework to Extract Roads from High-Resolution Multispectral Satellite Images[J]. IEEE Transactions on Geoscience & Remote Sensing, 2011, 49(10):3906-3931. doi: 10.1109/TGRS.2011.2136381.
doi: 10.1109/TGRS.2011.2136381
12 Liu Xiao, Wang Guanghui, Yang Huachao, et al. Road Extraction from Remote Sensing Image based on Fully Convolutional Networks[J]. Remote Sensing Information, 2018, 33 (1): 69-75.
12 刘笑, 王光辉, 杨化超,等. 全卷积神经网络遥感影像道路提取方法[J]. 遥感信息, 2018, 33(1):69-75.
13 He Hao, Wang Shicheng, Yang Dongfang, et al. A Road Extraction Method for Remote Sensing Image based on Encoder-decoder Network [J]. Journal of Surveying and Mapping, 2019, 48(3): 330-338.
13 贺浩,王仕成,杨东方,等. 基于Encoder-Decoder网络的遥感影像道路提取方法[J].测绘学报,2019,48(3):330-338.
14 Kampffmeyer M, Salberg A B, Jenssen R. Semantic Segmentation of Small Objects and Modeling of Uncertainty in Urban Remote Sensing Images Using Deep Convolutional Neural Networks[C]∥ Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. 2016: 1-9. doi: 10.1109/CVPRW.2016.90.
doi: 10.1109/CVPRW.2016.90
15 Zhong Z, Li J, Cui W, et al. Fully Convolutional Networks for Building and Road Extraction: Preliminary Results[C]∥ 2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2016: 1591-1594. doi: 10.1109/IGARSS.2016.7729406.
doi: 10.1109/IGARSS.2016.7729406
16 Fu G, Liu C, Zhou R, et al. Classification for High Resolution Remote Sensing Imagery Using a Fully Convolutional Network[J]. Remote Sensing, 2017, 9(5): 498. doi: 10.3390/rs9050498.
doi: 10.3390/rs9050498
17 Maggiori E, Tarabalka Y, Charpiat G, et al. Convolutional Neural Networks for Large-scale Remote-sensing Image Classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 55(2): 645-657. doi: 10.1109/TGRS. 2016. 2612821.
doi: 10.1109/TGRS. 2016. 2612821
18 Kestur R, Farooq S, Abdal R, et al. UFCN: A Fully Convolutional Neural Network for Road Extraction in RGB Imagery Acquired by Remote Sensing from An Unmanned Aerial Vehicle[J]. Journal of Applied Remote Sensing, 2018, 12(1): 016020. doi: 10.1117/1.JRS.12.016020.
doi: 10.1117/1.JRS.12.016020
19 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. doi: 10.1109/LGRS.2018.2802944.
doi: 10.1109/LGRS.2018.2802944
20 Panboonyuen T, Vateekul P, Jitkajornwanich K, et al. An Enhanced Deep Convolutional Encoder-decoder Network for Road Segmentation on Aerial Imagery[C]∥ International Conference on Computing and Information Technology. Springer, Cham, 2017: 191-201. doi: 10.1007/978-3-319-60663-7_18.
doi: 10.1007/978-3-319-60663-7_18
21 Ronneberger O, Fischer P, Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation[C]∥ International Conference on Medical Image Computing & Computer-assisted Intervention, 2015,9351:234-241. doi: 10.1007/978-3-319-24574-4_28.
doi: 10.1007/978-3-319-24574-4_28
22 Wang Yu,Yang Yi,Wang Baoshan,et al. Building Segmentation in High Resolution Remote Sensing Image by Deep ResNet[J] .Remote Sensing Technology and Application,2019,34(4):736-747.
22 王宇,杨艺,王宝山, 等.深度残差神经网络高分辨率遥感图像建筑物分割[J].遥感技术与应用,2019,34(4):736-747.
23 Cui Gang,Wu Jinsheng,Yu Zhen,et al. Comparison Analysis on Wheat Mapping Using Deep Learning Algorithm from Different Satellite Data Source [J]. Remote Sensing Technology and Appli cation,2019,34(4):748755.崔刚,吴金胜,于镇,等.深度学习对不同分辨率影像冬小麦识别的适用性研究[J].遥感技术与应用,2019,34(4):748-755.
