臺灣博碩士論文加值系統

　　在這網路普及的年代，隨著網路頻寬不斷地提昇，大家開始會透過網路互相分享資訊，例如：文章、照片和影片等，其中很多學校或老師會將教學影片分享在網路上，目的是為了能讓學生方便複習，或是作為遠距教學使用。
　　可是通常錄製一堂課的影片時間約為一小時，當學生想要利用影片複習課程中的某一片段，或是想從影片中找尋課程內容的資訊時，往往要將影片從頭到尾的搜尋一遍，這樣不僅耗時，也有可能在搜尋過程中不小心遺漏了，因此本論文將探討如何從教學影片中擷取課程資訊的內容，並利用此內容作為影片分段的依據。
　　現今的教學環境，老師為了更清楚地表達課程內容，通常都會搭配投影片上課，因此，我們可以從影片中找到投影片的區域，並擷取出來，然後偵測投影片的切換，其時間點作為影片分段的依據，這樣學生不僅可以透過投影片取得課程內容資訊，也可以快速地找到想要複習的片段。
　　本論文分成兩個研究課題：（1）找出投影片區域（2）投影片切換偵測。

In this era, network is wildly used, network bandwidth is improved, many information are shared, such as articles, pictures and video clips via internet; schools or teachers also share lecture video on internet, in order to let student review what they taught easily and for distance education as well.
Usually, it takes 1 hour for recording a lesson, when students intent to review a typical fragment or want to find specific resource of the lesson in the video, they are forced to look over entire video. In this way, students spend more time and possibly miss up the content they need. Therefore, this thesis is going to probe how to capture specific content in the video and use it as the basis for the video segments.
The teaching environment in nowadays, teachers like to use slide in the class, it makes the lesson more vivid and clear. We can define the slide region in the video, and capture it, then detect the transition of the slide to be the basis for the video segments. Then students can review the lesson and also find the necessary fragment immediately.
There are two research topics in the thesis: (1) slide region location. (2) slide transition detection.

中文摘要 i
英文摘要 iii
誌謝 v
目錄 vi
表目錄 viii
圖目錄 ix
一、 緒論 1
1.1 研究動機 1
1.2 研究議題 2
1.3 相關研究 2
1.4 論文架構 7
二、 研究方法 8
2.1擷取畫面中的投影片區域 9
2.1.1 影像二值化 11
2.1.2 邊緣偵測 16
2.1.3 找出投影片邊界 18
2.1.4 定義投影片區域 23
2.1.5 投影片擷取 24
2.2 投影片切換偵測 25
2.2.1 切割區塊 27
2.2.2 計算區塊平均灰階值 27
2.2.3 計算相鄰畫面灰階差 28
2.2.4 統計大於門檻值個數 28
2.2.5 偵測投影片的切換 29
2.3投影片切換偵測結果分類 29
2.3.1 單張投影片的切換 29
2.3.2 連續投影片的切換 30
2.4 去除偽投影片的切換 31
三、 實驗結果與討論 33
3.1 實驗結果 33
3.2 討論 36
3.2.1遺漏的投影片切換 36
3.2.2偽投影片切換 37
3.3 方法比較 38
四、 結論和未來展望 41
4.1 結論 41
4.2 未來展望 41
參考文獻 42

[1] J.S. Boreczky and L.A. Rowe. “Comparison of Video Shot Boundary Detection Techniques,” Journal of Electronic Imaging Volume 5, Issue 2, pp.122-128, April, 1996.

[2] C.-W. Ngo, T.-C. Pong and T.S. Huang. “Detection of slide transition for topic indexing,” Proceedings of the IEEE international conference on multimedia and expo, ICME’02. IEEE, pp. 533-536, 2002.

[3] G. Schroth, N.-M. Cheung, E. Steinbach and B. Girod. “Synchronization of presentation slides and lecture videos using bit rate sequences,” In 18th IEEE International Conference on Image Processing, pp.925-928, December, 2011.

[4] H.J. Jeong, T.-E. Kim, H.G. Kim and M.H. Kim. “Automatic detection of slide transitions in lecture videos,” In Multimedia Tools and Applications, Volume 74, Issue 18, pp.7537-7554, June, 2015.

[5] D.G. Lowe. “Distinctive Image Features from Scale-Invariant Keypoints,” International Journal of Computer Vision, Volume 60, Issue 2, pp.91-110, November, 2004.

[6] N. Otsu. “A Threshold Selection Method from Gray-Level Histograms,” IEEE Transactions on Systems, Man, and Cybernetics, Volume 9, Issue 1, pp.62-66, January, 1979.

[7] R.C. Gonzalez and R.E. Woods, Digital Image Processing, Third Edition, Pearson, 2008.

[8] P.V.C. Hough. “Method and means for recognizing complex patterns,” U.S. 3069654A, 1962.

[9] R.O. Duda and P.E. Hart. “Use of the hough transformation to detect lines and curves in pictures,” Communications of the ACM, Volume 15 Issue 1, pp.11–15, January, 1972.

[10] “Biostatistics” https://www.youtube.com/watch?v=ClCzrVTF5ic, Published on Nov 13, 2012.

[11] “CS321n” https://www.youtube.com/watch?v=2uiulzZxmGg, Published on May 14, 2016.

[12] “Kevin” https://www.youtube.com/watch?v=kiwwEn3LvcQ, Published on Oct 21, 2013.

[13] “Milian” https://www.youtube.com/watch?v=RXTfC-35Nx8, Published on Oct 11, 2015.

[14] “Science” https://www.youtube.com/watch?v=P_KPooIWQJ8, Published on Jun 4, 2014.