Coding of Facial Image Sequences by Model-Based Optical Flow

A model-based method for estimating the shapeand motion of 3D objects appearing in a videois described. This technique is used for modelbasedvideo coding #video compression#. Themethod is based on a new variant of optical #owand uses 3D computer graphics to represent anddisplay an object. Though the algorithm is general,this work concentrates on videos depictingthe human head and face because of its relevanceto videotelephony and teleconferencing. Rigidbody motion of the head and...