Automatic 3-D face model adaptation for model-based coding of videophone sequences

摘要: For model-based coding of videophone sequences at very low bit rates, an algorithm for the automatic adaptation a generic 3-D face model to individual faces in is presented. The adapted facial features person-namely, eye centers, corners, eyelids, irises, mouth center, lips, eyebrows, nose, and chin cheek contours. First, these are estimated from sequence. Then, using features, face. Applying presented Akiyo, Claire, Miss America, all required successfully estimated, models adapted. distance between manually determined feature points image projections corresponding vertices onto plane 1.5 3.5 pels.