Urban Building Change Detection Using VHR Satellite Imagery and Deep Learning

Taha Kamal; Haval A Sadeq

doi:10.23918/eajse.v12i1p1

Authors

Taha Kamal Department of Geomatics (Surveying) Engineering, College of Engineering, Salahaddin University, Erbil, Kurdistan Region, Iraq https://orcid.org/0009-0008-1741-2964
Haval A Sadeq Department of Geomatics (Surveying) Engineering, College of Engineering, Salahaddin University-Erbil, Kurdistan Region, Iraq. https://orcid.org/0000-0003-0636-473X

DOI:

https://doi.org/10.23918/eajse.v12i1p1

Keywords:

Urban Building Change Detection , Very-High-Resolution Satellite Imagery , Deep Learning, U-Net, Random Forest

Abstract

Urban building change detection using very high-resolution (VHR) satellite imagery plays an important role in urban planning and monitoring, particularly in rapidly developing areas. This study aims to evaluate the effectiveness of a deep learning-based architecture for detecting urban building changes using very high-resolution satellite imagery.

The analysis focuses on two distinct areas with various types of buildings where typologies are located in Erbil Governorate, Kurdistan Region of Iraq. The dataset consists of bi-temporal satellite images that were acquired by WorldView-2 (2010) and WorldView-3 (2025) with a spatial resolution of 0.5m. The first study area (Korian) covering 0.80km², second study area (Peshang) covering 0.88km², third study area (Spanish) covering 0.61km²and the fourth study area (Zerin) covering 0.34 km². In the experiment, the U-Net-based neural network architecture with a ResNet34 encoder was implemented and trained using a sparse labeling strategy, combined with masked binary cross-entropy loss to consider the problem that arises due to class differences and imitated labelling. The model performance was evaluated against a Random Forest classifier trained on manually engineered spectral features using identical training and testing conditions. The evaluation was conducted using standard performance metrics, including accuracy, F1-score, and Intersection over Union (IoU).

The findings indicate that the deep-learning model steadily outperforms the Random Forest classifier in all study areas. Overall accuracies were shown to be above 95% with great improvements in F1-score and IoU, which indicates improved robustness against class imbalance and improved spatial delineation of changes in the building.

However, the model's performance is influenced by the limited availability of training data, which restricts the model's generalization. Furthermore, variants in the model sensor and acquisition condition between the two data sets might lead to inconsistencies in feature detection.

Overall, the study shows the potentiality of using deep learning in changed detection for accurate and scalable urban growth monitoring, while emphasizing on focusing the requirement for larger training and more diverse datasets across different areas for the purpose of generalizing the model.

Downloads

Download data is not yet available.

References

[1] Hafner S, Fang H, Azizpour H, Ban Y. Continuous urban change detection from satellite image time series with temporal feature refinement and multi-task integration. IEEE Transactions on Geoscience and Remote Sensing. 2025 Jun 11. https://doi.org/10.48550/arXiv.2406.17458

[2] Jiang H, Peng M, Zhong Y, Xie H, Hao Z, Lin J, Ma X, Hu X. A survey on deep learning-based change detection from high-resolution remote sensing images. Remote Sensing. 2022 Mar 23;14(7):1552. https://doi.org/10.3390/rs14071552

[3] Asokan A, Anitha JJ. Change detection techniques for remote sensing applications: A survey. Earth Science Informatics. 2019 Jun 1;12(2):143-60. https://doi.org/10.1007/s12145-019-00380-5

[4] Rynkiewicz A, Hościło A, Aune-Lundberg L, Nilsen AB, Lewandowska A. Detection and quantification of vegetation losses with Sentinel-2 images using bi-temporal analysis of spectral indices and a transferable random forest model. Remote Sensing. 2025 Mar 11;17(6):979. https://doi.org/10.3390/rs17060979

