Please wait a minute...
 
国土资源遥感  2021, Vol. 33 Issue (2): 100-107    DOI: 10.6046/gtzyyg.2020230
  技术方法 本期目录 | 过刊浏览 | 高级检索 |
基于注意力增强全卷积神经网络的高分卫星影像建筑物提取
郭文1(), 张荞2()
1.自然资源部第三航测遥感院,成都 610100
2.西南石油大学地球科学与技术学院,成都 610500
Building extraction using high-resolution satellite imagery based on an attention enhanced full convolution neural network
GUO Wen1(), ZHANG Qiao2()
1. The Third Institute of Photogrammetry and Remote Sensing, Ministry of Natural Resources, Chengdu 610100, China
2. School of Geoscience and Technology, Southwest Petroleum University, Chengdu 610500, China
全文: PDF(8342 KB)   HTML  
输出: BibTeX | EndNote (RIS)      
摘要 

从卫星遥感影像中自动提取建筑物在国民经济社会发展中具有广泛的应用价值,由于卫星遥感影像存在地物遮挡、光照、背景环境等因素的影响,传统方法难以实现高精度建筑物提取。采用一种基于注意力增强的特征金字塔神经网络方法(FPN-SENet),利用多源高分辨率卫星影像和矢量成果数据快速构建大规模的像素级建筑物数据集(SCRS数据集),实现多源卫星影像的建筑物自动提取,并与常用的全卷积神经网络进行对比。研究结果表明: SCRS数据集的提取精度接近国际领先的卫星影像开源数据集,且假彩色数据精度高于真彩色数据; FPN-SENet的建筑物提取精度优于其他常用的全卷积神经网络; 采用交叉熵和dice系数之和为损失函数能够提升建筑物提取精度,最好的分类模型在测试数据上的分类总体精度为95.2%,Kappa系数为79.0%,F1分值和IoU分别达到了81.7%和69.1%。该研究可为高分辨率卫星影像建筑物自动提取提供参考。

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
郭文
张荞
关键词 国产高分卫星影像建筑物语义分割注意力增强    
Abstract

Automatic extraction of buildings from satellite remote sensing images has a wide range of applications in the development of economy and society. Due to the influence of mutual occlusion, illumination, background environment and other factors in satellite remote sensing images, it is difficult for traditional methods to achieve high-precision building extraction. This paper proposes an attention enhanced feature pyramid network (FPN-SENet) and constructs a large-scale pixel-wise building dataset (SCRS dataset) by using multi-source high-resolution satellite images and vector data to realize the automatic extraction of buildings from multi-source satellite images, and compares it with the other full convolution neural networks. The results show that the accuracy of building extracted from SCRS dataset is close to the world’s leading open source satellite image dataset, and the accuracy of Pseudo color data is higher than that of true color data The accuracy of FPN-SENet is better than that of other full convolution neural networks. The extraction of building can also be improved by using the sum of cross entropy and Dice coefficient as the loss function. The overall accuracy of the best classification model is 95.2%, Kappa coefficient is 79.0%, and F1-score and IoU are 81.7% and 69.1% respectively. This study can provide a reference for building automatic extraction from high-resolution satellite images.

