Abstract—Up to date, many advances have been made to 2D
face recognition (2D FR) due to its broad range of applications
in security and commercial areas as well as in smart devices.
However, 2D FR is still quite vulnerable under unconstrained
conditions of the image acquisition process. To overcome 2D FR
limitations, researchers shift to 3D face recognition technology
but this technology is computationally expensive and
inapplicable to real-world face recognition systems. Multimodal
2D-3D face recognition can combine the strength of both 2D
and 3D modalities. In this paper a multimodal 2D-3D face
recognition approach has been proposed based on geometric
and textural characteristics of 2D and 3D modalities. The
conducted experiments show that the proposed approach
achieved promising results with illumination and head pose
variations. The performance is evaluated using the landmark
Bosphorus facial databas
Index Terms—2D-3D face recognition, geometric invariants,
local binary pattern (LBP), k-nearest neighbor (kNN).
The authors are with Universiti Putra Malaysia (UPM), Malaysia Gawed.
M. Nagi (e-mail: gnagi2000@yahoo.com)
[PDF]
Cite:Gawed M. Nagi, Rahmita Rahmat, Muhamad Taufik, and Fatimah Khalid, "Multimodal 2D-3D Face Recognition," International Journal of Future Computer and Communication vol. 2, no. 6, pp. 687-691, 2013.
