Pemodelan Bangunan 3D Menggunakan Footprint Bangunan Hasil Ekstraksi Mask R-CNN Dan Dense Point Cloud Dari Foto Udara UAV

Open Access
Article Info
Submitted: 2021-12-09
Published:
Section: Articles
Language: ID
In Vol 1 (2021): PROSIDING FORUM ILMIAH TAHUNAN (FIT)-IKATAN SURVEYOR INDONESIA (ISI) 2021

Bangunan merupakan salah satu objek penting yang secara spasial dibutuhkan dalam berbagai pekerjaan khususnya untuk perencanaan dan pembangunan kota. Bangunan dalam representasi 3D telah terbukti mampu menunjang kegiatan perencanaan dengan baik mengingat dunia nyata berada dalam sistem 3D. Salah satu metode yang paling sederhana untuk membuat model bangunan 3D dalam cakupan wilayah yang luas adalah dengan melakukan ekstrusi footprint bangunan. Data yang umum digunakan dalam metode ini adalah footprint bangunan hasil digitasi manual pada orthomosaic dan komponen elevasi berupa point cloud dari Light Detection and Ranging (LiDAR). Pekerjaan digitasi manual umumnya memakan waktu yang relatif lama dan sumber daya manusia yang cenderung tinggi apabila data yang diproses semakin besar, selain itu hasil digitasi juga tidak konsisten relatif kepada keterampilan operator. Disisi lain, penggunaan point cloud LiDAR menyebabkan metode ini kurang terjangkau dari sisi biaya. Dalam penelitian ini, dilakukan pemodelan bangunan 3D menggunakan footprint bangunan yang dihasilkan secara otomatis dengan teknik Mask Region-based Convolutional Neural Network (Mask R-CNN) dan dense point cloud yang diperoleh dari pengolahan foto udara di kawasan kampus pusat Universitas Riau yang diakuisisi menggunakan Unmanned Aerial Vehicle (UAV). Metode yang diterapkan memberikan hasil yang cukup baik. Model Mask R-CNN yang dilatih dalam 25 epoch pembelajaran menghasilkan akurasi pembelajaran senilai 96,80% dan footprint bangunan yang dihasilkan memiliki nilai recall (kelengkapan) 88,83%, kepresisian 91,65%, dan nilai Intersection over Union (IoU) 91,90% ketika dibandingkan dengan data ground truth. Proses ekstrusi footprint bangunan hasil ekstraksi otomatis tersebut menghasilkan model bangunan 3D dalam Level of Detail (LOD) 2 dengan nilai Root Mean Square Error (RMSE) < 2 meter berdasarkan standar City Geography Markup Language (CityGML).

