Face Recognition Across Poses Using a Single 3D Reference Model

Gee-Sern HSU; Hsiao-Chia PENG; Ding-Yu LIN; Chyi-Yeu LIN

doi:10.1587/transinf.2014EDP7352

Face Recognition Across Poses Using a Single 3D Reference Model

Gee-Sern HSU, Hsiao-Chia PENG, Ding-Yu LIN, Chyi-Yeu LIN

Full Text Views

0

Share
Cite this

Summary :

Face recognition across pose is generally tackled by either 2D based or 3D based approaches. The 2D-based often require a training set from which the cross-pose multi-view relationship can be learned and applied for recognition. The 3D based are mostly composed of 3D surface reconstruction of each gallery face, synthesis of 2D images of novel views using the reconstructed model, and match of the synthesized images to the probes. The depth information provides crucial information for arbitrary poses but more methods are yet to be developed. Extended from a latest face reconstruction method using a single 3D reference model and a frontal registered face, this study focuses on using the reconstructed 3D face for recognition. The recognition performance varies with poses, the closer to the front, the better. Several ways to improve the performance are attempted, including different numbers of fiducial points for alignment, multiple reference models considered in the reconstruction phase, and both frontal and profile poses available in the gallery. These attempts make this approach competitive to the state-of-the-art methods.

Publication: IEICE TRANSACTIONS on Information Vol.E98-D No.6 pp.1238-1246

Publication Date: 2015/06/01

Publicized: 2015/02/24

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2014EDP7352

Type of Manuscript: PAPER

Category: Image Recognition, Computer Vision

Authors

Gee-Sern HSU
  National Taiwan University of Science and Technology
Hsiao-Chia PENG
  National Taiwan University of Science and Technology
Ding-Yu LIN
  National Taiwan University of Science and Technology
Chyi-Yeu LIN
  National Taiwan University of Science and Technology

Keyword

face recognition, face reconstruction, sparse representation classification

Cite this

Copy

Gee-Sern HSU, Hsiao-Chia PENG, Ding-Yu LIN, Chyi-Yeu LIN, "Face Recognition Across Poses Using a Single 3D Reference Model" in IEICE TRANSACTIONS on Information, vol. E98-D, no. 6, pp. 1238-1246, June 2015, doi: 10.1587/transinf.2014EDP7352.
Abstract: Face recognition across pose is generally tackled by either 2D based or 3D based approaches. The 2D-based often require a training set from which the cross-pose multi-view relationship can be learned and applied for recognition. The 3D based are mostly composed of 3D surface reconstruction of each gallery face, synthesis of 2D images of novel views using the reconstructed model, and match of the synthesized images to the probes. The depth information provides crucial information for arbitrary poses but more methods are yet to be developed. Extended from a latest face reconstruction method using a single 3D reference model and a frontal registered face, this study focuses on using the reconstructed 3D face for recognition. The recognition performance varies with poses, the closer to the front, the better. Several ways to improve the performance are attempted, including different numbers of fiducial points for alignment, multiple reference models considered in the reconstruction phase, and both frontal and profile poses available in the gallery. These attempts make this approach competitive to the state-of-the-art methods.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2014EDP7352/_p

Copy

@ARTICLE{e98-d_6_1238,
author={Gee-Sern HSU, Hsiao-Chia PENG, Ding-Yu LIN, Chyi-Yeu LIN, },
journal={IEICE TRANSACTIONS on Information},
title={Face Recognition Across Poses Using a Single 3D Reference Model},
year={2015},
volume={E98-D},
number={6},
pages={1238-1246},
abstract={Face recognition across pose is generally tackled by either 2D based or 3D based approaches. The 2D-based often require a training set from which the cross-pose multi-view relationship can be learned and applied for recognition. The 3D based are mostly composed of 3D surface reconstruction of each gallery face, synthesis of 2D images of novel views using the reconstructed model, and match of the synthesized images to the probes. The depth information provides crucial information for arbitrary poses but more methods are yet to be developed. Extended from a latest face reconstruction method using a single 3D reference model and a frontal registered face, this study focuses on using the reconstructed 3D face for recognition. The recognition performance varies with poses, the closer to the front, the better. Several ways to improve the performance are attempted, including different numbers of fiducial points for alignment, multiple reference models considered in the reconstruction phase, and both frontal and profile poses available in the gallery. These attempts make this approach competitive to the state-of-the-art methods.},
keywords={},
doi={10.1587/transinf.2014EDP7352},
ISSN={1745-1361},
month={June},}

Copy

TY - JOUR
TI - Face Recognition Across Poses Using a Single 3D Reference Model
T2 - IEICE TRANSACTIONS on Information
SP - 1238
EP - 1246
AU - Gee-Sern HSU
AU - Hsiao-Chia PENG
AU - Ding-Yu LIN
AU - Chyi-Yeu LIN
PY - 2015
DO - 10.1587/transinf.2014EDP7352
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E98-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 2015
AB - Face recognition across pose is generally tackled by either 2D based or 3D based approaches. The 2D-based often require a training set from which the cross-pose multi-view relationship can be learned and applied for recognition. The 3D based are mostly composed of 3D surface reconstruction of each gallery face, synthesis of 2D images of novel views using the reconstructed model, and match of the synthesized images to the probes. The depth information provides crucial information for arbitrary poses but more methods are yet to be developed. Extended from a latest face reconstruction method using a single 3D reference model and a frontal registered face, this study focuses on using the reconstructed 3D face for recognition. The recognition performance varies with poses, the closer to the front, the better. Several ways to improve the performance are attempted, including different numbers of fiducial points for alignment, multiple reference models considered in the reconstruction phase, and both frontal and profile poses available in the gallery. These attempts make this approach competitive to the state-of-the-art methods.
ER -