基于PRCUnet的高分遥感影像建筑物提取

doi:10.12082/dqxxkx.2021.210283

地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (10): 1838-1849.doi: 10.12082/dqxxkx.2021.210283

基于PRCUnet的高分遥感影像建筑物提取

徐佳伟¹(), 刘伟²^,^*(), 单浩宇¹, 史嘉诚¹, 李二珠¹, 张连蓬¹, 李行¹

1.江苏师范大学地理测绘与城乡规划学院,徐州 221116
2.中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室,北京 100101

收稿日期:2021-05-21 修回日期:2021-07-09 出版日期:2021-10-25 发布日期:2021-12-25
通讯作者: * 刘伟（1983— ）,男,安徽宿州人,副教授,主要从事空间数据质量检查、遥感图像处理以及GIS开发与应用研究。E-mail: liuw@jsnu.edu.cn
作者简介:徐佳伟（1997— ）,男,江苏南京人,硕士生,主要从事深度学习、遥感图像分析处理研究。E-mail: xujiawei@jsnu.edu.cn
基金资助:
江苏省研究生科研与实践创新计划项目(KYCX20_2364);江苏高校优势学科建设工程资助项目;徐州市重点研发计划（社会发展）项目(KC20172);徐州市重点研发计划（社会发展）项目(KC20172);资源与环境信息系统国家重点实验室开放基金项目;江苏省自然资源厅科技创新项目(2021046);江苏省地质矿产勘查局科研项目(2020KY11)

High-Resolution Remote Sensing Image Building Extraction based on PRCUnet

XU Jiawei¹(), LIU Wei²^,^*(), SHAN Haoyu¹, SHI Jiacheng¹, LI Erzhu¹, ZHANG Lianpeng¹, LI Xing¹

1. School of Geographic Mapping and Urban Rural Planning, Jiangsu Normal University, Xuzhou 221116, China
2. State Key Laboratory of Resources and Environmental Information Systems, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China

Received:2021-05-21 Revised:2021-07-09 Online:2021-10-25 Published:2021-12-25
Supported by:
Postgraduate Research & Practice Innovation Program of Jiangsu Province(KYCX20_2364);A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions;Xuzhou Science and Technology Key R&D Program (Social Development) under Project(KC20172);Xuzhou Science and Technology Key R&D Program(KC20172);Open Fund of National Key Laboratory of Resource and Environment Information System;Jiangsu Province Land and Resources Science and Technology Plan Project(2021046);Jiangsu Geology&Mineral Exploration Bureau Science and Technology Plan Project(2020KY11)

摘要/Abstract

摘要：

基于高分辨率遥感影像的建筑物提取具有重要的理论与实际应用价值,深度学习因其优异的深层特征提取能力,已经成为高分影像提取建筑物的主流方法之一。本文在改进深度学习网络结构的基础上,结合最小外接矩形与Hausdorff距离概念,对建筑物提取方法进行改进。本文主要改进内容为：① 基于Unet网络结构,利用金字塔池化模块 (Pyramid Pooling Module, PPM )的多尺度场景解析特点,残差模块(Residual Block, RB)的特征提取能力以及卷积块注意力模块(Convolutional Block Attention Module, CBAM)对空间信息和通道信息的平衡能力。将金字塔池化、残差结构以及卷积块注意力模块引入到Unet模型中,建立PRCUnet模型。PRCUnet模型更关注语义信息和细节信息,弥补Unet对小目标检测的欠缺;② 基于最小外接矩形与Hausdorff距离,改进建筑物轮廓优化算法,提高模型的泛化能力。实验表明,本文的建筑物提取方法在测试集上准确率、IoU、召回率均达到0.85以上,精度显著优于Unet模型,提取出的建筑物精度更高,对小尺度及不规则的建筑物有较好的提取效果,优化后的建筑物轮廓更接近真实的建筑物边界。

关键词: 深度卷积神经网络, 高分辨率遥感影像, 建筑物提取, Unet, 池化金字塔, 残差路径, 卷积块注意力机制, 建筑物轮廓优化

Abstract:

