針對數位教學錄影影像之圖例擷取

Legend Extraction for e-Learning Video Stream

邱致瑋(Ji-Wen Chio)

指導教授 ： 陳淑媛

元智大學/資訊學院/資訊工程學系/碩士(2004年)

摘要

本論文主要是提出一針對數位教學錄影影像擷取圖例的新方法。一般而言，授課老師會在重點部分利用圖例加以說明，也會花較長時間講解。基於此一前提，我們提出一針對數位教學錄影影像擷取圖例的新方法，藉以協助學生加速獲知重點所在，並在關鍵部分反覆學習，以提升學生學習效率，如此方能達到善用數位教學環境的目的。
所提方法分成訓練和擷取兩部份。在訓練的部份，先利用場景偵測來去除取樣影格資料的重複性，以得較為精確且簡潔之訓練樣本。再利用取樣影格資料建構背景對應圖並記錄錄影影像中常用色系，以協助後續前景圖例擷取。在擷取的部份，主要利用前述背景資訊，將影格中前景部分擷取，再輔以區塊標記及幾何驗證技巧，將前景圖例擷取出來。實驗結果證實所提方法確實有效。

關鍵字

以內容為基礎之影像檢索 ； 圖例檢索 ； 圖例擷取 ； 數位學習 ； 錄影影像 ； 背景建構 ； 前景擷取

並列摘要

The thesis proposes a legend extraction method for e-Learning video stream. Teachers usually use legends to explain the key point and may take much time to explain. Hence, we propose a method to extract legends from e-Learning video stream so as to aid student focusing on the important part. In this way, e-Learning environment can improve student’s learning efficiency.
The proposed method consists of two phases: training and extraction. In the training phase, shot boundary detection is performed first to resample the video stream so as to obtain non-redundancy training set. On the basis of the training set, background map will be generated and frequent label set will be constructed, which in turn are used for later legend extraction. In the extraction phase, foreground map is first generated for each frame within the video stream according to the background map and frequent label set. On the basis of the foreground map, it is easy to extract legend object by region labeling and geometric verification. The effectiveness and practicability of the proposed method have been demonstrated by various experiments.

並列關鍵字

content-based image retrieval ； legend retrieval ； legend extraction ； e-Learning ； background map generation ； foreground extraction

參考文獻

[1] R.T. Colllin, A.J. Lipton and T. Kanade, “Introduction to the special section on video surveillance,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 8, pp. 745-746, 2000.

[2] R.J. Radke, S. Andra, O. Al-Kofahi, and B. Roysam, “Image change detection algorithms: A system survey,” IEEE Trans. Image Processing, vol. 14, no. 3, pp. 294-307, 2005.

[5] C. Yong, Y. Xu, and X. De, “A method for shot boundary detection with automatic threshold,” in Proc. IEEE Conf. Computers, Communications, Control and Power Engineering, vol. 1, pp. 582-585, 2002.

[7] H. Zhiwei, L. Jilin and L. Peihong, “New method of background update for video-based vehicle detection,” in Proc. IEEE Conf. Intelligent transportation systems, pp. 580-584, 2004.

[8] D.S. Lee, “Effective Gaussian mixture learning for video background subtraction,” IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 27, no. 5, pp. 827-832, 2005.