Automatically estimating number of scenes for rushes summarization

ABSTRACT

This paper describes our video summarization system using a model selection technique to estimate the optimal number of scenes for a summary. It uses a minimum description length as a model selection criterion and carries out two-stage estimation. First, we estimate the number of scenes in each shot, and then we estimate the number of scenes in a whole video clip. We model a set of scenes with a Gaussian mixture model, where the mixture component is assumed to represent one scene. Our system was evaluated in the TRECVID 2008 rushes summarization task, where the test video set was unedited materials provided by the BBC. Our scores were about the same as the average of all the participants for the eight evaluation measures.

REFERENCES

• 1: H. Akaike. A new look at the statistical model identification. IEEE Transactions on Automatic Control, AC-19(6):716--723, 1974.
• 2: J.-Y. Bouget. Pyramidal implementation of the lucas kanade feature tracker description of the algorithm. Technical report, Intel Corporation Microprocessor Research Labs, 2000.
• 3: J. Lin. Divergence measures based on the shannon entropy. IEEE Transactions on Information Theory, 37(1):145--151, 1991.
• 4: Z. Liu, E. Zavesky, D. Gibbon, B. Shahraray, and P. Haffner. AT&T research at TRECVID 2007. In TREC Video Retrieval Evaluation Online Proceedings, 2007.
• 5: B. Lucas and T. Kanade. An iterative image registration technique with an application to stereo vision. Proc. 7th International Joint Conference on Artificial Intelligence, pages 674--679, 1981.
• 6: T. Nakamura, Y. Miyamura, K. Shinoda, and S. Furui. TokyoTech's TRECVID2006 notebook. In TREC Video Retrieval Evaluation Online Proceedings, 2006.
• 7: T. Nakamura, K. Shinoda, and S. Furui. TokyoTech's TRECVID2007 notebook. In TREC Video Retrieval Evaluation Online Proceedings, 2007.
• 8: Intel Open Source Computer Vision Library. http://www.intel.com/research/mrl/research/opencv/.
• 9: Paul Over , Alan F. Smeaton , George Awad, The trecvid 2008 BBC rushes summarization evaluation, Proceedings of the 2nd ACM TRECVid Video Summarization Workshop, p.1-20, October 31-31, 2008, Vancouver, British Columbia, Canada [doi>10.1145/1463563.1463564]
• 10: Chen-Ming Pan , Yung-Yu Chuang , Winston H. Hsu, NTU TRECVID-2007 fast rushes summarization system, Proceedings of the international workshop on TRECVID video summarization, p.74-78, September 28-28, 2007, Augsburg, Bavaria, Germany [doi>10.1145/1290031.1290045]
• 11: Z. Pan and C.-W. Ngo. Moving-object detection, association, and selection in home videos. IEEE Transactions on Multimedia, 9(2):268--279, February 2007.
• 12: J. Rissanen. A universal prior for integers and estimation by minimum description length. Ann. Statist., 11(2):416--431, 1983.
• 13: K. Shinoda, K. Ishihara, S. Furui, and T. Mochizuki. Automatic score scene detection for baseball video. International Symposium on Large-Scale Knowledge Resources (LKR2008), pages 226--240, March 2008.
• 14: Alan F. Smeaton , Paul Over , Wessel Kraaij, Evaluation campaigns and TRECVid, Proceedings of the 8th ACM international workshop on Multimedia information retrieval, October 26-27, 2006, Santa Barbara, California, USA [doi>10.1145/1178677.1178722]
• 15: E. Spyrou, P. Kapsalas, G. Tolias, P. Mylonas, and Y. A. et al. The COST292 experimental framework for TRECVID 2007. In TREC Video Retrieval Evaluation Online Proceedings, 2007.
• 16: Feng Wang , Chong-Wah Ngo, Rushes video summarization by object and event understanding, Proceedings of the international workshop on TRECVID video summarization, p.25-29, September 28-28, 2007, Augsburg, Bavaria, Germany [doi>10.1145/1290031.1290035]

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.1 Content Analysis and Indexing

General Terms:
Algorithms, Performance

Keywords:
mdl, model selection, rushes, trecvid, video summarization