Building extraction based on high-resolution remote sensing images has important theoretical and practical applications. Deep learning has become one of the mainstream methods for extracting buildings from high-resolution images because of its excellent deep feature extraction ability. In this paper, based on an improved structure of deep learning network, we combined the concept of minimum outer rectangle and Hausdorff distance to improve the building extraction method. The main improvements in this paper are: ① Based on the Unet network structure, we employed the multi-scale feature detection ability of Pyramid Pooling Module (PPM), the great feature extraction capability of Residual Block (RB), and the ability to balance spatial and channel information of Convolutional Block Attention Module (CBAM). The PPM, RB, and CBAM were introduced to the Unet model to build the PRCUnet model, which focuses more on semantic and detailed information and overcomes the limitation of Unet in small target detection; ② We improved the building contour optimization algorithm based on the minimum outer rectangle and Hausdorff distance to improve the generalization ability of the model. Experiments show that the accuracy, IoU, and recall of the building extraction method proposed in this paper reached above 0.85 using the test set, significantly higher than those of the Unet model. The PRCUnet model also had better extraction effect on small-scale and irregular buildings than Unet, and the optimized building contours were close to the real building boundaries.

Key words: deep convolution neural network, high-resolution remote sensing image, building extraction, unet, pyramid pooling, residual block, convolutional block attention module, building contour optimization

徐佳伟, 刘伟, 单浩宇, 史嘉诚, 李二珠, 张连蓬, 李行. 基于PRCUnet的高分遥感影像建筑物提取[J]. 地球信息科学学报, 2021, 23(10): 1838-1849.DOI:10.12082/dqxxkx.2021.210283

XU Jiawei, LIU Wei, SHAN Haoyu, SHI Jiacheng, LI Erzhu, ZHANG Lianpeng, LI Xing. High-Resolution Remote Sensing Image Building Extraction based on PRCUnet[J]. Journal of Geo-information Science, 2021, 23(10): 1838-1849.DOI:10.12082/dqxxkx.2021.210283

