IMPROVED 3-D FACIAL REPRESENTATION THROUGH STATISTICAL SHAPE MODEL

Abstract

This paper describes an improved 3-D facial representation method capable of modeling different types of faces without any constraint from expression, age, gender, or ethnic origin. Using the proposed technique, a 3-D face can be represented by a low dimensional shape space vector (SSV) of the statistical shape model (SSM), which is calculated through a model-based surface registration process. This model-based surface registration method consists of two major processing stages, model building and hierarchical model fitting. A statistical shape model is first built using a set of training faces. Then the model is deformed to match the new face by a modified iterative closest point (ICP)scheme. The experimental results on real 3-D facial data show that the proposed method can reasonably interpret the articulation of 3-D faces.