Automatic reconstruction method of 3D white models of buildings based on monocular remote sensing imagery

doi:10.13474/j.cnki.11-2246.2026.0319

Abstract

Abstract: With the development of real-world 3D,digital twin cities and smart cities,automatic building extraction and 3D modeling are of great significance in urban planning and visualization.In order to solve the problem that the existing methods rely on DSM or manual operation,a method of automatic reconstruction of three-dimensional white model of buildings based on monocular remote sensing images is proposed.In this method,which only relies on a single high-resolution satellite image,the high-precision segmentation of the building footprint and the roof is achieved through the improved U-Net model that fuses the CBAM attention mechanism and the SMU activation function,and then the building elevation is estimated by using the geometric relationship between the roof offset and the satellite image.Finally,the CGA modeling language is used to generate a 3D white model model of the building.Experimental results show that the overall accuracy of the improved model is 96.46%and 98.4%,and the mIoU is 90.93%and 92.15%,respectively,on the UBC and WHU datasets.In the elevation estimation experiment of 45 buildings,the average absolute error was 1.061 8 m,and 93.33%of the buildings had an error of less than 2 m.It verifies the feasibility and engineering application value of the method.

Key words: 3D building reconstruction, remote sensing image segmentation, footprint extraction, U-Net

CLC Number:

P208

QIU Zhiwei, QIN Jing, XIANG Qianjin, WANG Chenxi. Automatic reconstruction method of 3D white models of buildings based on monocular remote sensing imagery[J]. Bulletin of Surveying and Mapping, 2026, 0(3): 112-117,129.

References

[1] RONNEBERGER O,FISCHER P,BROX T.U-Net: convolutional networks for biomedical image segmentation[M]//Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015.Cham: Springer International Publishing,2015:234-241.
[2] 张永军,张祖勋,龚健雅.天空地多源遥感数据的广义摄影测量学[J].测绘学报,2021,50(1): 1-11.
[3] 龚健雅,季顺平.从摄影测量到计算机视觉[J].武汉大学学报(信息科学版),2017,42(11): 1518-1522,1615.
[4] QI C R,YI Li,SU Hao,et al.PointNet++:deep hierar-chical feature learning on point sets in a metric space[C]//Proceeding of the 31st International Conference on Neural Information Processing Systems.New York:ACM Press,2017:5105-5114.
[5] 张林杰,黄筱,饶维冬,等.网络RTK和PPK辅助水利工程免像控无人机倾斜摄影测量三维建模分析[J].测绘通报,2023(4): 115-120.
[6] XU Shibiao,PAN Xingjia,LI Er,et al.Automatic building rooftop extraction from aerial images via hierarchical RGB-D priors[J].IEEE Transactions on Geoscience and Remote Sensing,2018,56(12): 7369-7387.
[7] 朱梓萌,李少丹,郑东博,等.RH-CUnet: 嵌入边缘和角点的传统村落建筑物提取[J].遥感信息,2024,39(4): 166-173.
[8] 张越,程春泉,杨书成,等.融合双注意力机制模型的遥感影像建筑物提取[J].测绘科学,2022,47(4): 129-136.
[9] 周小迦.基于CityEngine的批量化三维建模方法研究[J].测绘与空间地理信息,2024,47(4): 214-216.
[10] 崔珂,田启川,廉露.基于U-Net变体的医学图像分割算法综述[J].计算机工程与应用,2024,60(11): 32-49.
[11] 邢素霞,李珂娴,方俊泽,等.深度学习下的医学图像分割综述[J].计算机工程与应用,2025,61(7): 25-41.
[12] 金澍,关沫,边玉婵,等.基于改进U-Net的遥感影像建筑物提取方法[J].激光与光电子学进展,2023,60(4): 59-65.
[13] 刘英,何雪,李单阳,等.注意力引导全尺度连接网络的高分辨率影像变化检测[J].遥感学报,2024,28(4): 1052-1065.
[14] HUANG Xingliang,REN Libo,LIU Chenglong,et al.Urban building classification(UBC)-a dataset for individual building detection and classification from satellite imagery[C]//Proceeding of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).New Orleans: IEEE,2022: 1412-1420.
[15] NOH H,HONG S,HAN B.Learning deconvolution network for semantic segmentation[C]//Proceeding of 2015 IEEE International Conference on Computer Vision (ICCV).Santiago: IEEE,2016: 1520-1528.
[16] SUN Ke,XIAO Bin,LIU Dong,et al.Deep high-resolution representation learning for human pose estimation[C]//Proceeding of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Long Beach: IEEE,2019: 5686-5696.
[17] AHMADIAN N,SEDAGHAT A,MOHAMMADI N.Building footprint extraction from remote sensing images with residual attention multi-scale aggregation fully convolutional network[J].Journal of the Indian Society of Remote Sensing,2024,52(11): 2417-2429.
[18] XIE Enze,WANG Wenhai,YU Zhiding,et al.SegFormer: simple and efficient design for semantic segmentation with transformers[EB/OL].[2025-04-12].https://arxiv.org/abs/2105.15203.
[19] LIU Yue,TIAN Yunjie,ZHAO Yuzhong,et al.VMamba: visual state space model[EB/OL].[2025-04-13].https://arxiv.org/abs/2401.10166.
[20] ZHU Enze,CHEN Zhan,WANG Dingkai,et al.UNetMamba: an efficient UNet-like mamba for semantic segmentation of high-resolution remote sensing images[J].IEEE Geoscience and Remote Sensing Letters,2025,22: 6001205.
[21] CHE Yangzi,LI Xuecao,LIU Xiaoping,et al.3D-GloBFP: the first global three-dimensional building footprint dataset[J].Earth System Science Data,2024,16(11): 5357-5374.