Loading...

Extraction of Important Events in Football Videos Using Video Summarization Techniques

Tavassolipour, Mostafa | 2011

562 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 42434 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Kasaei, Shohreh
  7. Abstract:
  8. Semantic video analysis and automatic concept extraction play an important role in several application including designing content-based search engines, video indexing, and video summarization. Since constructing an appropriate summarized video requires extraction of internal video concepts, video summarization is considered as an application of semantic analysis. The proposed system contains three main stages. In the first stage, the shot boundaries are detected, then using the hidden Markov model (HMM), the video is segmented into larger semantic units, called “play-break” sequences. In the next stage, several features are extracted from each of these units. Finally, in the last stage, in order to achieve high level semantic features (events and concepts), Bayesian networks are used. The basic part of the system is constructing the Bayesian network, for which the structure is estimated by using Chow-Liu tree. The joint distributions of random variables of the network are then modeled by applying the FGM family of Copulas. In fact, the extracted features of play-break sequences correspond to these random variables. Performance of the proposed system is evaluated on a large dataset of about 9 hours of soccer videos. The system is capable to detect seven different events in soccer videos; namely, goal, card, goal attempt, corner, foul, offside and non-highlights. Experimental results show the effectiveness and robustness of the proposed system on detecting the considered events.
  9. Keywords:
  10. Bayesian Network ; Copulas ; Semantic Analysis ; Video Summarization

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC