現今航空運輸業中，不論軍用或者民用，固定翼飛行器進行進場降落時通常必須要有特定跑道以及特定輔助降落資訊進行起降程序，以確保飛行安全。然而，當既有的飛行跑道或起降場之輔助設施無法提供飛行器安全的降落環境時，飛行器必須採取自主式降落等相關措施，以確保飛行安全。除了能保存飛行器完好外，也不會隨意讓飛行器落地，造成額外的人員與財物損失。
本文將透過影像辨識、人工智慧、導引系統，作為本文研究與應用重點，在衛星導航系統無法運作的情況下，如何藉由無人機上之儀器提供自主降落能力。透過掛載於機腹下方的機載鏡頭進行地面攝影，藉由SURF(Speed Up Robust Features)演算法進行影像拼接的技術建立起區域性的二維地圖場景，並且同時處理飛行器之定位以及在不同飛行高度下，雙尺度地圖之建構，以確認飛機於空間中的絕對位置。完成場域建立後，透過YOLO等影像辨識技術建立起影像學習的環境，進行後續降場區域之辨別。最後，使機載電腦依照標準降落程序進行降場之規劃並且透過即時影像導引航機至最終降落階段，使固定翼的無人飛行器能從巡行狀態下最終降落至危險性最低之降場或公路路段。

The purpose of this study was to build a vision-based automatic landing system for fixed-wing Unmanned Aerial Vehicle (UAV) due to a Global Positioning System (GPS) denied environment to ensure flight safety. Through the onboard camera installed on the center of the aircraft belly, orthogonal projection images were taken to implement the method of image stitching by SURF algorithm. By image stitching, a two-dimensional panoramic map was created as a global map; when the flight altitude changed, the instant image was assigned as a local map. With the two maps, a dual-scale image relation was connected to confirm the location of the aircraft in space.
After completing the built up of the map, the process of image recognition could be applied to detect the suitable landing candidates, e.g. runways or straight highway sections, by YOLO – a real-time object detecting system. With the selected region to land verified, the onboard computer could calculate the gliding route according to standard approach procedure and guide the aircraft to the final touch down by real-time images.
There are four goals to approach in this thesis:
(a)The realization of image stitching using orthogonal projection aerial images
(b)The recognition between dual-scale images and aircraft locations
(c)Integration of the technology about image identification and classification about runways and highways
(d)The guidance of the landing target

Abu-Jbara, Khaled & Alheadary, Wael & Sundaramorthi, Ganesh & Claudel, Christian. (2015). A robust vision-based runway detection and tracking algorithm for automatic UAV landing. 2015 International Conference on Unmanned Aircraft Systems, ICUAS 2015. 1148-1157. 10.1109/ICUAS.2015.7152407.
Alomran, Murtadha & Chai, Douglas. (2016). Feature-based panoramic image stitching. 1-6. 10.1109/ICARCV.2016.7838721.
Akbar, Javeria & Shahzad, Muhammad & Malik, Muhammad & Ul-Hasan, Adnan & Shafait, Fasial. (2019). Runway Detection and Localization in Aerial Images using Deep Learning. 1-8. 10.1109/DICTA47822.2019.8945889.
Asadzadeh, Maryam & Shivakumara, Palaiahnakote & Idris, Mohd & Anisi, Hossein & Lu, Tong & Blumenstein, Michael & N.M, Noor. (2019). An Automatic Zone Detection System for Safe Landing of UAVs. Expert Systems with Applications. 122. 10.1016/j.eswa.2019.01.024.
Bay, Herbert & Tuytelaars, Tinne & Van Gool, Luc. (2006). SURF: Speeded Up Robust Features. Computer Vision and Image Understanding - CVIU. 110. 404-417.
Brown, Matthew & Lowe, David. (2007). Automatic Panoramic Image Stitching using Invariant Features. International Journal of Computer Vision. 74. 59-73. 10.1007/s11263-006-0002-3.
Bachelder, Edward. (2017). Novel Helicopter Flight Director and Display. 10.1007/978-3-319-41956-5_31.
Chang, Jun-Dong & Yu, Shyr-Shen & Chen, Hong-Hao & Tsai, Chwei-Shyong. (2010). HSV-based Color Texture Image Classification using Wavelet Transform and Motif Patterns. Journal of Computers. 20.
