LOD-2 (Level of Detail 2) 三維建物模型已廣泛應用於電信、都市規劃、環境建模、地圖測繪、旅遊和導航等多種領域。LOD-2的獨特之處在於其詳細的表現形式，不僅描繪建築物的平面佔地面積，還納入垂直元素，如高度和屋頂構造。在精確度和逼真度要求極高的應用中，這樣的細緻度相當重要。LOD-2為進階的建物表示方式，包括準確的佔地面積和詳細的屋頂構造。這些模型包含準確的高度資訊，描繪建築物的垂直面向，並且包含表面紋理，以增加視覺真實感。這樣的詳細程度在都市規劃和視覺化專案等應用中扮演著重要的角色，可提供城市景觀的細緻描繪。
本研究使用無人飛行載具(UAV)影像產製三個不同時期資料集的LOD-2三維建物模型。首先進行攝影測量處理，由Metashape軟體產製數值表面模型DSM、數值地形模型DTM和真實正射影像。接著，透過網格計算出地物高程模型OHM，以確定場景中物體的高度。再結合深度學習和物件導向的影像分析OBIA兩種方法萃取出房屋平面圖Building Footprints。深度學習採用帶有U-Net架構的卷積神經網絡CNNs，並選擇ResNet-34作為骨幹。另一方面，OBIA利用深度學習的結果，整合OHM值和由Metashape生成的真實正射影像。這些方法的疊加應用相較於單一方法應用，結果呈現出更高的準確度。
在獲取房屋平面資訊後，後續階段涉及屋頂的分割，以分析研究區內各種類型的建物屋頂。屋頂分析的結果在後續生成LOD-2三維建物模型時發揮了關鍵作用。根據這個階段的研究結果，接續進行研究區內的變遷偵測。而變遷分析的結果顯示，在三個不同時期資料集中，多個建築物發生了明顯的改變，揭示了研究區隨著時間推移所發生的動態變化。

LOD-2 (Level of Detail 2) 3D building models have been used in various industries, including telecommunications, urban planning, environmental modeling, mapping, tourism, and mobile navigation. The distinctive feature of LOD-2 lies in its detailed representation, capturing not only the horizontal footprint of buildings but also incorporating vertical elements such as height and roof configurations. This level of detail is crucial for applications where accuracy and realism are paramount. LOD-2 represents an advanced stage of building representation, encompassing precise footprints and detailed roof configurations. The models incorporate accurate height information, capturing the buildings' vertical aspects, and include surface details for enhanced visual realism. This level of detail is vital for applications like urban planning and visualization projects, offering a nuanced depiction of urban landscapes.
In this study, LOD-2 3D building models were generated using Unmanned Aerial Vehicle (UAV) images across three distinct time series datasets. The initial step involved photogrammetric processing to derive DSM, DTM, and True Orthoimages derived by Metashape. Further, OHM was computed through raster calculations to determine the height of objects within the scene. Building footprint extraction was then performed by combining two methodologies: Deep Learning and OBIA. Deep learning procedures involved the utilization of CNNs with U-Net architecture, with ResNet-34 as the chosen backbone. OBIA, on the other hand, leveraged the outcomes of deep learning, integrating OHM values and True Orthoimages derived by Metashape. The synergistic application of these methods resulted in an enhanced level of accuracy compared to individual method applications.
