臺灣博碩士論文加值系統

近年來大規模開放式線上課程十分的熱門，目前有越來越多的學校加入此行列。由於大規模開放式線上課程並非面對面的實際教學，所以為了使學生更清楚了解教學內容，更需要一個良好的課程錄影方法。並且為了讓互動式教學錄影看起來更生動並能引發學生對課程的興趣和能有更方便的錄製教學影像，有許多的教學錄製系統陸續被提出。目前的相關研究提出使用多台攝影機進行視覺分析、聲音分析判別目前所需表現的攝影機畫面，或藉由使用臉部偵測與手勢辨識對教學畫面進行評估，分辨教師目前位置與手勢進行畫面的切換。這些研究強調在錄製的影像或過程中如何對分鏡進行切換讓觀看者更加舒適與專注，但並沒有對學生發言行為進行區別和增加課程中的師生的互動畫面與紀錄。所以我們開發了一個智慧教室導播系統與手控教學模式(投影片、指定、全自動模式)指示器，此系統定義之學生發言事件包含舉手與起立發言事件，並且研發強健型發言事件偵測技術，可以過濾學生走動的行為以及課程中可能會觸發發言事件的行為。系統架構分成全自動模式以及半自動模式，教師藉由手中的指示器切換全自動或半自動模式。在全自動模式下系統全自動的追蹤教師位置、偵測學生發言事件，半自動模式下系統依然會持續追蹤教師位置，教師可以藉由指示器指定學生發言。指示器會發射出具有固定頻率閃爍的紅外線光以避免並非所需的紅外線訊號，系統即可取得位置並且控制PTZ攝影機拍攝教師所指定的學生位置，藉由此方法可以在畫面中增加拍攝師生互動的畫面，在課程中達到良好的拍攝效果。此系統能達成同時的追蹤教師、學生發言事件偵測和紅外線事件偵測並且正確率達到90%且每秒能處理約15張影像，達成即時處理。

In recent years, Massive Open Online Courses (MOOCs) become a novel online education trend. However, to produce an online education course often require manual recording such that this work is time-consuming. Furthermore, the expertise director is needed to direct the recording processes and then the interactive behaviors between the teacher and students can be recorded well. Hence, the automatic classroom director systems are required to develop. The current researches about the automatic classroom director systems often focus the broadcasting quality by analyzing the visual contents and sound information to switch the teacher, students, and slides scenes with several cameras. Furthermore, the face detection and gesture recognition are applied to capture the teacher’s close-up view during the course. However, these systems don’t record the interactive behaviors, e.g., to raise a question (raise hand or stand up), assigned question and answer (teacher’s pointing action). To overcome the above mentioned problems, in this study we proposed an intelligence classroom director system that can detect and identify the interactive behaviors of raising a question and assigned question and answer. The system features of the proposed intelligent classroom director system consists of operation switching between automatic and semi-automatic modes that is controlled by using the IR pointer and robust detection and identification of the interactive behaviors between teacher and students. In automatic mode, the system can track teacher position and identify the event of raising a question from students. In semi-automatic mode, the teacher can assign a student to answer question with the assistance of IR pointer. The IR pointer cam emit a flashing infrared light image pattern with fixed frequency (10 Hz). By recognizing the specified infrared light image pattern, the position of emitted student can be located and then the close-up view can be captured and recorded by controlling the PTZ cameras. The simulation results show the accuracy of identifying the events of raising a question and assigned question and answer can approach 90% and the efficiency is about 15 fps.

摘要
ABSTRACT
致謝
目錄
圖目錄
第一章 簡介
第一節 研究動機
第二節 相關研究
第三節 研究貢獻
第四節 系統架構
第二章 發言事件偵測
第一節 移動物件偵測
第二節 舉手起立偵測
第三章 紅外線物件3D定位
第一節 紅外線影像偵測
第四章 實驗結果
第一節 發言行為偵測
第二節 移動干擾測試
第三節 移動干擾測試
第五章 結論與未來發展
參考文獻

[1] MOOCs, http://courses.taiwanmooc.org/ ,Taiwan磨課師
[2] 微型教室設置廠商,http://www.mutec.com.tw/front/bin/home.phtml,秦強實業有限公司
[3] 導播系統,http://www.shineoncast.com,北京中廣上洋科技有限公司
[4] 導播系統,http://www.whzgsy.com/index.asp,武漢中廣上洋科技有限公司
[5] D.W. Huang, A Research of Off-line Automatic Virtual Director for Lecture Video, 國立臺灣師範大學資訊工程研究所碩士論文, 2009.
[6] Y.Y. Kuo, Automated Lecture Recording System, 國立臺灣師範大學資訊工程研究所碩士論文, 2011.
[7] H.C. Liao, Virtual Audience Cameraman System, 國立臺灣師範大學資訊工程研究所碩士論文, 2014.
[8] Y.T. Tsai, An Automatic Virtual Cameraman System for Audience, 國立臺灣師範大學資訊工程研究所碩士論文, 2013
[9] A. Mathew and V. K. Asari, “Local Region Statistical Distance Measure for Tracking in Wide Area Motion Imagery,” IEEE International Conference on Systems, Man and Cybernetics, pp. 248-253, 2012.
[10] C. C. Lien, J. C. Wang and Y. M. Jiang, “Multi-Mode Target Tracking on a Crowd Scene,” Third International Conference on International Information Hiding and Multimedia Signal Processing, Vol. 02, pp. 427-430, 2007.
[11] A. Elgammal, D. Harwood and L. Davis, “Non-parametric Model for Background Subtraction,” European Conference on Computer Vision, pp. 751-767, 2000.
[12] X. Dai and S. Khorram, “Performance of Optical Flow Techniques,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 236-242, 1992.
[13] R. Jain, W. Martin and J. Aggarwal, “Segmentation Through the Detection of Changes due to Motion,” Compute Graph and Image Processing, Vol. 11, pp. 13-34, 1979.
[14] C.C. Lien, C.T. Hsieh and M.H Tsai, “Vehicle Counting without Background Modeling,” Journal of Marine Science and Technology, Vol. 21, No. 6, Dec. 2013, pp. 631-638.