图/表 14

图1

图2

图3

图4

图5

图6

图7

图8

表1

表2

图9

图10

图11

表3

参考文献 31

[1]	林祥国, 张继贤. 面向对象的形态学建筑物指数及其高分辨率遥感影像建筑物提取应用[J]. 测绘学报, 2017, 46(6):724-733.
	[ Lin X G, Zhang J X. Object-based morphological building index for building extraction from high resolution remote sensing imagery[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(6):724-733. ]
[2]	Chen Q, Wang L, Waslander S L, et al. An end-to-end shape modeling framework for vectorized building outline generation from aerial images[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 170:114-126. doi: 10.1016/j.isprsjprs.2020.10.008
[3]	Cheriyadat A M. Unsupervised feature learning for aerial scene classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2013, 52(1):439-451. doi: 10.1109/TGRS.2013.2241444
[4]	Huang J F, Zhang X C, Xin Q C, et al. Automatic building extraction from high-resolution aerial images and LiDAR data using gated residual refinement network[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2019, 151:91-105. doi: 10.1016/j.isprsjprs.2019.02.019
[5]	Zhang L P, Zhang L F, Du B. Deep learning for remote sensing data: A technical tutorial on the state of the art[J]. IEEE Geoscience and Remote Sensing Magazine, 2016Jun 7; 4(2):22-40. doi: 10.1109/MGRS.2016.2540798
[6]	Tong X Y, Xia G S, Lu Q K, et al. Land-cover classification with high-resolution remote sensing images using transferable deep models[J]. Remote Sensing of Environment, 2020, 237(11):13-22.
[7]	Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[EB/OL]. 2014: arXiv: 1409.1556. https://arxiv.org/abs/1409.1556
[8]	胡翔云, 巩晓雅, 张觅. 变分法遥感影像人工地物自动检测[J]. 测绘学报, 2018, 47(6):780-789.
	[ Hu X Y, Gong X Y, Zhang M. A variational approach for automatic man-made object detection from remote sensing images[J]. Acta Geodaetica et Cartographica Sinic, 2018, 47(6):780-789. ]
[9]	范荣双, 陈洋, 徐启恒, 等. 基于深度学习的高分辨率遥感影像建筑物提取方法[J]. 测绘学报, 2019, 48(1):38-45.
	[ Fan R S, Chen Y, Xu Q H, et al. A high-resolution remote sensing image building extraction method based on deep learning[J]. Acta Geodaetica et Cartographica Sinic, 2019, 48(1):38-45. ]
[10]	龚健雅, 季顺平. 摄影测量与深度学习[J]. 测绘学报, 2018, 47(6):693-704.
	[ Gong J Y, Ji S P. Photogrammetry and deep learning[J]. Acta Geodaetica et Cartographica Sinica, 2018, 47(6):693-704. ]
[11]	Mnih V. Machine learning for aerial image labeling[D]. University of Toronto (Canada), 2013.
[12]	Saito S, Yamashita Y, Aoki Y. Multiple object extraction from aerial imagery with convolutional neural networks[J]. Journal of Imaging Science & Technology, 2016, 60(1):10402.
[13]	Maggiori E, Tarabalka Y, Charpiat G, et al. High-resolution aerial image labeling with convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12):7092-7103. doi: 10.1109/TGRS.2017.2740362
[14]	Jonathan L, Evan S, Trevor D. Fully convolutional networks for semantic segmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 39(4):640-651. doi: 10.1109/TPAMI.2016.2572683
[15]	Ronneberger O, Fischer P, Brox T, et al. U-net: Convolutional networks for biomedical image segmentation[C]//International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015:234-241.
[16]	Guo M Q, Liu H, Xu Y Y, et al. Building extraction based on U-Net with an attention block and multiple losses[J]. Remote Sensing, 2020, 12(9):1400. doi: 10.3390/rs12091400
[17]	Kang W C, Xiang Y M, Wang F, et al. EU-Net: An efficient fully convolutional network for building extraction from optical remote sensing images[J]. Remote Sensing, 2019, 11(23):2813. doi: 10.3390/rs11232813
[18]	Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]// Proceedings of the IEEE international conference on computer vision, 2017:2980-2988.
[19]	Ye Z R, Fu Y Y, Gan M Y, et al. Building extraction from very high resolution aerial imagery using joint attention deep neural network[J]. Remote Sensing, 2019, 11:2970-2990. doi: 10.3390/rs11242970
[20]	Liu W, Xu J W, Guo Z H, et al. Building Footprint Extraction From Unmanned Aerial Vehicle Images Via PRU-Net:Application to Change Detection[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:2236-2248. doi: 10.1109/JSTARS.4609443
[21]	Glorot X, Bordes A, Bengio Y. Deep Sparse Rectifier Neural Networks[C]// Proceedings of the 14th International Conference on Artificial Intelligence and Statistics (AISTATS), 2011:315-323.
[22]	Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module[C]// Proceedings of the European conference on computer vision (ECCV), 2018:3-19.
[23]	He K M, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE conference on computer vision and pattern recognition, 2016:770-778.
[24]	Ibtehaz N, Rahman M S. MultiResUNet: Rethinking the U-Net architecture for multimodal biomedical image segmentation[J]. Neural Networks, 2020, 121:74-87. doi: S0893-6080(19)30250-3 pmid: 31536901
[25]	Liu Y B, Zhang Z X, Zhong R F, et al. Multilevel building detection framework in remote sensing images based on convolutional neural networks[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11(10):3688-3700. doi: 10.1109/JSTARS.4609443
[26]	Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[EB/OL]. 2017: arXiv: 1706.05587. https://arxiv.org/abs/1706.05587
[27]	He K, Zhang X, Ren S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9):1904-1916. doi: 10.1109/TPAMI.2015.2389824
[28]	Xie C W, Zhou H Y, Wu J X. Vortex pooling: Improving context representation in semantic segmentation[J]. 2018: arXiv: 1804.06242, https://arxiv.org/abs/1804.06242
[29]	柳娜, 孙晓亮, 谭毅华. 基于最小外接矩形的直角多边形拟合算法[J]. 计算机科学, 2017, 44(6):294-297,305.
	[ Liu N, Sun X L, Tan Y H. Polygon fitting algorithm based on minimum bounding rectangle[J]. Computer Science, 2017, 44(6):294-297,305. ]
[30]	常京新, 王双喜, 杨元维, 等. 高分遥感影像建筑物轮廓的逐级优化方法[J]. 中国激光, 2020, 47(10):249-262.
	[ Chang J X, Wang S X, Yang Y W, et al. Hierarchical optimization method of building contour in high-resolution remote sensing images[J]. Chinese Journal of Lasers, 2020, 47(10):249-262. ]
[31]	住房和城乡建设部信息中心. 全国试点地区房屋建筑和市政设施调查底图制备项目公开招标公告[A/OL]. http://www.ccgp.gov.cn/cggg/zygg/gkzb/202009/t20200929_15168096.html , 2020-10-20.
	[ Information Center of Ministry of Housing and Urban-Rural Development. National pilot area housing construction and municipal facilities survey base map preparation project public tender announcement[A/OL]. http://www.ccgp.gov.cn/cggg/zygg/gkzb/202009/t20200929_15168096.html , 2020-10-20. ]

