Segmentation of Buildings Using U-Net Model from Göktürk-1 Satellite Images

Yıl 2023, , 50 - 58, 30.06.2023

Duygu Arıkan , Ferruh Yıldız

https://doi.org/10.51489/tuzal.1300939

Öz

The increase in population has led to unplanned urbanization in urban areas, becoming a global issue. The identification and detection of these areas are of great importance for urban management and redevelopment planning. However, these processes can be costly and time-consuming when conducted on-site. Automatic detection and characterization of unplanned buildings in urban and rural areas using remote sensing imagery is a challenging task. Recently, with the advancements in deep learning methods, the detection of complex buildings has become possible. In this study, the building extraction process of a region from the Etimesgut district of Ankara was performed using the U-Net deep learning architecture. The Inria Aerial Image Labeling dataset, a publicly available dataset, was used for the process. Different numbers of images (500, 1000, 2500, 5000) were selected for the training process. The best learning outcome was tested with Göktürk-1 satellite imagery with a spatial resolution of 0.5 m. According to the results, the U-Net model achieved a Jaccard coefficient of 0.862 and a Dice similarity coefficient of 0.813 for building segmentation.The effectiveness and potential of deep learning methods were demonstrated using the U-Net model with the available dataset. This study showcased the efficiency and potential of deep learning methods in the detection and mapping of buildings in urban areas.

Anahtar Kelimeler

Deep Learning, U-Net Model, Gokturk-1 Satellite Image, Building Detection

Göktürk-1 Uydu Görüntülerinden U-Net Modeli Kullanılarak Binaların Segmentasyonu

Öz

Nüfus artışı, kentsel bölgelerde plansız yapılaşmanın ortaya çıkmasına yol açmaktadır. Bu durum dünya genelinde bir sorun haline gelmiştir. Bu alanların belirlenmesi ve tespit edilmesi, kentsel yönetim ve yeniden yapılanma planlaması için büyük öneme sahiptir. Ancak bu işlemler, arazide maliyetli ve zaman alıcı olabilmektedir. Uzaktan algılama görüntüleri kullanarak kentsel ve kırsal bölgelerde plansız yapılan binaları otomatik olarak tespit etmek ve karakterize etmek oldukça zordur. Son zamanlarda, derin öğrenme yöntemleri sayesinde karmaşık binaların tespiti mümkün hale gelmiştir. Bu çalışmada, Ankara'nın Etimesgut ilçesinden bir bölgenin bina çıkarımı işlemi, U-Net derin öğrenme mimarisi kullanılarak gerçekleştirilmiştir. İşlem için Inria Aerial Image Labeling adlı hazır bir veri seti kullanılmıştır. Eğitim işlemi için farklı sayıda görüntü (500, 1000, 2500, 5000) seçilmiştir. En iyi öğrenme sonucu, 0.5 m uzamsal çözünürlüğe sahip Göktürk-1 uydu görüntüleriyle test edilmiştir. Sonuçlara göre, U-Net modelinin bina segmentasyonunda Jaccard katsayısı 0.862, Dice benzerlik oranı 0.813 olarak bulunmuştur. Hazır veri seti kullanılarak U-Net modelinin derin öğrenme yöntemleri için kullanılabilir olduğu kanıtlanmıştır. Bu çalışma, kentsel alanlardaki binaların tespiti ve haritalanmasında derin öğrenme yöntemlerinin etkinliğini ve potansiyelini göstermiştir.

