Expression-Compensated 3D Face Recognition with Geodesically Aligned Bilinear Models

In this paper, we present a technique for addressing 3-dimensional face recognition in presence of facial expressions using a bilinear model. The bilinear model allows decoupling the impact of identity and expression on face appearance and encoding their contribution in separate control parameters. This is achieved by first representing faces as parametric surface models described by a fixed length parameter vector. A generic face model is fitted to each face based on a novel technique that relies on geodesic distances to find implicitly corresponding facial landmarks between the model and the face in hand. Model parameters are then used for bilinear decomposition. The experimental results on the publicly available BU-3DFE face database demonstrate the effectiveness of our technique.