3D face verification using a free-parts approach

This paper proposes a novel approach to 3D face verification which divides the 3D face into separate parts. This method, termed 3D free-parts, considers each part of the face independently and consequently the spatial relationship is discarded for the purpose of obtaining many observations from each face. Experiments illustrate the validity of the face verification system where the distribution of features are modelled robustly using Gaussian Mixture Models. This approach demonstrates a significant improvement over the eigenfaces approach, lowering the false rejection rate from 9.83% to 4.48% at a false acceptance rate of 0.1%, in tests conducted on 3D face data from the face recognition grand challenge database.