Generating Facial Expressions from Neutral Geometry Based on a Local Shape Model

In this paper, a method for generating 3D facial expressions from a specified neutral geometry based on a local shape model and a small set of training data has been presented. The facial geometry is pre-segmented into anatomically meaningful regions, and deformation parameters are estimated for each region based on data fit and boundary consistency. After approximating the neutral form, the expressions are synthesized and transferred back to the initial geometry while preserving the anatomical structure. The proposed method ensures high expressiveness and consistency between expressions while preserving individual facial features. Experimental results demonstrate its superiority over existing methods under data-limited conditions.