口唇動作と音声の共起に着目した被写体と話者の不一致検出--ニュース映像への適用と評価

Bibliographic Information

Other Title
  • 口唇動作と音声の共起に着目した被写体と話者の不一致検出 : ニュース映像への適用と評価(萌芽セッション,エンタテインメントのためのメディアとリアリティ)
  • コウシン ドウサ ト オンセイ ノ キョウキ ニ チャクモク シタ ヒシャタイ ト ワシャ ノ フイッチ ケンシュツ ニュース エイゾウ エ ノ テキヨウ ト ヒョウカ
  • Detection of Inconsistency between Face and Speaker Focusing on the Co-occurrence of Lip Motion and Audio : An Application to News Video and its Evaluation

Search this article

Description

Speech scenes in news videos contain a wealth of multimedia information, and are valuable as archived material. In order to extract speech scenes from news videos, there is an approach that uses the position and size of a face region. However, it is difficult to extract them with only the approach, since news videos contain scenes where the speakers are not the subjects such as in narration scenes. To solve this problem, we have been proposing a method to detect the inconsistency between face and speaker focusing on the co-occurrence of the lip motion and the speech. However, the evaluations for the proposed method were performed in an ideal condition without much noise. In this paper, we report the investigation on the performance of the proposed method not only with videos captured in ideal conditions but also with actual broadcasted news videos. Their results showed the effectiveness and the usefulness of our method.

Journal

Citations (1)*help

See more

References(11)*help

See more

Details 詳細情報について

Report a problem

Back to top