臺灣博碩士論文加值系統

機器人攝影師(Robot Photographer) 是近年來在機器人相關研究應用上另一項重要
的發展。然而，以往機器人攝影師只應用於拍攝出好的照片，但如何拍攝出好的影
片等相關研究少之又少。本論文的目標是開發一能夠自動攝影的機器人攝影師，能
夠使用一般攝影師常用的運鏡手法並且能夠構圖拍攝出好的影片。此外，此系統也
會評估拍攝得影片品質的好壞(Quality Assessment)，並過濾出品質較低的影片。在
本研究中，攝影機器人是以Pioneer-3DX 移動式機器人為基礎構成的。使用筆記型
電腦作為主要的控制運算設備。為了使用運鏡手法，我們採用Kinect 感測器獲取在
攝影機畫面中人的骨架資訊。此外，為了拍攝構圖，採用人臉偵測演算法獲取必要
的人臉位置資訊。在本系統裡包含了pan/tile 設備平台和可用電腦透過LANC(local
application control bus system) 控制的攝影機。並在實際的環境中測試實驗此系統。最
後結果顯示，攝影機器人可以成功使用許多運鏡手法自動拍攝影片並且可以刪除品
質較低的影片。我們以主觀評量方法進行評估，結果顯示使用攝影機器人拍攝取得
的影片是具有視覺上的吸引力。

Robot photographer is a recent important development in the field of robotics. However,
conventional robot photographer systems mainly aimed to take high quality still images. Very
few studies focus on how to shoot high quality films. The goal of this thesis is to develop an
automatic camera robot mimicking a professional camera man who can select proper camera
works and composition rules during filming. Furthermore, the camera robot will assess the
quality of each shot to filter out low quality video clips. In this study, the camera robot is
constructed based on a Pioneer-3DX mobile robot. A notebook computer is used as the main
controller. A Kinect sensor is adopted to gather skeleton information of people in the camera
field of view for selecting camera works. Also, a face detection algorithm is introduced to
acquire necessary information for shooting composition. The system also includes a pan-tilt
unit and a camcorder controlled by the computer through the local application control bus
system. The implemented system has been tested in a real environment. The results show
that the camera robot can successfully select different camera works for automatic filming
and delete low quality video clips. A subjective evaluation has been conducted and the results
show that the videos acquired with the camera robot are visually appealing.

致謝 i
論文摘要 ii
Abstract iii
目錄 iv
圖目錄 vi
第一章緒論 1
1.1 研究動機 1
1.2 相關研究與應用 2
1.3 研究目的 3
1.4 論文架構 3
第二章系統介紹 4
2.1 系統設備 4
2.1.1 深度攝影機(RGB-D Camera) 5
2.1.2 PowerPod-LANC Robotic Tripod 5
2.1.3 DV Camera(HDR-CX260V) 7
2.1.4 pan/tilt 平台(pan/tilt table) 8
2.1.5 Pioneer P3-DX Robot 9
2.2 系統架構 10
第三章研究方法 11
3.1 攝影運鏡 11
3.1.1 搖攝鏡頭(pan) 11
3.1.2 Zoom in/out 12
3.1.3 Dolly in 13
3.1.4 跟隨拍攝(Follow) 14
3.1.5 平行拍攝(Track) 15
3.1.6 Dolly Zoom 16
3.2 攝影構圖 17
3.2.1 三分法則(Rule of Thirds) 17
3.2.2 頭部空間(Head room) 18
3.2.3 水平構圖(Horizontal composition) 18
3.3 影像品質評估(Quality Assessment) 19
3.3.1 不穩定程度(Unstableness) 19
3.3.2 晃動程度(Jerkiness) 21
3.3.3 明暗程度(Brightness) 22
3.3.4 色彩多樣程度(Infidelity) 23
3.3.5 模糊程度(Blurring) 25
3.3.6 傾斜程度(Orientation) 27
3.3.7 Rule-Based Method 30
第四章實驗 31
4.1 實驗平台 31
4.2 實驗結果 33
4.2.1 實驗一 33
4.2.2 實驗二 33
4.2.3 實驗三 34
4.2.4 實驗四 39
4.2.5 實驗五 40
4.2.6 實驗六 41
4.2.7 實驗七 45
第五章結論與未來方向 46
5.1 結論 46
5.2 未來方向 47
參考文獻 48

[1] H. Ahn, D. Kim, J. Lee, S. Chi, K. Kim, J. Kim, M. Hahn, and H. Kim, “A robot photographer with user interactivity,” in Proceedings of the
IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 5637–5643, 2006.
[2] Y. Matsushita, E. Ofek, X. Tang, and H.-Y. Shum, “Full-frame video stabilization,” in Proceedings of the IEEE Computer Society Conference
on Computer Vision and Pattern Recognition, vol. 1, pp. 50– 57, 2005.
[3] W.-H. Cheng, Y.-Y. Chuang, Y.-T. Lin, C.-C. Hsieh, S.-Y. Fang, B.-Y. Chen, and J.-L.Wu, “Semantic analysis for automatic event recognition
and segmentation of wedding ceremony videos,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 18, pp. 1639–1650, 2008.
[4] Y. Y. Xiang and M. S. Kankanhalli, “Automated aesthetic enhancement of videos,” in Proceedings of the international conference on
Multimedia, MM ’10, (New York, NY,USA), pp. 281–290, ACM, 2010.
[5] http://www.sony.jp/cyber-shot/products/IPT-DS2/feature_1.html.
[6] http://www.imaging-resource.com/news/2012/05/24/the-camera-bag-robotic-smarttripod-is-a-drone-videographer.
[7] 丘錦榮, “Dv 拍攝完全探索-腳本、拍攝、剪接、輸出一次成.”
[8] T. Mei, X.-S. Hua, C.-Z. Zhu, H.-Q. Zhou, and S. Li, “Home video visual quality assessment with spatiotemporal factors,” IEEE Transactions
on Circuits and Systems for Video Technology, vol. 17, pp. 699– 706, 2007.
[9] http://tw.asus.com/Multimedia/Motion_Sensor/Xtion_PRO_LIVE/.
[10] http://plazma.kapsi.fi/diy/lanc_controller/.
[11] http://www.sony.com.tw/.
[12] http://www.studio1productions.com/lanc-sa.htm.
[13] http://librairie.immateriel.fr/fr/read_book/9780596516130/ch10s05.
[14] C. Mak and W. Cham, “Global motion estimation using modified least median of square error,” in Proceedings of 2004 International
Symposium on Intelligent Multimedia, Video and Speech Processing, pp. 591– 594, 2004.
[15] J.-G. Kim, H. S. Chang, J. Kim, and H.-M. Kim, “Efficient camera motion characterization for mpeg video indexing,” in Proceedings of the
IEEE International Conference on Multimedia and Expo, vol. 2, pp. 1171– 1174, 2000.
[16] H. Tong, H. Z. Mingjing Li, and C. Zhang, “Blur detection for digital images using wavelet transform,” in Proceedings of 2004 IEEE
International Conference on Multimedia and Expo, vol. 1, pp. 17 –20, 2004.
[17] M. L. Lei Zhang and H.-J. Zhang, “Boosting image orientation detection with indoor vs. outdoor classification,” in Proceedings of the
Sixth IEEE Workshop on Applications of Computer Vision, pp. 95– 99, 2002.
[18] http://opencv.willowgarage.com/documentation/python/objdetect_cascade_classification.html.
[19] ”NITE Algorithms 1.3 Note”, PrimeSense Inc., 2010.