Methods for Quality Determination of Papillary Lines in ...

Methods for Quality Determination of Papillary Lines in Fingerprints

NIST – Biometric Quality Workshop II, November 7-8, 2007 , Gaithersburg, USA

Martin Drahansky ( www.fit.vutbr.cz/~drahan )Brno University of Technology, Faculty of Information Technology

Bozetechova 2, 612 66, Brno, Czech Republic

Overview

2/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007 � Technologies of fingerprint sensors

� Quality measures derived from the image histogram

� Information entropy in the fingerprint

� Quality estimation of a papillary line

1 – Technologies of Fingerprint Sensors

3/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Optical Technology

Capacitive Technology

Ultrasound Technology

E-Field Technology

Electro-optical Technology

Pressure Technology

Thermal Technology

Fingerprint Sensor Technologies I.

4/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

� Optical Technology

Protective Glass

Lens

CCD-Camera

Lichtquelle Lichtquelle

Finger

� Capacitive Technology

Silicon Dioxide

↓

Finger

V

δδδδQC1 Cout

V0

~65µ

Metal Plate

Identix

Veridicom

Fingerprint Sensor Technologies II.

5/25

� Ultrasonic Technology

� Electro-optical Technology

xy

T

Isolation Layer

Black Coaxial Layer

Light Emitting Phosphorus Layer

Basic Layer

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Optel

AuthentTec

Fingerprint Sensor Technologies III.

6/25

� Pressure-Sensitive Technology

� Thermal Technology

Finger

Non-Conductive Layer

Electro-Conductive Layer

Insulation Layer

Pyro-electric Cell

Raw Image

Image Reconstruction

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

BMF

Atmel

2 – Quality Measures from the Image Histogram

7/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

8/25

� Histogram – Example (Suprema SFM 3020 – B&W)

� Histogram – Example (Suprema SFM 3050 – gaps)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Histogram of a Fingerprint Image I.

9/25

� Histogram – Access Control Systems

� Histogram – Dactyloscopic (Identification) Systems

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Histogram of a Fingerprint Image II.

Quality of Fingerprints I.

10/25

� Image Contrast

� Weber Contrast

� Michelson Contrast

� Average Value of Grayscale (∅ Full G-Scale = 128)

� Separability of Histogram Peaks (B&W Maximums)

� Mean Value of the Histogram (Usage of PDF)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Weber

LC

L

∆=

L

LCWeber

∆=

)(

)(

minmax

minmax

LL

LLCMichelson +

−=

Quality of Fingerprints II.

11/25

� Michelson Contrast – Results by SupremaN

IST

–B

iom

etric

Qua

lity

Wor

ksho

p II

–20

07

Weber

LC

L

∆=

0

0,2

0,4

0,6

0,8

1

1,2

1 63 125 187 249 311 373 435 497 559 621 683 745 807 869 931 993 1055 1117 1179

No. of Fingerprint

Mic

hels

on C

ontr

ast

SFM3000 SFM3010 SFM3020 SFM3050

� Mean Value of the Histogram (SFM3000 & SFM3050)

0

20

40

60

80

100

120

1 63 125 187 249 311 373 435 497 559 621 683 745 807 869 931 993 1055 1117 1179

No. of Fingerprint

Mea

n V

alue

0

10

20

30

40

50

60

1 63 125 187 249 311 373 435 497 559 621 683 745 807 869 931 993 1055 1117 1179

No. of Fingerprint

Mea

n V

alue

3 – Information Entropy in the Fingerprint

12/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

http://www.edcampbell.com/PalmD-History.htg/fbi-adob.jpg

Resolution of a Papillary Line

13/25

σσFF = 0,33 = 0,33 mmmm

σσSS = 0,043 = 0,043 mmmm

σσFF = 7,7= 7,7××σσSS

FσF

FσF

σS

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Maximal Amount of Minutiae

14/25

MaximumMaximumMM = 368 = 368 MinutiaeMinutiae

MM << 368<< 368!

15

15 m

mm

m=

46

= 4

6 ×σ×σ

FF10 10 mmmm = 31= 31×σ×σFF

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

More in: Drahansky, M.: Biometric Security Systems – Fingerprint Recognition Technology, Dissertation, Brno University of Technology, Faculty of Information Technology, 2005

Real Amounts of Minutiae + Entropies

15/25

Amount of minutiae (Bergdata)

0

20

40

60

80

100

120

1 2 3 4 5 6 7 8 9 10

User [No]

Num

ber

of m

inut

iae

Maximum Minimum Average

Amount of minutiae (Dactyloscopic)

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10

User [No]

Num

ber

of m

inut

iae


Amount of minutiae (Veridicom)

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10

User [No]

Num

ber

of m

inut

iae


NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Entropy Computation:

Maximal Entropy (368 minutiae):

Minimal Entropy (12 minutiae):

Quantization = Reduction of Entropy !!!

4 – Quality Estimation of a Papillary Line

16/25

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

http://162.105.71.163/people/chenxg/fip.jpg

Center Computation I.

17/25

� Finding of Fingerprint’s Center

� Gravity Center of the Minutiae (M1)

� Vertical & Horizontal Maximums of Ridge Count (M2)

� Middle of the Orientation Field (M3)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

CoreCore

DeltaDelta

Type-Lines

18/25


� Gravity Center of the Minutiae (M1)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Center Computation II.

19/25


� Vertical & Horizontal Maximums of Ridge Count (M2)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

37,0031,0051,0024,00Vertical maximum

25,6722,7933,1813,26Vertical average

11,0011,003,005,00Vertical minimum

31,0030,0027,0023,00Horizontal maximum

21,2519,9319,3012,86Horizontal average

10,009,006,006,00Horizontal minimum

050020010000Sensor(Suprema

SFM3xxx)

Center Computation III.

20/25


� Middle of the Orientation Field (M3)

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Center Computation IV.

Crosscut Through the Fingerprint Image I.

21/25

� Crosscut Through the Fingerprint in the Center

Valley(background)

Ridge(foreground)

� Reason: Perpendicularity of the Cut to Papillary Line

� Same for Horizontal and Vertical Cuts

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Crosscut Through the Fingerprint Image II.

22/25

� Comparison with the Sine Function

� Range:

� Deviation of the Curvature from the Sine Function

2

3,

2

ππ−

sin

1 100%FAA

AD

A

= − ⋅

( )E

S

x

FA

x

A f x dx= ∫ sin sin( )E

S

x

x

A x dx= ∫&…

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Crosscut Through the Fingerprint Image III.

23/25

� Thickness of the Papillary Line

� Deviation of the Thickness from the Defined One (0,33 mm)

[ ]2,54Pix

DPI

Th N cmR

= ⋅

1 100%0,033Th

ThD

= − ⋅

18%

18%

18%

178195 187

5,5 Pixels 8 Pixels 7 Pixels

[ ]2,54Pix

DPI

Th N cmR

= ⋅

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Crosscut Through the Fingerprint Image IV.

24/25

� Steepness of the Papillary Line

� Deviation of the Steepness from Ideal Triangle

Px1 Px2

Py

α β

1

1

arcsin x

x y

P

P Pα

= +

2

2

arcsin x

x y

P

P Pβ

= +

%10060

60⋅

°°−

=β

βD%10060

60⋅

°°−

=α

αD

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

The End

25/25

�� Thank you for your attention!

�� Thanks to MSM 0021630528 – Security-Oriented Researc h in IT

NIS

T –

Bio

met

ric Q

ualit

y W

orks

hop

II –

2007

Methods for Quality Determination of Papillary Lines in ...

Documents