Recombination of 2D and 3D Images for Multimodal 2D + 3D Face Recognition

Exploiting the dependency between 2D and 3D information of a face at the early stage is more effective than at the later stage. Early fusion data contains richer information about the input biometric than the compressed features or matching scores. We propose an image recombination for face recognition that explores the dependency between modalities at the image level. Facial cues from 2D and 3D images are recombined into more discriminating data using PCA- and FLD-based recombination transforms. Our experimental results show that the proposed image recombination outperforms the pixel- and score-level fusions.