Paper Abstract and Keywords |
Presentation |
2006-02-20 14:15
Video Retrieval System with Query Generation Assistance Using Relevance Feedback Yasuhiko Fujisaki, Takamichi Miyata (Tokyo Tech.), Yasuhiro Inazumi (Kanagawa Univ.), Aki Kobayashi, Katsunori Yamaoka, Yoshinori Sakai (Tokyo Tech.) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In content based video retrieval, it is di±cult to obtain a query (not keywords) from a user who only has an ambiguous memory about what he wants. On the other hand, we have already proposed the user interface by using relevance feedback that achieves automatic query generation for still image retrieval. In this paper, we proposed a video retrieval system by introducing motion feature and motion histogram matching to our conventional system. The proposed system presents nine automatically generated videos for a user as candidates of the query. Then, a user selects video which are similar to what he wants. After user’s selection, the system creates new nine videos based on the video selected by user. This interaction process allows us to create the appropriate query easily. Experimental results show the e®ectiveness of our proposed system. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Video Retrieval System / Query Generation Assistance / Relevance Feedback / Feature Weighting / User Interface / / / |
Reference Info. |
IEICE Tech. Rep., vol. 105, no. 610, IE2005-270, pp. 95-100, Feb. 2006. |
Paper # |
IE2005-270 |
Date of Issue |
2006-02-13 (ITS, IE) |
ISSN |
Print edition: ISSN 0913-5685 |
Download PDF |
|