Key wordsChinese high-resolution satellite imagery    buildings    semantic segmentation    attention enhancement
收稿日期: 2020-07-21      出版日期: 2021-07-21
ZTFLH:  TP751P237  
基金资助:四川省自然资源科研项目“基于深度注意网络的多云多雨地区土地利用精准提取方法”(KJ-2020-4);国家基础测绘科技与标准计划“信息化测绘基地建设方案设计与论证”(2018KJ0304)
通讯作者: 张荞
作者简介: 郭 文(1964-),男,高级工程师,主要从事摄影测量与遥感研究。Email: 451362006@qq.com
引用本文:   
郭文, 张荞. 基于注意力增强全卷积神经网络的高分卫星影像建筑物提取[J]. 国土资源遥感, 2021, 33(2): 100-107.
GUO Wen, ZHANG Qiao. Building extraction using high-resolution satellite imagery based on an attention enhanced full convolution neural network. Remote Sensing for Land & Resources, 2021, 33(2): 100-107.
链接本文:  
https://www.gtzyyg.com/CN/10.6046/gtzyyg.2020230      或      https://www.gtzyyg.com/CN/Y2021/V33/I2/100
Fig.1  FPN-SENet网络框架
Fig.2  SE模块单元
Fig.3  二次样条窗函数
Fig.4  卫星遥感影像建筑物样本
Fig.5  图斑综合样例
Fig.6  建筑物测试数据
数据集 OA Kappa recall precision F1/% IoU/
%
WHU 0.995 0.804 0.772 0.845 80.7 67.6
SCRS(真彩色) 0.946 0.751 0.742 0.827 78.2 64.2
SCRS(假彩色) 0.952 0.784 0.778 0.847 81.1 68.2
Tab.1  SCRS数据集与WHU数据集的比较
方法 OA Kappa recall precision F1/% IoU/
%
FCN-8s 0.932 0.672 0.639 0.800 71.0 55.1
Segnet 0.934 0.681 0.645 0.809 71.8 56.0
U-net 0.941 0.735 0.707 0.818 75.8 61.1
PSPNet 0.936 0.689 0.645 0.827 72.5 56.8
FPN-
SENet
0.946 0.751 0.742 0.827 78.2 64.2
Tab.2  不同网络模型的比较
损失函数 OA Kappa recall precision F1/% IoU/
%
L_bce 0.952 0.784 0.778 0.847 81.1 68.2
L_dice 0.948 0.779 0.833 0.786 80.9 67.9
L_bce-dice 0.952 0.790 0.82 0.814 81.7 69.1
Tab.3  不同损失函数的比较
Fig.7  平滑预测的效果
Fig.8  测试数据提取效果
[1] Blaschke T. Object based image analysis for remote sensing[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2010, 65(1):2-16.
doi: 10.1016/j.isprsjprs.2009.06.004
[2] Yang Y, Newsam S. Geographic image retrieval using local invariant features[J]. IEEE Transactions on Geoscience and Remote Sensing, 2013, 51(2):818-832.
doi: 10.1109/TGRS.2012.2205158
[3] Li E, Femiani J, Xu S B, et al. Robust rooftop extraction from visible band images using higher order CRF[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(8):4483-4495.
doi: 10.1109/TGRS.2015.2400462
[4] Melgani F, Bruzzone L. Classification of hyperspectral remote sensing images with support vector machines[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(8):1778-1790.
doi: 10.1109/TGRS.2004.831865
[5] Hinton G E, Osindero S, Teh Y. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18:1527-1554.
doi: 10.1162/neco.2006.18.7.1527
[6] Lecun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition[C]. Proceedings of the IEEE, 1998, 86(11):2278-2324.
[7] Graves A, Liwicki M, Fernandez S, et al. A novel connectionist system for unconstrained handwriting recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31:855-868.
doi: 10.1109/TPAMI.2008.137 pmid: 19299860
[8] Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[C]// International Conference on Neural Information Processing Systems, 2012.
[9] Mikolov T, Deoras A, Kombrink S, et al. Empirical evaluation and combination of advanced language modeling techniques[C]// Florence:Conference of the International Speech Communication Association, 2011.
[10] Dahl G E, Yu D, Deng L, et al. Context-dependent pre-trained deep neural networks for large-vocabulary speech recognition[J]. IEEE Transactions on Audio, Speech and Language Processing, 2012, 20(1):30-42.
doi: 10.1109/TASL.2011.2134090
[11] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]// IEEE Conference on Computer Vision and Pattern Recognition,Boston,MA,USA, 2015.
[12] Wu G M, Shao X W, Guo Z L, et al. Automatic building segmentation of aerial imagery using multi-constraint fully convolutional networks[J]. Remote Sensing, 2018, 10(3):407-424.
doi: 10.3390/rs10030407
[13] Zhang W K, Huang H, Schmitz M, et al. Effective fusion of multi-modal remote sensing data in a fully convolutional network for semantic labeling[J]. Remote Sensing, 2018, 10(52):1-14.
doi: 10.3390/rs10010001
