| 摘要 |
The system for videoconferencing with automatic speaker detection and tracking comprises first means (21) to pan and zoom on a speaker automatically, to attend a "head and shoulder" mode, provided that second means (20) detect that a potential speaker rises his/her hand to initiate the tracking. The tracking operates on a "first-come, first-served" basis, requiring other speakers to wait for their turn. The schedule for the speakers can also be changed interactively by the person who chairs the meeting. The system can be implemented by pure software and runs real-time. It is only limited by the panning speed of the step motors moving the camera. |