Anahtar Kelimeler

Derin Öğrenme, U-Net mimarisi, Göktürk-1 Uydu Görüntüsü, Bina çıkarımı

Kaynakça

• Abdollahi, A., & Pradhan, B. (2021). Integrating semantic edges and segmentation information for building extraction from aerial images using UNet. Machine Learning with Applications, 6, 100194.
• Atlan, F., Hançer, E., & Pençe, İ. (2020). U-Net ile Çekirdek Segmentasyonunda Hiper Parametre Optimizasyonu Etkisinin Değerlendirilmesi. Avrupa Bilim ve Teknoloji Dergisi, 60-69.
• Ataş, İ. (2023). Performance Evaluation of Jaccard-Dice Coefficient on Building Segmentation from High Resolution Satellite Images. Balkan Journal of Electrical and Computer Engineering, 11(1), 100-106.
• Bayraktar, U. (2018). Derin Öğrenme Tabanlı Kanserli Hücre Tespiti. no. December, 2019.
• Hou, Y., Liu, Z., Zhang, T., & Li, Y. (2021). C-UNet: Complement UNet for remote sensing road extraction. Sensors, 21(6), 2153.
• Ji, S., Wei, S., & Lu, M. (2018). Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set. IEEE Transactions on Geoscience and Remote Sensing, 57(1), 574-586.
• Kattenborn, T., Leitloff, J., Schiefer, F., & Hinz, S. (2021). Review on Convolutional Neural Networks (CNN) in vegetation remote sensing. ISPRS Journal of photogrammetry and remote sensing, 173, 24-49.
• Kuffer, M., Pfeffer, K., & Sliuzas, R. (2016). Slums from space—15 years of slum mapping using remote sensing. Remote Sensing, 8(6), 455.
• Li, L., Wang, C., Zhang, H., & Zhang, B. (2019). Residual UNet for urban building change detection with Sentinel-1 SAR data. Paper presented at the IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium.
• Li, X., Zhang, G., Cui, H., Hou, S., Chen, Y., Li, Z., . . . Wang, H. (2023). Progressive fusion learning: A multimodal joint segmentation framework for building extraction from optical and SAR images. ISPRS Journal of photogrammetry and remote sensing, 195, 178-191.
• Maas, A. L., Hannun, A. Y., & Ng, A. Y. (2013). Rectifier nonlinearities improve neural network acoustic models. Paper presented at the Proc. icml.
• Madhu, G., Kautish, S., Alnowibet, K. A., Zawbaa, H. M., & Mohamed, A. W. (2023). NIPUNA: A Novel Optimizer Activation Function for Deep Neural Networks. Axioms, 12(3), 246.
• Maggiori, E., Tarabalka, Y., Charpiat, G., & Alliez, P. (2017). Can semantic labeling methods generalize to any city? the inria aerial image labeling benchmark. Paper presented at the 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS).
• Mahabir, R., Croitoru, A., Crooks, A. T., Agouris, P., & Stefanidis, A. (2018). A critical review of high and very high-resolution remote sensing approaches for detecting and mapping slums: Trends, challenges and emerging opportunities. Urban Science, 2(1), 8.
• Mitra, P., Shankar, B. U., & Pal, S. K. (2004). Segmentation of multispectral remote sensing images using active support vector machines. Pattern recognition letters, 25(9), 1067-1074.
• Mnih, V. (2013). Machine learning for aerial image labeling: University of Toronto (Canada).
• Nwankpa, C., Ijomah, W., Gachagan, A., & Marshall, S. (2018). Activation functions: Comparison of trends in practice and research for deep learning. arXiv preprint arXiv:1811.03378.
• Oliphant, T. E. (2006). A guide to NumPy (Vol. 1): Trelgol Publishing USA.
• Öz, M. (2021). Eye segmentation using deep neural networks.
• Pal, M. (2005). Random forest classifier for remote sensing classification. International journal of remote sensing, 26(1), 217-222.
• Pan, Z., Xu, J., Guo, Y., Hu, Y., & Wang, G. (2020). Deep learning segmentation and classification for urban village using a worldview satellite image based on U-Net. Remote Sensing, 12(10), 1574.
• Ramachandran, P., Zoph, B., & Le, Q. V. (2017). Searching for activation functions. arXiv preprint arXiv:1710.05941.
• Ronneberger, O., Fischer, P., & Brox, T. (2015a). U-net: Convolutional networks for biomedical image segmentation. Paper presented at the Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18.
• Ronneberger, O., Fischer, P., & Brox, T. (2015b). U-net: Convolutional networks for biomedical image segmentation. Paper presented at the International Conference on Medical image computing and computer-assisted intervention.
• Sariturk, B., & Seker, D. Z. (2022). Comparison of Residual and Dense Neural Network Approaches for Building Extraction from High-Resolution Aerial Images. Advances in Space Research.
• Tuermer, S., Kurz, F., Reinartz, P., & Stilla, U. (2013). Airborne vehicle detection in dense urban areas using HoG features and disparity maps. IEEE Journal of selected topics in applied earth observations and remote sensing, 6(6), 2327-2337.
• Wu, G., Shao, X., Guo, Z., Chen, Q., Yuan, W., Shi, X., . . . Shibasaki, R. (2018). Automatic building segmentation of aerial imagery using multi-constraint fully convolutional networks. Remote Sensing, 10(3), 407.
• Wurm, M., Stark, T., Zhu, X. X., Weigand, M., & Taubenböck, H. (2019). Semantic segmentation of slums in satellite images using transfer learning on fully convolutional neural networks. ISPRS Journal of photogrammetry and remote sensing, 150, 59-69.
• Yan, X., Tang, H., Sun, S., Ma, H., Kong, D., & Xie, X. (2022). After-unet: Axial fusion transformer unet for medical image segmentation. Paper presented at the Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision.
• Zeiler, M. D., Ranzato, M., Monga, R., Mao, M., Yang, K., Le, Q. V., . . . Dean, J. (2013). On rectified linear units for speech processing. Paper presented at the 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.
• Zhang, L., Zhang, L., Tao, D., & Huang, X. (2011). On combining multiple features for hyperspectral remote sensing image classification. IEEE Transactions on Geoscience and Remote Sensing, 50(3), 879-893.
• Zhang, J., Du, J., Liu, H., Hou, X., Zhao, Y., & Ding, M. (2019). LU-NET: An improved U-Net for ventricular segmentation. IEEE Access, 7, 92539-92546.
• Zhu, X. X., Tuia, D., Mou, L., Xia, G.-S., Zhang, L., Xu, F., & Fraundorfer, F. (2017). Deep learning in remote sensing: A comprehensive review and list of resources. IEEE geoscience and remote sensing magazine, 5(4), 8-36.