Top Banner
1 hael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein
22

1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

Dec 19, 2015

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

1Michael Bronstein 3D face recognition

Face recognition:New technologies, new challenges

Michael M. Bronstein

Page 2: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

2Michael Bronstein 3D face recognition

The coin that betrayed Louis XVI

Page 3: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

3Michael Bronstein 3D face recognition

Modern challenges

=?

Is this the same person?

Page 4: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

4Michael Bronstein 3D face recognition

+

GEOMETRIC(3D)

PHOTOMETRIC(2D)

What is a face?

=

Page 5: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

5Michael Bronstein 3D face recognition

What is more important: 2D or 3D?

+ =

Page 6: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

6Michael Bronstein 3D face recognition

What is more important: 2D or 3D?

+ =

Page 7: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

7Michael Bronstein 3D face recognition

Conclusion 1

3D data conceals valuable information about identity Less sensitive to external factors (light, pose, makeup)More difficult to forge

Page 8: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

8Michael Bronstein 3D face recognition

The curse of expressions

Page 9: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

9Michael Bronstein 3D face recognition

Is geometry sensitive to expressions?

A

B

A′

B′

EUCLIDEAN DISTANCES: |A B| |A′ B′|

Page 10: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

10Michael Bronstein 3D face recognition

Is geometry sensitive to expressions?

A

B

A′

B′

GEODESIC DISTANCES: d(A,B) d′(A′,B′)

Page 11: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

11Michael Bronstein 3D face recognition

Conclusion 2

Extrinsic (Euclidean) geometry is sensitive to expressionsIntrinsic (Riemannian) geometry is insensitive to expressionsExpression-invariant face recognition using intrinsic

geometry

-60 -40 -20 0 20 40 600

0.2

0.4

0.6

0.8

1

ERROR DISTRIBUTION

Page 12: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

12Michael Bronstein 3D face recognition

Mapmaker’s nightmare

SPHERE(RIEMANNIAN)

PLANE(EUCLIDEAN)

A

B

A′

B′

d(A,B) |A′ B′|

Find a planar map of the Earth which preserves the geodesic distances in the best way

Page 13: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

13Michael Bronstein 3D face recognition

Isometric embedding

RIEMANNIAN EUCLIDEAN

A B A′ B′

EMBEDDING

Expression-invariant representation of face = canonical form

Page 14: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

14Michael Bronstein 3D face recognition

A remark from Gauss

Result: the embedding is only approximately

isometric, and therefore, introduces an error.

Carl Friedrich Gauss  (1777-1855)

Theorema Egregium (Remarkable Theorem):

A face has non-zero curvature, therefore, it is

not isometric to the plane.

Page 15: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

15Michael Bronstein 3D face recognition

How to canonize a person?

3D SURFACE ACQUISITION

SMOOTHING CANONIZATIONCROPPING

Page 16: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

16Michael Bronstein 3D face recognition

Examples of canonical forms

Page 17: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

17Michael Bronstein 3D face recognition

ORIGINAL SURFACES CANONICAL FORMS

Canonical forms

MichaelAlex

Page 18: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

18Michael Bronstein 3D face recognition

Telling identical twins apart

MichaelAlex

Page 19: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

19Michael Bronstein 3D face recognition

Page 20: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

20Michael Bronstein 3D face recognition

CAMERA

PROJECTOR

MONITOR

CARD READER

Page 21: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

21Michael Bronstein 3D face recognition

SCANNED FACE

CANONICAL FORM

DISTANCES

Page 22: 1 Michael Bronstein 3D face recognition Face recognition: New technologies, new challenges Michael M. Bronstein.

22Michael Bronstein 3D face recognition

Towards more accurate recognition

Embed one surface into another insteadof using a common embedding space

Avoid representation error

Beautiful theory: related to the Gromov-Hausdorff metric