AUGMENTED REALITY
Wearables
Perspective registration
Video Augmentation
The WristCam

RECENT PUBLICATION
Patents
Journals
Conferences

RESOURCES
Media Tech Resources

RETURN TO:
Userport homepage
York homepage

Our object-based analysis/synthesis coding scheme uses muscle-based facial models to represent a videophone picture. As in other object-based schemes, picture content is interpreted in terms of a wire-frame head-and-shoulders model, whose facets are textures mapped from an original "neutral" picture of the person in view. At the transmitter, incoming frames are analysed to deduce appropriate model parameters, and at the receiver the deformed model is resynthesized from those parameters plus extra information in the eye and mouth regions. In this way moving pictures are transmitted at below 10 kbit/s.

Our main contribution was a conjugate-gradient descent algorithm for estimating the model parameters.