簡易檢索 / 詳目顯示

回結果列表
研究生： 翟德雲
Jair, De-Yun
論文名稱： 基於向量式遮蔽偵測法產製無遮蔽橋梁表面正射影像
Implementation of a Vector-based Occlusion Detection Method to Generate "Occlusion-free" Bridge Surface Orthoimages
指導教授： 饒見有
Rau, Jiann-Yeou
學位類別： 碩士
Master
系所名稱： 工學院 - 測量及空間資訊學系
Department of Geomatics
論文出版年： 2024
畢業學年度： 112
語文別： 英文
論文頁數： 105
中文關鍵詞： 遮蔽偵測 、立體模型 、正射影像 、攝影測量 、無人機
外文關鍵詞： Occlusion Detection, 3D CAD Model, Orthoimage, Photogrammetry, UAV
相關次數： 點閱：82  下載：5 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 裂縫等橋樑劣化會對橋梁造成傷害並且造成重大的事故。因此，有必要偵測並且處理橋梁上的各種劣化。然而，一般的人工偵測相當耗時且成本高。相對地，人工智慧(Artificial Intelligence)可以在短時間處理大量的資料並且進行影像辨識。此外，無人機(UAV)的應用相當廣泛，可以用來取得近景影像。因此，本研究嘗試建立由多個矩形多邊形所構成的立體橋樑模型，並利用無人機所獲得的橋梁影像製作這些多邊形的正射影像。這些無遮蔽的正射影像在未來可以利用人工智慧，進行橋梁正射影像的劣化偵測，並獲得這些劣化在橋樑上的位置。
    在此研究中，橋樑模型由帶有地理座標的多邊形所組成。接著利用無人機拍攝高空間解析度的橋樑影像以製作橋樑的正射影像。在取得橋梁影像時，會產生遮蔽的情形。因此，這些遮蔽必須被偵測出以防止重複映射(Double Mapping)的情形出現在正射影像上。本研究提出了一個向量式的遮蔽偵測法，用以偵測出橋梁影像的遮蔽資訊，並得到非遮蔽的資訊。一張無遮蔽的正射影像是由這些非遮蔽資訊製作，並且橋梁的其他影像會用來補償遮蔽以製作出無遮蔽的橋梁正射影像。最後，色彩平衡會被應用在正射影像上以平衡主影像(Master Image)和從影像(Slave Image)的色彩。研究成果顯示，此研究所提出的向量式遮蔽偵測用以製作正射影像以及色彩平衡能夠改善橋梁正射影像的正確率以及視覺呈現。

    Defects, such as cracks, in a bridge may cause damage to the bridge, and they may cause catastrophic incidents. Thus, it is essential to detect those defects in bridges and take actions against them. However, it costs a lot of money and time to perform visual inspection on bridges. On the contrary, artificial intelligence can not only deal with large amount of data but also perform image recognition in a short period of time. Besides, it is considered that UAV is a flexible platform for close-range image acquisition. Therefore, a 3D CAD model of a bridge composed of 3D rectangular polygons is generated first, and an orthoimage for each 3D polygon is created using images acquired by the UAV. Defect detection with AI is then able to be applied to the orthoimages in the future, and the spatial positions of those defects on the entire bridge can be obtained.
    In this study, the surface of a bridge is decomposed into numerous 3D polygons with geospatial coordinates to present the 3D model of a bridge. After that, very high spatial resolution images of a bridge are taken by UAV, and generate orthoimages for the bridge surface. During the image acquisition, occlusion of some parts of the bridge may occur. Thus, it is important to detect those occluded areas in case the source images have occlusion effect causing double mapping in orthoimages. A vector-based occlusion detection algorithm is proposed to determine non-occluded area of each 2D target polygon in every source image. An occlusion-free orthoimage is created using non-occluded area of 2D polygons, and other available source images are used to compensate for the occluded area in order to create an occlusion-free orthoimage. Finally, color balancing is performed to balance the color between the master and slave images. Experimental results show that the proposed vector-based occlusion detection and image compensation together with the color balancing processes improve the overall completeness ratio and the visual presentation of orthoimages.

    摘要 I ABSTRACT II 致謝 III TABLE OF CONTENTS IV List of Tables VI List of Figures VII Chapter 1 Introduction 1 1.1 Background 1 1.2 Motivation 1 1.3 Research Approach 1 1.4 Thesis Structure 2 Chapter 2 Literature Review 3 2.1 Occlusion Detection 3 2.2 Orthoimage Generation 4 Chapter 3 Methodology 5 3.1 Materials 7 3.1.1 UAV Images 7 3.1.2 3D Bridge Model 8 3.2 Preprocessing 10 3.2.1 Filtering Parameters 10 3.2.2 Back-Projection 11 3.3 Occlusion Detection 14 3.3.1 Vector-Based Occlusion Detection 15 3.3.2 Using NetTopologySuite 19 3.4 Calculate Total Non-Occluded Area 19 3.5 Create Orthoimage 20 3.5.1 Determine Size of Orthoimage 21 3.5.2 Using Bilinear Transformation 23 3.5.3 Using Collinearity Equations 26 3.5.4 Orthoimage Compensation 26 3.6 Color Balancing 27 3.6.1 Color Balancing to an Entire Orthoimage 29 3.6.2 Color Balancing to Seamline Buffer Zone 29 3.6.3 Average Filter 31 3.6.4 Color Balancing to Bridge Components 32 Chapter 4 Result and Analysis 33 4.1 Results of Occlusion Detection 33 4.2 Results of Orthoimage 42 4.2.1 Results of Orthoimage Without Occlusion Detection 45 4.2.2 Results of Orthoimage With Occlusion Detection 49 4.2.3 Results of Orthoimage Without Color Balancing 53 4.2.4 Results of Orthoimage With Color Balancing 54 4.3 Comparisons of Orthoimages 57 4.3.1 Comparisons of Orthoimages Between With and Without Occlusion Detection 57 4.3.2 Comparisons Between Orthoimages Using Bilinear Transformation and Collinearity Equations Without Color Balancing 58 4.3.3 Comparisons of Orthoimages Between With and Without Color Balancing 78 4.4 Comparisons of Bridge Orthoimages 80 4.4.1 Comparisons of Bridges Between With and Without Occlusion Detection 80 4.4.2 Comparisons of Bridges Between With and Without Color Balancing to Bridge Components 83 4.5 Special Cases 88 Chapter 5 Conclusions 92 References 93

