基于深度学习的建筑识别技术在城市体检中的应用*
Exploring City Examination Using Deep Learning-based Building Detection
晏龙旭
同济大学建筑与城市规划学院 自然资源部国土空间智能规划技术重点实验室 助理教授，博士
王 勇
上海市上规院城市规划设计有限公司 工程师，硕士
张扬帆
上海同济城市规划设计研究院有限公司 工程师
张雨迪
同济大学建筑与城市规划学院
刘 骝
同济大学建筑与城市规划学院 副教授
张尚武
同济大学建筑与城市规划学院 自然资源部国土空间智能规划技术重点实验室 教授，博士生导师
王 德（通信作者）
同济大学建筑与城市规划学院 高密度人居环境生态与节能教育部重点实验室 教授，博士生导师
涂鸿昌
同济大学建筑与城市规划学院 硕士研究生
摘要: 建筑变化是城市体检的重要维度之一，但实际工作常受制于建筑数据可获取性。探索基于深度学习和高精度卫星影像
数据的建筑识别技术在城市体检中的应用。首先提出由城乡建设基本情况、政策落实与风险预警、规划实施评估等目标
构成的建筑变化视角下城市体检评估框架；然后介绍基于深度学习和高精度卫星影像的建筑识别与分析方法，并提取
2014年、2019年的建筑轮廓和高度信息；最后以上海为例，从建筑存量变化基本情况、当前城市体检关注的重点指标、上
海“十三五”规划实施情况3个方面开展评估。结果发现：“十三五”期间上海建筑存量有所增长，关于工业用地减量化
和空间布局优化的政策实施效果显著，推进新城、分类推进镇的建设等政策实施较好，中心城区建筑总量控制、城市开
发边界外工业用地减量复垦两项政策有待进一步落实。
Abstract: Building change assessment is one of the fundamental aspects of city examination, which is, however, frequently subjected to
data accessibility. This paper explores building change assessment using high-resolution satellite imagery and deep learning
models. We firstly propose an assessment framework under the perspective of buildings. Then, we introduce the high-resolution
satellite imagery and deep learning-based building extraction approach, and identify the buildings in Shanghai in 2014 and
2019. Finally, the building changes in Shanghai are evaluated from three aspects. The result shows that Shanghai's building
floor area is still growing, and policies regarding industrial land use renewal are implemented effectively, while policies with
respect to construction control in the central city and reclamation of industrial land out of the concentrated construction area
are not well put into practice.
关键词：城市体检；建筑变化；深度学习；高精度卫星影像；上海
Keyword: city examination; building changes; deep learning; high-resolution satellite imagery; Shanghai
中图分类号：TU984
文献标识码: A
资金资助
中央高校基本科研业务费专项资金 “面 向智慧共享出行的规划决策支持系统研究” 22120210541
林文棋，蔡玉蘅，李栋，等. 从城市体检到动态监
测——以上海城市体征监测为例[J]. 上海城市规
划，2019（3）：23-29.
LIN Wenqi, CAI Yuheng, LI Dong, et al. From city
medical to dynamic diagnosis: a case of Shanghai
city sign diagnosis[J]. Shanghai Urban Planning
Review, 2019(3): 23-29.
PAISITKRIANGKRAI S, SHERRAH J, JANNEY
P, et al. Semantic labeling of aerial and satellite
imagery[J]. IEEE Journal of Selected Topics in
Applied Earth Observations and Remote Sensing,
2016, 9(7): 2868-2881.
YUAN J. Learning building extraction in aerial
scenes with convolutional networks[J]. IEEE
Transactions on Pattern Analysis and Machine
Intelligence, 2017, 40(11): 2793-2798.
DIXIT M, CHAURASIA K, MISHRA V K. Dilated-
ResUnet: a novel deep learning architecture for building extraction from medium resolution multispectral
satellite imagery[J]. Expert Systems with
Applications, 2021(184): 115530.
WAGNER F H, DALAGNOL R, TARABALKA Y,
et al. U-Net-Id, an instance segmentation model for
building extraction from satellite images: case study
in the Joanópolis City, Brazil[J]. Remote Sensing,
2020, 12(10): 1544.
ZHANG Z, GUO W, LI M, et al. GIS-supervised
building extraction with label noise-adaptive fully
convolutional neural network[J]. IEEE Geoscience
and Remote Sensing Letters, 2020, 17(12): 2135-
2139.
CHEN L, ZHU Y, PAPANDREOU G, et al. Encoderdecoder
with atrous separable convolution for
semantic image segmentation[C]//Computer vision –
ECCV 2018. New York: Springer, 2018: 801-818.
AZAD R, ASADI-AGHBOLAGHI M, FATHY
M, et al. Attention Deeplabv3+: multi-level
context attention mechanism for skin lesion
segmentation[C]//European Conference on Computer
Vision. Cham: Springer, 2020: 251-266.
BAHETI B, INNANI S, GAJRE S, et al. Semantic
scene segmentation in unstructured environment
with modified DeepLabV3+[J]. Pattern Recognition
Letters, 2020(138): 223-229.
LI Z, WANG R, ZHANG W, et al. Multiscale
features supported DeepLabV3+ optimization
scheme for accurate water semantic segmentation[J].
IEEE Access, 2019(7): 155787-155804.
许泽宇，沈占锋，李杨，等. 结合模糊度和形态学
指数约束的深度学习建筑物提取[J]. 地球信息科
学学报，2021，23（5）：918-927.
XU Zeyu, SHEN Zhanfeng, LI Yang, et al. Building
extraction by deep learning method combined with
ambiguity and morphological index constraints[J].
Journal of Geo-Information Science, 2021, 23(5):
918-927.
MAGGIORI E, TARABALKA Y, CHARPIAT G,
et al. Can semantic labeling methods generalize
to any city? The INRIA aerial image labeling
benchmark[C]//2017 IEEE International Geoscience
and Remote Sensing Symposium (IGARSS). Fort
Worth, 2017: 3226-3229.