	Unet	Unet+ResBlock	Unet+ResBlock+PPM	PRCUnet
迭代次数	61	63	63	58
学习率	1e-4	1e-4	1e-4	1e-4
批次大小	16	16	16	16
图像增强	是	是	是	是

模型	Accuracy	IoU	Recall
Unet	0.873	0.753	0.801
Unet+RB	0.886	0.774	0.825
Unet+RB+PPM	0.905	0.821	0.853
PRCUnet	0.921	0.851	0.877
PRCUnet+轮廓优化	-	0.882	-

基于PRCUnet的高分遥感影像建筑物提取

High-Resolution Remote Sensing Image Building Extraction based on PRCUnet

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 31

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	唐璎, 刘正军, 杨懿, 顾海燕, 杨树文. 基于特征增强和ELU的神经网络建筑物提取研究[J]. 地球信息科学学报, 2021, 23(4): 692-709.
[2]	郭紫甜, 王春梅, 刘欣, 庞国伟, 朱梦阳, 王晋卿. 基于小流域抽样单元的中国FROM-GLC30数据精度评价[J]. 地球信息科学学报, 2021, 23(3): 524-535.
[3]	郭峰, 毛政元, 邹为彬, 翁谦. 融合LiDAR数据与高分影像特征信息的建筑物提取方法[J]. 地球信息科学学报, 2020, 22(8): 1654-1665.
[4]	蔡博文,王树根,王磊,邵振峰. 基于深度学习模型的城市高分辨率遥感影像不透水面提取[J]. 地球信息科学学报, 2019, 21(9): 1420-1429.
[5]	李泽宇,明冬萍,范莹琳,赵林峰,刘思民. 遥感影像监督分割评价指标比较与分析[J]. 地球信息科学学报, 2019, 21(8): 1265-1274.
[6]	刘懿兰, 黄晓霞, 李红旮, 柳泽, 陈崇, 王新歌. 基于卷积神经网络与条件随机场方法提取乡镇非正规固体废弃物[J]. 地球信息科学学报, 2019, 21(2): 259-268.
[7]	刘浩, 骆剑承, 黄波, 杨海平, 胡晓东, 徐楠, 夏列钢. 基于特征压缩激活Unet网络的建筑物提取[J]. 地球信息科学学报, 2019, 21(11): 1779-1789.
[8]	罗星, 徐伟铭, 王佳. 基于对象BOW特征的高分辨率遥感影像变化检测方法[J]. 地球信息科学学报, 2018, 20(8): 1150-1159.
[9]	邵磊, 董广军, 于英, 姚强强, 张阿龙. 车载激光扫描数据中建筑物立面快速提取[J]. 地球信息科学学报, 2018, 20(4): 462-470.
[10]	王成军, 毛政元, 徐伟铭, 翁谦. 超像素与主动学习相结合的遥感影像变化检测方法[J]. 地球信息科学学报, 2018, 20(2): 235-245.
[11]	曲畅, 任玉环, 刘亚岚, 李娅. POI辅助下的高分辨率遥感影像城市建筑物功能分类研究[J]. 地球信息科学学报, 2017, 19(6): 831-837.
[12]	苏腾飞, 张圣微, 李洪玉. 基于优化合并的高分辨率遥感影像分割算法[J]. 地球信息科学学报, 2016, 18(7): 931-940.
[13]	吴田军, 骆剑承, 沈占锋, 夏列钢, 徐楠. 高分辨率遥感土地覆盖分类技术的应用研究——以重庆市黔江贫困区为例[J]. 地球信息科学学报, 2016, 18(3): 353-361.
[14]	施文灶, 毛政元. 基于图分割的高分辨率遥感影像建筑物变化检测研究[J]. 地球信息科学学报, 2016, 18(3): 423-432.
[15]	杜欣, 黄晓霞, 李红旮, 沈利强. 基于投影寻踪学习网络算法的植物群落高分遥感分类研究[J]. 地球信息科学学报, 2016, 18(1): 124-132.