After acquiring building footprint data, a subsequent phase involved the segmentation of rooftops to analyze the various types of building rooftops within the study area. The outcomes of this rooftop analysis played a pivotal role in the subsequent generation of Level of Detail 2 (LOD-2) 3D building models. Expanding upon the findings obtained in this phase of the study, a comprehensive change detection process was initiated. The results of the change detection analysis revealed discernible alterations in several buildings across the three distinct time series datasets, shedding light on the dynamic transformations occurring within the study area over time

Abbas, N., Abbas, Z., Liu, X., Khan, S.S., Foster, E.D., Larkin, S., 2023. A Survey: Future Smart Cities Based on Advance Control of Unmanned Aerial Vehicles (UAVs). Appl. Sci. 13. https://doi.org/10.3390/app13179881
Alsabhan, W., Alotaiby, T., 2022. Automatic Building Extraction on Satellite Images Using Unet and ResNet50. Comput. Intell. Neurosci. 2022. https://doi.org/10.1155/2022/5008854
Amir, A., Karim, A., Sameer, R.A., 2017. Comparing the Main Approaches of Image Segmentation. Iraqi J. Sci. 58, 2211–2221. https://doi.org/10.24996/ijs.2017.58.4b.24
Anjani, I.A., Pratiwi, Y.R., Norfa Bagas Nurhuda, S., 2021. Implementation of Deep Learning Using Convolutional Neural Network Algorithm for Classification Rose Flower. J. Phys. Conf. Ser. 1842. https://doi.org/10.1088/1742-6596/1842/1/012002
Benchabana, A., Kholladi, M.K., Bensaci, R., Khaldi, B., 2023. Building Detection in High-Resolution Remote Sensing Images by Enhancing Superpixel Segmentation and Classification Using Deep Learning Approaches. Buildings 13. https://doi.org/10.3390/buildings13071649
Biljecki, F., Ledoux, H., Stoter, J., 2016a. An improved LOD specification for 3D building models. Comput. Environ. Urban Syst. 59, 25–37. https://doi.org/10.1016/j.compenvurbsys.2016.04.005
Biljecki, F., Ledoux, H., Stoter, J., Vosselman, G., 2016b. The variants of an LOD of a 3D building model and their influence on spatial analyses. ISPRS J. Photogramm. Remote Sens. 116, 42–54. https://doi.org/10.1016/j.isprsjprs.2016.03.003
Biljecki, F., Ledoux, H., Stoter, J., Zhao, J., 2014. Formalisation of the level of detail in 3D city modelling, Computers, Environment and Urban Systems. https://doi.org/10.1016/j.compenvurbsys.2014.05.004
Biljecki, F., Stoter, J., Ledoux, H., Zlatanova, S., Çöltekin, A., 2015. Applications of 3D city models: State of the art review. ISPRS Int. J. Geo-Information 4, 2842–2889. https://doi.org/10.3390/ijgi4042842
Bisheng, Y., Qingquan, L.I., Deren, L.I., 1999. Building modeling for 3d city model. Geo-Spatial Inf. Sci. 2, 109–114. https://doi.org/10.1007/BF02826729
Boeters, R., Arroyo Ohori, K., Biljecki, F., Zlatanova, S., 2015. Automatically enhancing CityGML LOD2 models with a corresponding indoor geometry. Int. J. Geogr. Inf. Sci. 29, 2248–2268. https://doi.org/10.1080/13658816.2015.1072201
Borba, P., de Carvalho Diniz, F., da Silva, N.C., de Souza Bias, E., 2021. Building Footprint Extraction Using Deep Learning Semantic Segmentation Techniques: Experiments and Results. Int. Geosci. Remote Sens. Symp. 2021-July, 4708–4711. https://doi.org/10.1109/IGARSS47720.2021.9553855
Dong, J., Tan, J., Zhao, Q., He, L., Li, S., She, J., 2021. Structure-level 3D building model encoding method for progressive transmission. ISPRS Int. J. Geo-Information 10. https://doi.org/10.3390/ijgi10050306
Dukai, B., Ledoux, H., Stoter, J.E., 2019. A multi-height LOD1 model of all buildings in the Netherlands. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 4, 51–57. https://doi.org/10.5194/isprs-annals-IV-4-W8-51-2019
Ge, L., 2019. Generalization of LOD2 buildings with different roof structures. J. Spat. Sci. 64, 319–340. https://doi.org/10.1080/14498596.2018.1474812
George, V., Dijkman, S., 2001. 3D building model reconstruction from point clouds and ground plans. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. XXXIV, 22–24.
Hossain, M.D., Chen, D., 2019. Segmentation for Object-Based Image Analysis (OBIA): A review of algorithms and challenges from remote sensing perspective. ISPRS J. Photogramm. Remote Sens. 150, 115–134. https://doi.org/10.1016/j.isprsjprs.2019.02.009
Jiang, S., Jiang, C., Jiang, W., 2020. Efficient structure from motion for large-scale UAV images: A review and a comparison of SfM tools. ISPRS J. Photogramm. Remote Sens. 167, 230–251. https://doi.org/10.1016/j.isprsjprs.2020.04.016
Jiang, X., 2009. Feature extraction for image recognition and computer vision. Proc. - 2009 2nd IEEE Int. Conf. Comput. Sci. Inf. Technol. ICCSIT 2009 1–15. https://doi.org/10.1109/ICCSIT.2009.5235014
Kaur, J., Singh, W., 2022. Tools, techniques, datasets and application areas for object detection in an image: a review. Multimed. Tools Appl. 81, 38297–38351. https://doi.org/10.1007/s11042-022-13153-y
Labetski, A., Vitalis, S., Biljecki, F., Arroyo Ohori, K., Stoter, J., 2023. 3D building metrics for urban morphology. Int. J. Geogr. Inf. Sci. 37, 36–67. https://doi.org/10.1080/13658816.2022.2103818
Lane, S.N., Gentile, A., Goldenschue, L., 2020. Combining UAV-based SfM-MVS photogrammetry with conventional monitoring to set environmental flows: Modifying dam flushing flows to improve alpine stream habitat. Remote Sens. 12, 1–35. https://doi.org/10.3390/rs12233868
Lang, S., 2008. Object-based image analysis for remote sensing applications: Modeling reality – dealing with complexity. Lect. Notes Geoinf. Cartogr. 0, 3–27. https://doi.org/10.1007/978-3-540-77058-9_1
Lee, J., Zlatanova, S., 2008. A 3D data model and topological analyses for emergency response in urban areas. Geospatial Inf. Technol. Emerg. Response 143–167.