24 Zhao H S, Shi J P, Qi X J, et al. Pyramid Scene Parsing Net⁃ work[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2017:2881-2890. doi: 10.1109/CVPR.2017.660.
doi: 10.1109/CVPR.2017.660
25 Cheng G, Zhu F, Xiang S, et al. Road Centerline Extraction via Semisupervised Segmentation and Multidirection Nonmaximum Suppression[J]. IEEE Geoscience & Remote Sensing Letters, 2017, 13(4):545-549. doi: 10.1109/LGRS. 2016. 2524025.
doi: 10.1109/LGRS. 2016. 2524025
26 Simonyan K, Zisserman A. Very Deep Convolutional Networks for Large-scale Image Recognition[J]. arXiv Preprint arXiv:1409.1556, 2014.
27 Ioffe S, Szegedy C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift[J]. arXiv Preprint arXiv:1502.03167, 2015.
28 Srivastava N, Hinton G, Krizhevsky A, et al. Dropout: A Simple Way to Prevent Neural Networks from Overfitting[J]. The Journal of Machine Learning Research, 2014, 15(1): 1929-1958.
29 Djork-Arné Clevert, Unterthiner T, Hochreiter S. Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)[J]. arXiv Preprint arXiv:1511.07289, Computer Science, 2015.
[1] 连喜红,祁元,王宏伟,张金龙,杨瑞. 基于面向对象的青海湖环湖区居民地信息自动化提取[J]. 遥感技术与应用, 2020, 35(4): 775-785.
[2] 丛铭, 段晨曦, 许妙忠, 陶翊婷. 基于格式塔形状分析的高分辨率遥感影像道路提取[J]. 遥感技术与应用, 2019, 34(3): 595-601.
[3] 曹云刚,王志盼,杨磊. 高分辨率遥感影像道路提取方法研究进展[J]. 遥感技术与应用, 2017, 32(1): 20-26.
[4] 费鲜芸,王婷,魏雪丽. 基于多尺度分割的遥感影像滨海湿地分类[J]. 遥感技术与应用, 2015, 30(2): 298-303.
[5] 张勇,谷正气,刘水长,罗伦,李程. 基于高分辨率遥感影像的城市群交通路网监测技术研究[J]. 遥感技术与应用, 2015, 30(1): 76-81.
[6] 钱家航,王金亮,马如彪,邓钰. 基于QuickBird影像城市道路特征语义信息提取[J]. 遥感技术与应用, 2014, 29(4): 653-659.
[7] 顾徐博君,虞思逸,金叶蒙,周坚华 . 基于数学形态学和Hough变换的复杂形态线状地物提取方法[J]. 遥感技术与应用, 2014, 29(4): 689-694.
[8] 吴健生,林 倩,李卫锋,刘建政,彭 建. 基于特征选择的土地整理地物识别规则的动态提取[J]. 遥感技术与应用, 2013, 28(5): 799-806.
[9] 高利鹏,史文中,吕志勇,张华. 基于机载LiDAR和高分辨率遥感影像的城市道路网提取[J]. 遥感技术与应用, 2013, 28(4): 562-568.
[10] 刘 辉,谢天文. 基于PCA与HIS模型的高分辨率遥感影像阴影检测研究[J]. 遥感技术与应用, 2013, 28(1): 78-84.
[11] 张 扬,周子勇. 基于关联规则的面向对象高分辨率影像分类[J]. 遥感技术与应用, 2012, 27(3): 339-346.
[12] 周安发,周家香. 一种高分辨率遥感影像道路信息提取方法[J]. 遥感技术与应用, 2012, 27(1): 94-99.
[13] 杨先武,蚩志锋,姚高伟. 一种形态重建的遥感影像城市道路提取方法研究[J]. 遥感技术与应用, 2011, 26(6): 758-762.
[14] 张华,赵传燕,张勃,彭守璋. 高分辨率遥感影像GeoEye-1在黑河下游柽柳生物量估算中的应用 [J]. 遥感技术与应用, 2011, 26(6): 713-718.
[15] 杨康,李满春,刘永学,程亮,江冲亚. 遥感影像道路的多点同时快速行进提取方法[J]. 遥感技术与应用, 2011, 26(3): 294-302.