[14] Badrinarayanan V, Kendall A, Cipolla R. Segnet:A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(12):2481-2495.
doi: 10.1109/TPAMI.34
[15] Ronneberger O, Fischer P, Brox T. U-Net:Convolutional networks for biomedical image segmentation[C]// Medical Image Computing and Computer-Assisted Intervention, 2015.
[16] Zhao H S, Shi J P, Qi X J, et al. Pyramid scene parsing network[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2017.
[17] Chen L C, Papandreou G, Kokkinos I, et al. Deeplab:Semantic image segmentation with deep convolutional nets,atrous convolution,and fully connected CRFS[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2018, 40:834-848.
[18] Mnih V. Machine learning for aerial image labeling[M]. Toronto:Toronto University of Toronto, 2013.
[19] Maggiori E, Tarabalka Y, Charpiat G, et al. Can semantic labeling methods generalize to any city? The inria aerial image labeling benchmark[C]// IEEE International Geoscience and Remote Sensing Symposium(IGARSS),Fort Worth,United States, 2017.
[20] 季顺平, 魏世清. 遥感影像建筑物提取的卷积神经元网络与开源数据集方法[J]. 测绘学报, 2019, 48(4):448-459.
Ji S P, Wei S Q. Building extraction via convolutional neural networks from an open remote sensing building dataset[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(4):448-459.
[21] Lin S Y, Doll'ar P, Girshick R, et al. Feature pyramid networks for object detection[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2017.
[22] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]// IEEE Conference on Computer Vision and Pattern Recognition, 2018.
[23] 杨建宇, 周振旭, 杜贞容, 等. 基于segnet语义模型的高分辨率遥感影像农村建设用地提取[J]. 农业工程学报, 2019, 35(5):251-258.
Yang J Y, Zhou Z X, Du Z R, et al. Rural construction land extraction from high spatial resolution remote sensing image based on segnet semantic segmentation model[J]. Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(5):251-258.
[24] 国务院第一次全国地理国情普查领导小组办公室. 地理国情普查内容与指标[M]. 北京: 测绘出版社, 2013.
The First National Geographic National Conditions Census Leading Group Office of State Council. Geographic national conditions census content and indicator[M]. Beijing: Surveying and Mapping Press, 2013.
[25] Congalton R G. A review of assessing the accuracy of classifications of remotely sensed data[J]. Remote Sensing of Environment, 1991, 37(1):35-46.
doi: 10.1016/0034-4257(91)90048-B
[26] 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, 2017, 55:645-657.
doi: 10.1109/TGRS.2016.2612821
[1] 林佳惠, 刘广, 范景辉, 赵红丽, 白世彪, 潘宏宇. 联合改进U-Net模型和D-InSAR技术采矿沉陷提取方法[J]. 自然资源遥感, 2023, 35(3): 145-152.
[2] 刘立, 董先敏, 刘娟. 顾及地学特征的遥感影像语义分割模型性能评价方法[J]. 自然资源遥感, 2023, 35(3): 80-87.
[3] 李志新, 王梦飞, 贾伟洁, 纪松, 王宇飞. 一种结合阴影信息的建筑物层数识别方法[J]. 自然资源遥感, 2023, 35(3): 97-106.
[4] 王珏, 郭振, 张志卫, 徐文学, 许昊. 基于倾斜摄影测量的三维景观指数构建——以山东田横岛为例[J]. 自然资源遥感, 2023, 35(2): 112-121.
[5] 赵凌虎, 袁希平, 甘淑, 胡琳, 丘鸣语. 改进Deeplabv3+的高分辨率遥感影像道路提取模型[J]. 自然资源遥感, 2023, 35(1): 107-114.
[6] 孟琮棠, 赵银娣, 韩文泉, 何晨阳, 陈锡秋. 基于RandLA-Net的机载激光雷达点云城市建筑物变化检测[J]. 自然资源遥感, 2022, 34(4): 113-121.
[7] 沈骏翱, 马梦婷, 宋致远, 柳汀洲, 张微. 基于深度学习语义分割模型的高分辨率遥感图像水体提取[J]. 自然资源遥感, 2022, 34(4): 129-135.
[8] 王华俊, 葛小三. 一种轻量级的DeepLabv3+遥感影像建筑物提取方法[J]. 自然资源遥感, 2022, 34(2): 128-135.
[9] 廖廓, 聂磊, 杨泽宇, 张红艳, 王艳杰, 彭继达, 党皓飞, 冷伟. 基于多维卷积神经网络的多源高分辨率卫星影像茶园分类[J]. 自然资源遥感, 2022, 34(2): 152-161.
[10] 宋仁波, 朱瑜馨, 郭仁杰, 赵鹏飞, 赵珂馨, 朱洁, 陈颖. 基于多源数据集成的城市建筑物三维建模方法[J]. 自然资源遥感, 2022, 34(1): 93-105.
[11] 武宇, 张俊, 李屹旭, 黄康钰. 基于改进U-Net的建筑物集群识别研究[J]. 国土资源遥感, 2021, 33(2): 48-54.
[12] 卢麒, 秦军, 姚雪东, 吴艳兰, 朱皓辰. 基于多层次感知网络的GF-2遥感影像建筑物提取[J]. 国土资源遥感, 2021, 33(2): 75-84.
[13] 刘钊, 赵桐, 廖斐凡, 李帅, 李海洋. 基于语义分割网络的高分遥感影像城市建成区提取方法研究与对比分析[J]. 国土资源遥感, 2021, 33(1): 45-53.
[14] 蔡祥, 李琦, 罗言, 齐建东. 面向对象结合深度学习方法的矿区地物提取[J]. 国土资源遥感, 2021, 33(1): 63-71.
[15] 仇一帆, 柴登峰. 无人工标注数据的Landsat影像云检测深度学习方法[J]. 国土资源遥感, 2021, 33(1): 102-107.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
京ICP备05055290号-2
版权所有 © 2015 《自然资源遥感》编辑部
地址:北京学院路31号中国国土资源航空物探遥感中心 邮编:100083
电话:010-62060291/62060292 E-mail:zrzyyg@163.com
本系统由北京玛格泰克科技发展有限公司设计开发