Facial Feature Detection
—
filed under:
Biometric Person Recognition,
Computer Vision
Downhill Simplex method
This system tracks a face in a video sequence using a 3D face model. The shape of the person to be tracked is learned during a training phase, as well as an appearance model for 20 features located on the eyebrows, eyes and mouth. Optimization is performed by the downhill simplex method. It minimizes the difference between the observed features
