Narrative Structure Analysis of Lecture Video with Hierarchical Hidden Markov Model for E-Learning
In E-learning, structure analysis of lecture video is the first step for effective and efficient indexing, browsing and retrieval. A hierarchical model of narrative structure for lecture video is introduced. The root is lecture video; the next is layer of narrative elements conveying meaningful information in semantics; then is narrative features layer closely to both visual and auditory physical features. A framework is proposed to analyze narrative structure. Extraction of narrative features is described as well. Hierarchical hidden Markov model is introduced to determine the parameters and detect narrative elements automatically.
KeywordsAudio Signal Dynamic Bayesian Network Narrative Structure Lecture Video Narrative Element
Unable to display preview. Download preview PDF.
- 1.Dorai, C., Oria, V., Neelavalli, V.: Structuralizing educational videos based on presentation content. In: ICME 2003 (2003)Google Scholar
- 4.Murphy, K., Paskin, M.: Linear time inference in hierarchical hidden Markov models. In: Proceedings of Neural Information Processing Systems (2001)Google Scholar
- 5.Naphade, M., Huang, T.: Discovering recurrent events in video using unsupervised methods. In: Proc. Intl. Conf. Image Processing (2002)Google Scholar
- 6.Xie, L., Chang, S.-F., Divakaran, A., Sun, H.: Unsupervised discovery of multilevel statistical video structures using hierarchical hidden Markov models. In: Interational Conference on Multimedia and Expo. (2003)Google Scholar