Image mapped 3D model based coding for videoconferencing type applications

One of the important reasons for model based coding scheme to be attractive for encoding videoconferencing type sequences at very low bit rates is the possibility of using a single generic facial model, and appropriately adding the identity of the subject in the sequence, In this paper, we present a method of adding the identity of the subject to the model using multiple views of the subject that allows synthesis of a perceptually equivalent version of the input sequence.