A Fourier series based expression deformation model for 3D face recognition

This paper presents a Fourier series based expression deformation model for 3D face recognition. Given a set of training 3D face scans with sufficient facial expressions, these face scans are first preprocessed and represented as a series of Fourier series coefficients. Then, the shape residues between the non-neutral and neutral face scans of the same subject are calculated. These residues are supposed to contain the expression deformation patterns and PCA is applied to learn these patterns. The eigenvector with top eigenvalue in the generated lower dimensional subspace of PCA is then used to build the expression deformation model. Experimental results show the feasibility and merits of the proposed expression deformation model in the recognition scenario.