Monji-Azad, S., Hesser, J., Löw, N., 2023. A review of non-rigid transformations and learning-based 3D point cloud registration methods. ISPRS J. Photogramm. Remote Sens. 196, 58–72. https://doi.org/10.1016/j.isprsjprs.2022.12.023
Patil, A., Rane, M., 2021. Convolutional Neural Networks: An Overview and Its Applications in Pattern Recognition. Smart Innov. Syst. Technol. 195, 21–30. https://doi.org/10.1007/978-981-15-7078-0_3
Pepe, M., Costantino, D., Alfio, V.S., Angelini, M.G., Garofalo, A.R., 2020. A CityGML multiscale approach for the conservation and management of cultural heritage: The case study of the old town of Taranto (Italy). ISPRS Int. J. Geo-Information 9. https://doi.org/10.3390/ijgi9070449
Peters, R., Dukai, B., Vitalis, S., van Liempt, J., Stoter, J., 2022. Automated 3D Reconstruction of LoD2 and LoD1 Models for All 10 Million Buildings of the Netherlands. Photogramm. Eng. Remote Sensing 88, 165–170. https://doi.org/10.14358/PERS.21-00032R2
Prieto-Garzón, J.B., Lima-Machado, Á.M., Scheer-Hainosz, F., 2019. Detection and classification of changes in buildings by direct comparison of multi-temporal LIDAR systems data. DYNA 86, 206–214. https://doi.org/10.15446/dyna.v86n209.74248
Salau, A.O., Jain, S., 2019. Feature Extraction: A Survey of the Types, Techniques, Applications. 2019 Int. Conf. Signal Process. Commun. ICSC 2019 158–164. https://doi.org/10.1109/ICSC45622.2019.8938371
Shafiq, M., Gu, Z., 2022. Deep Residual Learning for Image Recognition: A Survey. Appl. Sci. 12, 1–43. https://doi.org/10.3390/app12188972
Shen, P., Mei, L., Ye, Z., Wang, Y., Zhang, Q., Hong, B., Yin, X., Yang, W., 2023. Building Change Detection in Remote Sensing Imagery with Focal Self-Attention and Multi-Level Feature Fusion. Electron. 12. https://doi.org/10.3390/electronics12132796
Shirinyan, E., Petrova-Antonova, D., 2022. Modeling Buildings in CityGML LOD1: Building Parts, Terrain Intersection Curve, and Address Features. ISPRS Int. J. Geo-Information 11, 1–18. https://doi.org/10.3390/ijgi11030166
Singh, S.P., 2013. VIRTUAL 3D CITY MODELING : TECHNIQUES AND APPLICATIONS XL, 27–29.
Stylianidis, E., Evangelidis, K., Vital, R., Dafiotis, P., Sylaiou, S., 2022. 3D Documentation and Visualization of Cultural Heritage Buildings through the Application of Geospatial Technologies. Heritage 5, 2818–2832. https://doi.org/10.3390/heritage5040146
Su, T., 2019. Unsupervised evaluation-based region merging for high resolution remote sensing image segmentation. GIScience Remote Sens. 56, 811–842. https://doi.org/10.1080/15481603.2019.1568679
Sun, Q., Zhou, X., Hou, D., 2020. A simplified citygml-based 3d indoor space model for indoor applications. Appl. Sci. 10, 1–22. https://doi.org/10.3390/app10207218
Wolf, P.R., Dewitt, B.A., Wilkinson, B.E., 2014. Elements of Photogrammetry with Application in GIS, Third Edit. ed. Springer International Publishing.
Zhang, C., Fan, H., Kong, G., 2021. VGI3D: an Interactive and Low-Cost Solution for 3D Building Modelling from Street-Level VGI Images. J. Geovisualization Spat. Anal. 5. https://doi.org/10.1007/s41651-021-00086-7
Zhao, W., 2019. Multitemporal SAR images denoising and change detection : applications to Sentinel-1 data.