[5] Shafique A, Cao G, Khan Z, Asad M, Aslam M. Deep learning-based change detection in remote sensing images: A review. Remote Sensing. 2022 Feb 11;14(4):871. https://doi.org/10.3390/rs14040871

[6] Cohen WB, Healey SP, Yang Z, Zhu Z, Gorelick N. Diversity of algorithm and spectral band inputs improves Landsat monitoring of forest disturbance. Remote Sensing. 2020 May 23;12(10):1673. https://doi.org/10.3390/rs12101673

[7] Daudt RC, Le Saux B, Boulch A, Gousseau Y. Urban change detection for multispectral earth observation using convolutional neural networks. InIGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium 2018 Jul 22 (pp. 2115-2118). Ieee. https://doi.org/10.48550/arXiv.1810.08468

[8] Basavaraju KS, Sravya N, Lal S, Nalini J, Reddy CS, Dell’Acqua F. UCDNet: A deep learning model for urban change detection from bi-temporal multispectral Sentinel-2 satellite images. IEEE Transactions on Geoscience and Remote Sensing. 2022 Mar 23; 60:1-0. https://doi.org/10.1109/TGRS.2022.3161337

[9] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical Image Computing and Computer-Assisted Intervention 2015 Oct 5 (pp. 234-241). Cham: Springer International Publishing. https://doi.org/10.1007/978-3-319-24574-4_28

[10] Alsabhan W, Alotaiby T, Dudin B. Detecting Buildings and Nonbuildings from Satellite Images Using U‐Net. Computational Intelligence and Neuroscience. 2022;2022(1):4831223. https://doi.org/10.1155/2022/4831223

[11] He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2016 (pp. 770-778). https://doi.org/10.1109/CVPR.2016.90

[12] Khankeshizadeh E, Mohammadzadeh A, Mohsenifar A, Moghimi A, Pirasteh S, Feng S, Hu K, Li J. Building detection in VHR remote sensing images using a novel dual attention residual-based U-Net (DAttResU-Net): An application to generating building change maps. Remote Sensing Applications: Society and Environment. 2024 Nov 1; 36:101336. https://doi.org/10.1016/j.rsase.2024.101336

[13] Maxar Technologies. WorldView satellite constellation [Internet]. Westminster (CO): Maxar Technologies; c2024 [cited 2025 Jun 9]. Available from: https://www.maxar.com/constellation

[14] Rizeei HM, Pradhan B. Urban mapping accuracy enhancement in high-rise built-up areas deployed by 3D-orthorectification correction from WorldView-3 and LiDAR imagery. Remote Sensing. 2019 Mar 22;11(6):692. https://doi.org/10.3390/rs11060692

[15] Alaska Satellite Facility. SAR data archive [Internet]. Fairbanks (AK): Alaska Satellite Facility; c2015 [cited 2025 Jun 9]. Available from: https://asf.alaska.edu

[16] Alganci U, Besol B, Sertel E. Accuracy assessment of different digital surface models. ISPRS International Journal of Geo-Information. 2018 Mar 15;7(3):114. https://doi.org/10.3390/ijgi7030114

[17] Abdullah SH, Sadeq HA, Salih DM. 3D Buildings Change Detection from Aerial and Satellite Stereo Imagery Using the Kullback–Leibler Divergence Algorithm. Wasit Journal of Engineering Sciences. 2022;10(4):13-26. https://doi.org/10.31185/ejuow.Vol10.Iss4.454

[18] Lu D, Mausel P, Brondizio E, Moran E. Change detection techniques. International journal of remote sensing. 2004 Jun 1;25(12):2365-401. https://doi.org/10.1080/0143116031000139863

[19] Myint SW, Gober P, Brazel A, Grossman-Clarke S, Weng Q. Per-pixel vs. object-based classification of urban land cover extraction using high spatial resolution imagery. Remote sensing of the environment. 2011 May 15;115(5):1145-61. https://doi.org/10.1016/j.rse.2010.12.017

