Synthetic natural hybrid video processings for virtual teleconferencing systems

In this paper, we propose powerful virtual image processing tools, which can be useful for designing new virtual teleconferencing systems with limited bandwidth requirements. We present an integrated approach that mixes real images with computer graphics in order to obtain an efficient and highly realistic model-based coding scheme, while offering new possibilities from the point of view of the ergonomics. In particular, we focus on original or extended existing algorithms to address the speakers’ representation problem via face cloning, and the virtual environment creation issue viavideo spatialization.