Characterizing smiles in the context of video phone data compression
Li-an Tang, Thomas S. Huang · 1996
We propose an approach to characterizing human smiles based on the principal component analysis of the facial motion parameters. Motion vectors of a set of feature points are derived from neutral and smiling face images and a feature space is built based on those motion vectors. The principal component analysis technique is used for dimension reduction and some preliminary results of smile characterization are obtained.