On Broadcasted Game Video Analysis: Event Detection, Highlight Detection, and Highlight Forecast
| Autor: | Yung-Chieh Chou, 周雍傑 |
|---|---|
| Rok vydání: | 2015 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 103 Efficient access of broadcasted game videos is urgently demanded due to the emergence of live streaming platforms. The popularity of game video streaming builds a big market showing commercial potentials and arising many technical challenges. In this thesis, we facilitate efficient access from two aspects: event detection and game highlight detection. By recognizing designated text displayed on screen when important events occur, we associate game events with time stamps, and accordingly develop an interface to facilitate direct access. For highlight detection, we jointly consider visual features, event features, and viewer’s behaviors to construct two highlight models, i.e., the psychophysiological approach based on the arousal model and the data-driven approach based on support vector machine (SVM). In addition, we propose an optimal subset selection based on the genetic algorithm for video sub-segment selection. The concatenated highlight segments enable compact game video presentation. To facilitate adaptive bitrate streaming, a highlight forecast model is built so that the streaming system can allocate higher bitrate for more important segments on the fly. Comprehensive experiments demonstrate effectiveness of the proposed methods from various perspectives. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|