Model-Based Coding of 3D Head Sequences
L. Granai, T. Vlachos, Miroslav Hamouz, J. Rafael Tena, Thomas Idwal Davies · 2007
This paper describes a novel technique for model-based coding of 3D head and head-and-shoulders sequences. First, 3D frames are analyzed and registered using a 3D face model, fixed and known also at the decoder side. Then, shape and texture information are compressed in a lossy fashion in order to reduce spatial and temporal redundancy. Results show that efficient compression can be achieved for this type of data.