References

  1. Ammour, N., Alhichri, H., Bazi, Y., Benjdira, B., Alajlan, N., & Zuair, M. (2017). Deep learning approach for car detection in UAV imagery. Remote Sensing, 9(4). https://doi.org/10.3390/rs9040312
  2. Biljecki, F., Ledoux, H., & Stoter, J. (2016). An improved LOD specification for 3D building models. In Computers, Environment and Urban Systems (Vol. 59).
  3. https://doi.org/10.1016/j.compenvurbsys.2016.04.005
  4. Biljecki, F., Ledoux, H., & Stoter, J. (2017). Generating 3D city models without elevation data. Computers, Environment and Urban Systems, 64, 1–18. https://doi.org/10.1016/j.compenvurbsys.2017.01.001
  5. Chen, M., & Li, J. (2019). Deep convolutional neural network application on rooftop detection for aerial image. arXiv, 0–3.
  6. Cowton, J., Kyriazakis, I., & Bacardit, J. (2019). Automated Individual Pig Localisation, Tracking and Behaviour Metric Extraction Using Deep Learning. IEEE Access, 7, 108049–108060. https://doi.org/10.1109/ACCESS.2019.2933060
  7. de Carvalho, O. L. F., de Carvalho, O. A., Albuquerque, A. O. de, Bem, P. P. de, Silva, C. R., Ferreira, P. H. G., de Moura, R. D. S., Gomes, R. A. T., Guimarães, R. F., & Borges, D. L. (2021). Instance segmentation for large, multi-channel remote sensing imagery using mask-RCNN and a mosaicking approach. Remote Sensing, 13(1), 1–24. https://doi.org/10.3390/rs13010039
  8. Fawcett, D., Azlan, B., Hill, T. C., Kho, L. K., Bennie, J., & Anderson, K. (2019). Unmanned aerial vehicle (UAV) derived structure-from-motion photogrammetry point clouds for oil palm (Elaeis guineensis) canopy segmentation and height estimation. International Journal of Remote Sensing, 40(19), 7538–7560. https://doi.org/10.1080/01431161.2019.1591651
  9. Gómez, C., White, J. C., & Wulder, M. A. (2016). Optical remotely sensed time series data for land cover classification: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 116, 55–72. https://doi.org/10.1016/j.isprsjprs.2016.03.008
  10. Gröger, G., Kolbe, T. H., Nagel, C., & Häfele, K.-H. (2012). OpenGIS City Geography Markup Language (CityGML) Encoding Standard, Version 2.0.0. OGC Document No. 12-019, 344. https://portal.opengeospatial.org/files/?artifact_id=47842
  11. He, K., Gkioxari, G., Dollár, P., & Girshick, R. (2018). Mask R-CNN. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(2), 386–397. https://doi.org/10.1109/TPAMI.2018.2844175
  12. Hemanth, D. J., & Estrela, V. V. (2017). DEEP LEARNING FOR IMAGE PROCESSING APPLICATIONS. IOS Press.
  13. Khan, S., Rahmani, H., Shah, S. A. A., & Bennamoun, M. (2018). A Guide to Convolutional Neural Networks for Computer Vision. In Synthesis Lectures on Computer Vision (Vol. 8, Nomor 1). https://doi.org/10.2200/s00822ed1v01y201712cov015
  14. Kraff, N. J., Wurm, M., & Taubenbock, H. (2020). Uncertainties of Human Perception in Visual Image Interpretation in Complex Urban Environments. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13, 4229–4241. https://doi.org/10.1109/jstars.2020.3011543
  15. Kwak, E., Habib, A., & Al-Durgham, M. (2011). Model-based automatic 3d building model generation by integrating lidar and aerial images. Archiwum Fotogrametrii, Kartografii i Teledetekcji, 22(January 2015), 187–200.
  16. Miroslaw-Swiatek, D., Szporak-Wasilewska, S., Michalowski, R., Kardel, I., & Grygoruk, M. (2016). Developing an algorithm for enhancement of a digital terrain model for a densely vegetated floodplain wetland. Journal of Applied Remote Sensing, 10(3), 036013. https://doi.org/10.1117/1.jrs.10.036013
  17. Mlambo, R., Woodhouse, I. H., Gerard, F., & Anderson, K. (2017). Structure from motion (SfM) photogrammetry with drone data: A low cost method for monitoring greenhouse gas emissions from forests in developing countries. Forests, 8(3). https://doi.org/10.3390/f8030068
  18. Monteiro, F. C., & Campilho, A. C. (2005). Performance evaluation of image segmentation algorithms. Handbook of Pattern Recognition and Computer Vision, 3rd Edition, 525–542. https://doi.org/10.1142/9789812775320_0028
  19. Peeroo, U., Idrees, M. O., & Saeidi, V. (2017). Building extraction for 3D city modelling using airborne laser scanning data and high-resolution aerial photo. South African Journal of Geomatics, 6(3), 363. https://doi.org/10.4314/sajg.v6i3.7
  20. Sornapudi, S., Long, L. R., Almubarak, H., & Antani, S. K. (2018). Deep Learning Nuclei Detection in Digitized Histology Images by Superpixels. Journal of Pathology Informatics, 9(1). https://doi.org/10.4103/jpi.jpi
  21. Uba, N. (2016). Land Use and Land Cover Classification Using Deep Learning Techniques. Arizona State University.
  22. Zhang, A., Lipton, Z. C., Li, M., & Smola, A. J. (2020). Dive Into Deep Learning. https://doi.org/10.1016/j.jacr.2020.02.005
  23. Zhao, K., Kang, J., Jung, J., & Sohn, G. (2018). Building extraction from satellite images using mask R-CNN with building boundary regularization. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2018-June, 242–246. https://doi.org/10.1109/CVPRW.2018.00045

  1. Tata Bahtera Negara  Departemen Teknik Geodesi-Fakultas Teknik Universitas Gadjah Mada, Indonesia
  2. Harintaka Harintaka  Departemen Teknik Geodesi-Fakultas Teknik Universitas Gadjah Mada, Indonesia