Towards 3D Face Recognition in the Real: A Registration-Free Approach Using Fine-Grained Matching of 3D Keypoint Descriptors

Handle URI:
http://hdl.handle.net/10754/566183
Title:
Towards 3D Face Recognition in the Real: A Registration-Free Approach Using Fine-Grained Matching of 3D Keypoint Descriptors
Authors:
Li, Huibin; Huang, Di; Morvan, Jean-Marie; Wang, Yunhong; Chen, Liming
Abstract:
Registration algorithms performed on point clouds or range images of face scans have been successfully used for automatic 3D face recognition under expression variations, but have rarely been investigated to solve pose changes and occlusions mainly since that the basic landmarks to initialize coarse alignment are not always available. Recently, local feature-based SIFT-like matching proves competent to handle all such variations without registration. In this paper, towards 3D face recognition for real-life biometric applications, we significantly extend the SIFT-like matching framework to mesh data and propose a novel approach using fine-grained matching of 3D keypoint descriptors. First, two principal curvature-based 3D keypoint detectors are provided, which can repeatedly identify complementary locations on a face scan where local curvatures are high. Then, a robust 3D local coordinate system is built at each keypoint, which allows extraction of pose-invariant features. Three keypoint descriptors, corresponding to three surface differential quantities, are designed, and their feature-level fusion is employed to comprehensively describe local shapes of detected keypoints. Finally, we propose a multi-task sparse representation based fine-grained matching algorithm, which accounts for the average reconstruction error of probe face descriptors sparsely represented by a large dictionary of gallery descriptors in identification. Our approach is evaluated on the Bosphorus database and achieves rank-one recognition rates of 96.56, 98.82, 91.14, and 99.21 % on the entire database, and the expression, pose, and occlusion subsets, respectively. To the best of our knowledge, these are the best results reported so far on this database. Additionally, good generalization ability is also exhibited by the experiments on the FRGC v2.0 database.
KAUST Department:
Visual Computing Center (VCC)
Publisher:
Springer Nature
Journal:
International Journal of Computer Vision
Issue Date:
12-Nov-2014
DOI:
10.1007/s11263-014-0785-6
Type:
Article
ISSN:
09205691
Appears in Collections:
Articles; Visual Computing Center (VCC)

Full metadata record

DC FieldValue Language
dc.contributor.authorLi, Huibinen
dc.contributor.authorHuang, Dien
dc.contributor.authorMorvan, Jean-Marieen
dc.contributor.authorWang, Yunhongen
dc.contributor.authorChen, Limingen
dc.date.accessioned2015-08-12T09:31:30Zen
dc.date.available2015-08-12T09:31:30Zen
dc.date.issued2014-11-12en
dc.identifier.issn09205691en
dc.identifier.doi10.1007/s11263-014-0785-6en
dc.identifier.urihttp://hdl.handle.net/10754/566183en
dc.description.abstractRegistration algorithms performed on point clouds or range images of face scans have been successfully used for automatic 3D face recognition under expression variations, but have rarely been investigated to solve pose changes and occlusions mainly since that the basic landmarks to initialize coarse alignment are not always available. Recently, local feature-based SIFT-like matching proves competent to handle all such variations without registration. In this paper, towards 3D face recognition for real-life biometric applications, we significantly extend the SIFT-like matching framework to mesh data and propose a novel approach using fine-grained matching of 3D keypoint descriptors. First, two principal curvature-based 3D keypoint detectors are provided, which can repeatedly identify complementary locations on a face scan where local curvatures are high. Then, a robust 3D local coordinate system is built at each keypoint, which allows extraction of pose-invariant features. Three keypoint descriptors, corresponding to three surface differential quantities, are designed, and their feature-level fusion is employed to comprehensively describe local shapes of detected keypoints. Finally, we propose a multi-task sparse representation based fine-grained matching algorithm, which accounts for the average reconstruction error of probe face descriptors sparsely represented by a large dictionary of gallery descriptors in identification. Our approach is evaluated on the Bosphorus database and achieves rank-one recognition rates of 96.56, 98.82, 91.14, and 99.21 % on the entire database, and the expression, pose, and occlusion subsets, respectively. To the best of our knowledge, these are the best results reported so far on this database. Additionally, good generalization ability is also exhibited by the experiments on the FRGC v2.0 database.en
dc.publisherSpringer Natureen
dc.subject3D keypoint descriptorsen
dc.subjectExpression, pose and occlusionen
dc.subjectFine-grained matchingen
dc.subjectRegistration-free 3D face recognitionen
dc.titleTowards 3D Face Recognition in the Real: A Registration-Free Approach Using Fine-Grained Matching of 3D Keypoint Descriptorsen
dc.typeArticleen
dc.contributor.departmentVisual Computing Center (VCC)en
dc.identifier.journalInternational Journal of Computer Visionen
dc.contributor.institutionSchool of Mathematics and Statistics, Xi’an Jiaotong University, Xi’anShaanxi, Chinaen
dc.contributor.institutionBeijing Center for Mathematics and Information Interdisciplinary Sciences (BCMIIS)Beijing, Chinaen
dc.contributor.institutionLaboratory of Intelligent Recognition and Image Processing, School of Computer Science and Engineering, Beihang UniversityBeijing, Chinaen
dc.contributor.institutionDépartement de Mathématiques, Université Claude Bernard Lyon 1Lyon, Franceen
dc.contributor.institutionDépartement de Mathématiques et Informatique, UMR CNRS 5205, Ecole Centrale LyonLyon, Franceen
kaust.authorMorvan, Jean-Marieen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.