[20] Salehi B, Zhang Y, Zhong M, Dey V. Object-based classification of urban areas using VHR imagery and height points ancillary data. Remote Sensing. 2012 Aug 3;4(8):2256-76. https://doi.org/10.3390/rs4082256

[21] Hussain M, Chen D, Cheng A, Wei H, Stanley D. Change detection from remotely sensed images: From pixel-based to object-based approaches. ISPRS Journal of photogrammetry and remote sensing. 2013 Jun 1; 80:91-106. https://doi.org/10.1016/j.isprsjprs.2013.03.006

[22] Chen H, Shi Z. A spatial-temporal attention-based method and a new dataset for remote sensing image change detection. Remote sensing. 2020 May 22;12(10):1662. https://doi.org/10.3390/rs12101662

[23] Zhang Z, Liu Q, Wang Y. Road extraction by deep residual U-Net. IEEE Geoscience and Remote Sensing Letters. 2018 Mar 8;15(5):749-53. https://doi.org/10.1109/LGRS.2018.2802944

[24] Chen H, Qi Z, Shi Z. Remote sensing image change detection with transformers. IEEE Transactions on Geoscience and Remote Sensing. 2021 Jul 20; 60:1-4. https://doi.org/10.1109/TGRS.2021.3095166

[25] Fang S, Li K, Shao J, Li Z. SNUNet-CD: A densely connected Siamese network for change detection of VHR images. IEEE Geoscience and Remote Sensing Letters. 2021 Feb 17; 19:1-5. https://doi.org/10.1109/LGRS.2021.3056416

[26] Wang J, Lin T, Zhang C, Peng J. FIBTNet: Building Change Detection for Remote Sensing Images Using Feature Interactive Bi-Temporal Network. Computers, Materials & Continua. 2024 Sep 1;80(3). https://doi.org/10.32604/cmc.2024.053206

[27] Feng W, Guan F, Tu J, Sun C, Xu W. Detection of changes in buildings in remote sensing images via self-supervised contrastive pre-training and historical geographic information system vector maps. Remote Sensing. 2023 Dec 8;15(24):5670. https://doi.org/10.3390/rs15245670

[28] Li J, He W, Li Z, Guo Y, Zhang H. Overcoming the uncertainty challenges in detecting building changes from remote sensing images. ISPRS Journal of Photogrammetry and Remote Sensing. 2025 Feb 1; 220:1-7. https://doi.org/10.1016/j.isprsjprs.2024.11.017

[29] Guo H, Du B, Zhang L, Su X. A coarse-to-fine boundary refinement network for building footprint extraction from remote sensing imagery. ISPRS Journal of Photogrammetry and Remote Sensing. 2022 Jan 1; 183:240-52. https://doi.org/10.1016/j.isprsjprs.2021.11.005

[30] Lin H, Hao M, Luo W, Yu H, Zheng N. BEARNet: A novel building edge-aware refined network for building extraction from high-resolution remote sensing images. IEEE Geoscience and Remote Sensing Letters. 2023 May 2; 20:1-5. https://doi.org/10.1109/LGRS.2023.3272353

[31] Kaur G, Afaq Y. Developments in deep learning for change detection in remote sensing: A review. Transactions in GIS. 2024 Apr;28(2):223-57. https://doi.org/10.1111/tgis.13133

[32] Cheng G, Huang Y, Li X, Lyu S, Xu Z, Zhao H, Zhao Q, Xiang S. Change detection methods for remote sensing in the last decade: A comprehensive review. Remote Sensing. 2024 Jun 27;16(13):2355. https://doi.org/10.3390/rs16132355

[33] Brodersen KH, Ong CS, Stephan KE, Buhmann JM. The balanced accuracy and its posterior distribution. In2010 20th international conference on pattern recognition 2010 Aug 23 (pp. 3121-3124). IEEE. https://doi.org/10.1109/ICPR.2010.764