Abstract
Large pose variation is considered a major research concern due to its significant impact on the performance of face recognition systems. In this paper, we present a new efficient Machine-Learning Based System that identifies a person's face in 3D under Large pose variation. The proposed system can automatically detect and recognize people under translations and rotation by leveraging the anthropometric methodology and geometric descriptors. We evaluated our system on three well-known 3D face databases: 3DFPE, GAVAB, and 3DFRAV. These databases have a considerable change in the position of the face, even for the same individual. Additionally, we use the ICP algorithm to align the face to the front view. Our experimental simulation results show that the proposed system posed great robustness for significant changes in the pose and proved its advantages over state-of-the-art systems in terms of identification accuracy, specificity, and sensitivity.