Chen, Kaili & Wang, Meiling. (2014). Image stitching algorithm research based on OpenCV. Proceedings of the 33rd Chinese Control Conference, CCC 2014. 7292-7297. 10.1109/ChiCC.2014.6896208.
Cheng, Gong & Han, Junwei & Lu, Xiaoqiang. (2017). Remote Sensing Image Scene Classification: Benchmark and State of the Art. Proceedings of the IEEE. 105. 1865-1883. 10.1109/JPROC.2017.2675998.
Duan, Dagao & Xie, Meng & Mo, Qian & Han, Zhongming & Wan, Yueliang. (2010). An improved Hough transform for line detection. ICCASM 2010 - 2010 International Conference on Computer Application and System Modeling, Proceedings. 2. V2-354. 10.1109/ICCASM.2010.5620827.
Dehshibi, Mohammad & Fahimi, Mohammad & Mashhadi, Mohsen. (2015). Vision-Based Site Selection for Emergency Landing of UAVs. Advances in Intelligent Systems and Computing. 361. 10.1007/978-3-319-19024-2_14.
Guo, Xufeng & Denman, Simon & Fookes, Clinton & Mejias, Luis & Sridharan, Sridha. (2015). Automatic UAV Forced Landing Site Detection Using Machine Learning. 2014 International Conference on Digital Image Computing: Techniques and Applications, DICTA 2014. 10.1109/DICTA.2014.7008097.
Huertas, Andres & Cole, William & Nevatia, Ramakant. (1987). Detecting Runways in Aerial Images. 712-717.
Heipke, C. & Mayer, Helmut & Wiedemann, C. & Jamet, Olivier. (1997). Evaluation of Automatic Road Extraction. Inter. Arch. Photogramm. Remote Sens. 32.
He, Kaiming & Chang, Huiwen & Sun, Jian. (2013). Rectangling Panoramic Images via Warping. ACM Transactions on Graphics (TOG). 32. 10.1145/2461912.2462004.
Ibraheem, Noor & Hasan, Mokhtar & Khan, Rafiqul Zaman & Mishra, Pramod. (2012). Understanding Color Models: A Review. ARPN Journal of Science and Technology. 2.
Kathuria, Ayoosh (2018), What’s new in YOLO v3? Retrieved from https://towardsdatascience.com/yolo-v3-object-detection-53fb7d3bfe6b
Lowe, David. (2001). Object Recognition from Local Scale-Invariant Features. Proceedings of the IEEE International Conference on Computer Vision. 2.
Laiacker, Maximilian & Kondak, Konstantin & Schwarzbach, Marc & Muskardin, Tin. (2013). Vision aided automatic landing system for fixed wing UAV. IEEE International Conference on Intelligent Robots and Systems. 2971-2976. 10.1109/IROS.2013.6696777.
Mistry, Dr & Banerjee, Asim. (2017). Comparison of Feature Detection and Matching Approaches: SIFT and SURF. GRD Journals- Global Research and Development Journal for Engineering. 2. 7-13.
Poleg, Yair & Peleg, Shmuel. (2012). Alignment and Mosaicing of Non-Overlapping Images. 2012 IEEE International Conference on Computational Photography, ICCP 2012. 10.1109/ICCPhot.2012.6215214.
Redmon, Joseph. (2013-2016). Darknet : Open Source Neural Networks in C. Retrieved from http://pjreddie.com/darknet/
Redmon, Joseph & Divvala, Santosh & Girshick, Ross & Farhadi, Ali. (2016). You Only Look Once: Unified, Real-Time Object Detection. 779-788. 10.1109/CVPR.2016.91.
Redmon, Joseph & Farhadi, Ali. (2018). YOLOv3: An Incremental Improvement.
Radovic, Matija & Adarkwa, Offei & Wang, Qiaosong. (2017). Object Recognition in Aerial Images Using Convolutional Neural Networks. Journal of Imaging. 3. 21. 10.3390/jimaging3020021.
Shi, Huaitao & Guo, Lei & Tan, Shuai & Li, Gang & Sun, Jie. (2019). Improved Parallax Image Stitching Algorithm Based on Feature Block. Symmetry. 11. 348. 10.3390/sym11030348.