    Albertz, J., & Wolf, B. (2004). Generating true orthoimages without a 3D surface model. International Archives of Photogrammetry and Remote Sensing.
    Amhar, F., Jansa, J., & Ries, C. (1998). The generation of true orthophotos using a 3D building model in conjunction with a conventional DTM. International Journal of Photogrammetry and Remote Sensing, 32, 16-22.
    Barazzetti, L., Brovelli, M., & Scaioni, M. (2008). Generation of True-Orthophotos with Lidar High Res. The Photogrammetric Journal of Finland,, 21.
    Barazzetti, L., Brumana, R., Oreni, D., Previtali, M., & Roncoroni, F. (2014). True-orthophoto generation from UAV images: Implementation of a combined photogrammetric and computer vision approach. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, II-5, 57-63.
    Chen, L. C., Teo, T. A., Wen, J. Y., & Rau, J. Y. (2007). Occlusion‐Compensated True Orthorectification For High‐Resolution Satellite Images. The Photogrammetric Record, 22(117), 39-52.
    de Oliveira, H. C., Dal Poz, A. P., Galo, M., & Habib, A. F. (2018). Surface Gradient Approach for Occlusion Detection Based on Triangulated Irregular Network for True Orthophoto Generation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 11(2), 443-457.
    de Oliveira, H. C., Galo, M., & Dal Poz, A. P. (2015). Height-Gradient-Based Method for Occlusion Detection in True Orthophoto Generation. IEEE Geoscience and Remote Sensing Letters, 12(11), 2222-2226.
    Ebrahimikia, M., & Hosseininaveh, A. (2022). True Orthophoto Generation Based on Unmanned Aerial Vehicle Images Using Reconstructed Edge Points. The Photogrammetric Record, 37(178), 161-184.
    Günay, A., Hossein, A., & Michael, H. (2007). True orthophoto production using lidar data. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 36.
    Habib, A. F., Kim, E. M., & Kim, C. J. (2007). New Methodologies for True Orthophoto. Photogrammetric Engineering & Remote Sensing, 73, 25-36.
    Kuzmin, Y., Korytnik, S., & Long, O. (2004). Polygon-based true orthophoto generation. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 35.
    Lindenberger, P., Sarlin, P., & Pollefeys, M. (2023). LightGlue Local: Feature Matching at Light Speed. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 17627-17638.
    Marsetic, A. (2021). Robust Automatic Generation of True Orthoimages From Very High-Resolution Panchromatic Satellite Imagery Based on Image Incidence Angle for Occlusion Detection. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, 3733-3749.
    Oda, K., Lu, W., Uchida, O., & Doihara, T. (2004). Triangle-based visibility analysis and true orthoimage. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 623-628.
    Rau, J. Y., Chen, N. Y., & Chen, L. C. (2002). True orthophoto generation of built-up areas using multi-view images. Photogrammetric Engineering & Remote Sensing, 68, 581-588.
    Rau, J. Y., & Chu, C. Y. (2011). Vector-Based Occlusion Detection for Automatic Facade Texture Mapping. 2011 International Workshop on Multi-Platform/Multi-Sensor Remote Sensing and Mapping, 1-6.
    Sheng, Y. (2007). Minimising algorithm‐induced artefacts in true ortho‐image generation: a direct method implemented in the vector domain. The Photogrammetric Record, 22(118), 151-163.
    Shin, Y. H., Lee, D. C., & Jung, H. S. (2021). True Orthoimage Generation Using Airborne LiDAR Data with Generative Adversarial Network-Based Deep Learning Model. Journal of Sensors, 2021, 1-25.
    Wang, Q., Yan, L., Sun, Y., Cui, X., Mortimer, H., & Li, Y. (2018). True orthophoto generation using line segment matches. The Photogrammetric Record, 33(161), 113-130.
    Wang, X., & Xie, J. (2012). A method for true orthophoto generation based on projection and iteration strategy. SPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, I-4.
    Widyaningrum, E., Lindenbergh, R. C., Gorte, B. G. H., & Zhou, K. (2018). Extraction of Building Roof Edges from Lidar Data to Optimize the Digital Surface Model for True Orthophoto Generation. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLII-2, 1199-1205.
    Xie, W., & Zhou, G. (2008). Experimental realization of urban large-scale true orthoimage generation. In Proceedings of the ISPRS Congress, Beijing, China.
    Yoo, E. J., & Lee, D. C. (2016). True orthoimage generation by mutual recovery of occlusion areas. GIScience & Remote Sensing, 53(2), 227-246.
    Zhou, G., Wang, Y., Yue, T., Ye, S., & Wang, W. (2016). Building Occlusion Detection From Ghost Images. IEEE Transactions on Geoscience and Remote Sensing, 55(2), 1074-1084.

    下載圖示 校內：立即公開
    校外：立即公開
    QR CODE