Adaptive Quotient Image with 3D Generic Elastic Models for Pose and Illumination Invariant Face Recognition
Large pose and illumination variations are very challenging for face recognition. In this paper, we address this challenge by combining an Adaptive Quotient Image method with 3D Generic Elastic Models (AQI-GEM). Frontal, neutral light face is re-rendered virtually under varying illumination conditions by AQI. Nearly accurate 3D models are constructed from each re-rendered image by GEM so as to virtually synthesize images under varying poses and illumination conditions. Pose-specific metrics are learnt for recognition. Experiments on MultiPIE demonstrate that it outperforms state-of-the-art face recognition methods, with much simpler parameter tuning, and much less training data.
KeywordsFace recognition Pose and illumination Face re-rendering 3D face construction
Unable to display preview. Download preview PDF.
- 1.Heo, J.: Generic Elastic Models for 2D Pose Synthesis and Face Recognition. Ph.D thesis, Department of Electrical and Computer Engineering, Carnegie Mellon University (2009)Google Scholar
- 7.Milborrow, S., Nicolls, F.: Active shape models with sift descriptors and MARS. J. VISAPP 1(2), 5 (2014)Google Scholar
- 8.Xiong, X. De la Torre, F.: Supervised descent method and its applications to face alignment. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 532–539 (2013)Google Scholar
- 9.Taigman, Y., Yang, M., Ranzato, M.A., et al.: Deepface: Closing the gap to human-level performance in face verification. In: 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1701–1708 (2014)Google Scholar
- 12.Zhu, Z., Luo, P., Wang, X., et al.: Deep learning identity-preserving face space. In: 2013 IEEE International Conference on Computer Vision (ICCV), pp. 113–120 (2013)Google Scholar
- 13.Jung, J.Y.H., Yoo, B.I., Choi, C., et al.: Rotating Your Face Using Multi-task Deep Neural Network (2015)Google Scholar