3D face alignment without correspondences

Zsolt Sánta, Z. Kato

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

A novel correspondence-less approach is proposed to find a thin plate spline map between a pair of 3D human faces represented by triangular surface meshes. The proposed method works without landmark extraction and feature correspondences. The aligning transformation is simply found by solving a system of nonlinear equations. Each equation is generated by integrating a non-linear function over the surfaces represented as fuzzy sets of triangles. We derive approximating formulas for the efficient computation of these integrals. Based on a series of comparative tests on a standard 3D face dataset, our triangular mesh-based algorithm outperforms state of the art methods in terms of computing time while maintaining accuracy.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2016 Workshops, Proceedings
PublisherSpringer Verlag
Pages521-535
Number of pages15
Volume9914 LNCS
ISBN (Print)9783319488806
DOIs
Publication statusPublished - 2016
Event14th European Conference on Computer Vision, ECCV 2016 - Amsterdam, Netherlands
Duration: Oct 8 2016Oct 16 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9914 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other14th European Conference on Computer Vision, ECCV 2016
CountryNetherlands
CityAmsterdam
Period10/8/1610/16/16

Fingerprint

Alignment
Correspondence
Face
Thin-plate Spline
Triangular Mesh
System of Nonlinear Equations
Landmarks
Fuzzy sets
Nonlinear Function
Nonlinear equations
Splines
Fuzzy Sets
Triangular
Triangle
Mesh
Series
Computing
Human
Standards

Keywords

  • 3D face
  • Alignment
  • Correspondenceless
  • Thin plate spline

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Sánta, Z., & Kato, Z. (2016). 3D face alignment without correspondences. In Computer Vision – ECCV 2016 Workshops, Proceedings (Vol. 9914 LNCS, pp. 521-535). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9914 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-48881-3_36

3D face alignment without correspondences. / Sánta, Zsolt; Kato, Z.

Computer Vision – ECCV 2016 Workshops, Proceedings. Vol. 9914 LNCS Springer Verlag, 2016. p. 521-535 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9914 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sánta, Z & Kato, Z 2016, 3D face alignment without correspondences. in Computer Vision – ECCV 2016 Workshops, Proceedings. vol. 9914 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9914 LNCS, Springer Verlag, pp. 521-535, 14th European Conference on Computer Vision, ECCV 2016, Amsterdam, Netherlands, 10/8/16. https://doi.org/10.1007/978-3-319-48881-3_36
Sánta Z, Kato Z. 3D face alignment without correspondences. In Computer Vision – ECCV 2016 Workshops, Proceedings. Vol. 9914 LNCS. Springer Verlag. 2016. p. 521-535. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-48881-3_36
Sánta, Zsolt ; Kato, Z. / 3D face alignment without correspondences. Computer Vision – ECCV 2016 Workshops, Proceedings. Vol. 9914 LNCS Springer Verlag, 2016. pp. 521-535 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{68ee57a18bb348e691357ecdceaac48d,
title = "3D face alignment without correspondences",
abstract = "A novel correspondence-less approach is proposed to find a thin plate spline map between a pair of 3D human faces represented by triangular surface meshes. The proposed method works without landmark extraction and feature correspondences. The aligning transformation is simply found by solving a system of nonlinear equations. Each equation is generated by integrating a non-linear function over the surfaces represented as fuzzy sets of triangles. We derive approximating formulas for the efficient computation of these integrals. Based on a series of comparative tests on a standard 3D face dataset, our triangular mesh-based algorithm outperforms state of the art methods in terms of computing time while maintaining accuracy.",
keywords = "3D face, Alignment, Correspondenceless, Thin plate spline",
author = "Zsolt S{\'a}nta and Z. Kato",
year = "2016",
doi = "10.1007/978-3-319-48881-3_36",
language = "English",
isbn = "9783319488806",
volume = "9914 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "521--535",
booktitle = "Computer Vision – ECCV 2016 Workshops, Proceedings",

}

TY - GEN

T1 - 3D face alignment without correspondences

AU - Sánta, Zsolt

AU - Kato, Z.

PY - 2016

Y1 - 2016

N2 - A novel correspondence-less approach is proposed to find a thin plate spline map between a pair of 3D human faces represented by triangular surface meshes. The proposed method works without landmark extraction and feature correspondences. The aligning transformation is simply found by solving a system of nonlinear equations. Each equation is generated by integrating a non-linear function over the surfaces represented as fuzzy sets of triangles. We derive approximating formulas for the efficient computation of these integrals. Based on a series of comparative tests on a standard 3D face dataset, our triangular mesh-based algorithm outperforms state of the art methods in terms of computing time while maintaining accuracy.

AB - A novel correspondence-less approach is proposed to find a thin plate spline map between a pair of 3D human faces represented by triangular surface meshes. The proposed method works without landmark extraction and feature correspondences. The aligning transformation is simply found by solving a system of nonlinear equations. Each equation is generated by integrating a non-linear function over the surfaces represented as fuzzy sets of triangles. We derive approximating formulas for the efficient computation of these integrals. Based on a series of comparative tests on a standard 3D face dataset, our triangular mesh-based algorithm outperforms state of the art methods in terms of computing time while maintaining accuracy.

KW - 3D face

KW - Alignment

KW - Correspondenceless

KW - Thin plate spline

UR - http://www.scopus.com/inward/record.url?scp=84996772743&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84996772743&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-48881-3_36

DO - 10.1007/978-3-319-48881-3_36

M3 - Conference contribution

AN - SCOPUS:84996772743

SN - 9783319488806

VL - 9914 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 521

EP - 535

BT - Computer Vision – ECCV 2016 Workshops, Proceedings

PB - Springer Verlag

ER -