隨著手機攝影功能的普及讓「自拍」成為時下人們的新文化。加上多軸飛行器技術的快速發展，飛行穩定度越來越好，藉由其靈活飛行的特性，加裝鏡頭在其上，形成了現在人們所稱的「空拍機」，因其能夠拍攝更許多手持相機、自拍棒等，無法拍攝出的畫面，而逐漸廣受人們喜愛。但要靈活操控空拍機需要大量的練習，因此我們希望設計出一套系統，藉由簡單的步驟便能讓無操控經驗的使用者利用空拍機進行自拍。在此系統中，我們讓使用者先拍攝自拍時想合照的目標物，然後依據使用者心中原本的構圖來對照上述所拍攝的影像圖，進而計算出能拍攝出原構圖結果的位置，並讓空拍機飛往該定點進行拍攝，因此如何計算出準確的位置，是我們想要著手研發系統的核心。本論文針對此部分進行了研究，設計出兩種計算方法並進行實驗比較。

Nowadays, flying technology of drone is more and more stable. Camera-equipped drones are more popular today. Using drone to selfie can take a picture with more possible shot angle and distance. However, controlling a drone require a longer training. In this paper, We firstly survey the approaches about drone, and the state-of-the-art monocular visual positioning approaches. We want to invest a system which let unexperienced drone users can selfie with drone. In our system, user can take an initial shot, and composing this shot. System calculate a position to drone. Drone can shot at this position and get a photo which similar to user composition result. So how to calculate the target camera position is the most important part of our paper. We design two methods, and evaluate them by experiment to decide the most performance method.

口試委員會審定書 i
誌謝 ii
中文摘要 iii
ABSTRACT iv
目錄 v
圖目錄 vii
表目錄 ix
第一章 緒論 1
1.1 研究動機 1
1.2 初步研究 1
1.3 研究流程 2
第二章 文獻探討 4
2.1 空拍機相關文獻 4
2.2 基於視覺之定位 5
第三章 系統總覽 8
3.1 系統架構 8
3.2 系統流程簡介 9
3.2.1 初始照片（Initial Shot） 9
3.2.2 構圖（Manual Composition） 10
3.2.3 畫面微調（Manual Refinement） 11
第四章 定位技術 12
4.1 ORB-SLAM 12
4.1.1 Tracking 12
4.1.2 Local Mapping 13
4.1.3 Loop Closing 13
4.2 空拍機建構場景模型之盤旋方法探討 13
4.2.1 盤旋方法設計 13
4.2.2 實驗與結果 14
4.3 動態環境下之場景模型建構 17
4.3.1 動態環境 17
4.3.2 實驗與結果 17
第五章 空拍機目標位置計算 18
5.1 方法一：基於模型之構圖結果定位 18
5.1.1 系統架構與方法 18
5.1.2 實驗與結果 18
5.2 方法二：基於幾何學之位置推算法 22
5.2.1 系統架構 22
5.2.2 座標系 22
5.2.3 目標物與初始照片拍攝位置距離算法介紹與條件 23
5.2.4 位置計算 24
5.2.5 實驗與結果 25
5.3 兩方法結果比較與分析 29
第六章 結論與未來發展 32
參考文獻 33

[1] Atap.google.com. (2017). Ara. [online] Available at: https://atap.google.com/ara/ [Accessed 18 Jul. 2017].
[2] Motorola. (2017). Explore the world of Moto Mods™. [online] Available at: https://www.motorola.com/us/moto-mods [Accessed 18 Jul. 2017].
[3] Higuchi K, Shimada T, Rekimoto J. Flying sports assistant. Proceedings of the 2nd Augmented Human International Conference on - AH ’11. 2011. doi:10.1145/1959826.1959833
[4] Joubert, N., Roberts, M., Truong, A., Berthouzoz, F., & Hanrahan, P. (2015). An interactive tool for designing quadrotor camera shots. ACM Transactions on Graphics (TOG), 34(6), 238.
[5] Chen, C. F., Liu, K. P., & Yu, N. H. (2015, November). Exploring interaction modalities for a selfie drone. In SIGGRAPH Asia 2015 Posters (p. 25). ACM.
[6] Mur-Artal, R., Montiel, J. M. M., & Tardos, J. D. (2015). ORB-SLAM: a versatile and accurate monocular SLAM system. IEEE Transactions on Robotics, 31(5), 1147-1163.
[7] Rublee, E., Rabaud, V., Konolige, K., & Bradski, G. (2011, November). ORB: An efficient alternative to SIFT or SURF. In Computer Vision (ICCV), 2011 IEEE international conference on (pp. 2564-2571). IEEE.
[8] Liang, Q. Visual-Inertial Ego-Positioning for Flying Cameras. M.S. thesis, National Taiwan University, 2016
[9] Andrew J Davison, Ian D Reid, Nicholas D Molton, and Olivier Stasse. Monoslam:Real-time single camera slam. IEEE transactions on pattern analysis and machine intelligence, 29(6):102–1067, 2007.
[10] Javier Civera, Andrew J Davison, and JM Martinez Montiel. Inverse depth parametrization for monocular slam. IEEE transactions on robotics, 24(5):932–945, 2008.
[11] Ethan Eade and Tom Drummond. Scalable monocular slam. In 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR’06), volume 1, pages 469–476. IEEE, 2006.
[12] Jakob Engel, Thomas Schóps, and Daniel Cremers. Lsd-slam: Large-scale direct monocular slam. In European Conference on Computer Vision, pages 834–849. Springer, 2014.
[13] Richard A Newcombe, Steven J Lovegrove, and Andrew J Davison. Dtam: Dense tracking and mapping in real-time. In 2011 international conference on computer vision, pages 2320–2327. IEEE, 2011.
[14] Jan Stúhmer, Stefan Gumhold, and Daniel Cremers. Real-time dense geometry from a handheld camera. In Joint Pattern Recognition Symposium, pages 11–20. Springer, 2010.
[15] Matia Pizzoli, Christian Forster, and Davide Scaramuzza. Remode: Probabilistic, monocular dense reconstruction in real time. In 2014 IEEE International Conference on Robotics and Automation (ICRA), pages 2609–2616. IEEE, 2014.
[16] Jakob Engel, Jurgen Sturm, and Daniel Cremers. Semi-dense visual odometry for a monocular camera. In Proceedings of the IEEE international conference on computer vision, pages 1449–1456, 2013.
[17] Thomas Schöps, Jakob Engel, and Daniel Cremers. Semi-dense visual odometry for ar on a smartphone. In Mixed and Augmented Reality (ISMAR), 2014 IEEE International Symposium on, pages 145–150. IEEE, 2014.
[18] Georg Klein and David Murray. Parallel tracking and mapping for small ar workspaces. In Mixed and Augmented Reality, 2007. ISMAR 2007. 6th IEEE and ACM International Symposium on, pages 225–234. IEEE, 2007.
[19] Kuan-Wen Chen, Chun-Hsin Wang, Xiao Wei, Qiao Liang, Chu-Song Chen, Ming-Hsuan Yang, and Yi-Ping Hung. Vision-based positioning for internet-of-vehicles. IEEE Transactions on Intelligent Transportation Systems, 2016.
[20] Vicon. Vicon bonita. http://www.vicon.com/products/camerasystems/bonita.