|
概要
|
In most of the current teleconferencing systems, a conference room is taken in one fixed video camera. For that reason, a face image of speakers becomes small. To take both full view images of a conference room and face images of speakers, it is required to develop a multi camera system with human operations. In this paper, we develop a new video camera for a teleconferencing without human operations in above mentioned situations. The proposed video camera is constructed with 3 NTSC video cameras and outputs the images which composed with panoramic images generated with images from these video cameras and the facial images of speakers. In this paper, Active Contour Models for panoramic images is adopted and directions of speakers are estimated by voice and hardware realization the system is mentioned.
|