Presentation 2006-05-25
Automatic speaker detection for videoconferencing
Takeshi SAITOH, Hayato KOBAYASHI, Ryosuke KONISHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper proposes an automatic speaker detection method with visual information for the videoconferencing. Recently, some papers reported the same aim using microphone array method. We pay attention the fact that the movement of the speaker's mouth is large compared with other participants'. Then we detect the mouth and compute its movement. After the person's face region is detected, the location of between the eyes, left and right eyes, and nose are detected. Finally, the location of the mouth is detected. The mouth tracking is applied the template matching with updating. We identify the speaker who is the large error. We took 16 simulated scenes of videoconferencing, and obtained all parts of all 42 persons. Moreover, the resulting detection rate was 100% (as the frame was 99.1%) and we confirm the validity of our method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Speaker detection / videoconferencing / mouth detection
Paper # PRMU2006-5,MI2006-5
Date of Issue

Conference Information
Committee PRMU
Conference Date 2006/5/18(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Automatic speaker detection for videoconferencing
Sub Title (in English)
Keyword(1) Speaker detection
Keyword(2) videoconferencing
Keyword(3) mouth detection
1st Author's Name Takeshi SAITOH
1st Author's Affiliation Tottori University()
2nd Author's Name Hayato KOBAYASHI
2nd Author's Affiliation Tottori University
3rd Author's Name Ryosuke KONISHI
3rd Author's Affiliation Tottori University
Date 2006-05-25
Paper # PRMU2006-5,MI2006-5
Volume (vol) vol.106
Number (no) 72
Page pp.pp.-
#Pages 6
Date of Issue