Indexing and Classification of News Video Articles by Speech, Character and Image Recognition

10 Citations 17 References
  • ARIKI Y.
    Faculty of Science and Technology, Ryukoku University
  • SUGIYAMA Y.
    Faculty of Science and Technology, Ryukoku University
  • ISHIKAWA N.
    Faculty of Science and Technology, Ryukoku University
  • TERANISHI T.
    Faculty of Science and Technology, Ryukoku University
  • SAKURAI M.
    Faculty of Science and Technology, Ryukoku University

Bibliographic Information

Other Title
  • ニュース映像中の記事に対する音声・文字・映像を用いた索引付けと分類

Search this article

Description

We are studying about article classification of TV news video by extracting indices from the character, speech and image information after segmenting news video into individual article. Articles are automatically extracted on the detected scene cut points, according to a loop syntax that each article starts from main studio, followed by field reportings and ends at the main studio again. We obtained 99.2% of news article extraction rate for 30 days news video data. From the news telop, characters are recognized and words are extracted as indices. Using these indices, articles are classified into 10 categories. For the same TV news data, article classification rate was 62.5%. From the news speech, keywords are extracted as indices and articles are classified into 10 categories. For the TV news data, the article classification rate was 50.7%. From the images, human faces are extracted, tracked and recognized by a subspace method.

Journal

  • Technical report of IEICE. PRMU

    Technical report of IEICE. PRMU 96 (385), 31-38, 1996-11-22

    The Institute of Electronics, Information and Communication Engineers

Citations (10)*help

See more

References(17)*help

See more

Keywords

Details 詳細情報について

  • CRID
    1570572702486582656
  • NII Article ID
    110003274578
  • NII Book ID
    AN10541106
  • Text Lang
    ja
  • Data Source
    • CiNii Articles

Report a problem

Back to top