Iris Recognition Part I: Patrick Gampp Part II: Andrea Sereinig - - PowerPoint PPT Presentation

▶
iris recognition
SMART_READER_LITE
LIVE PREVIEW

Iris Recognition Part I: Patrick Gampp Part II: Andrea Sereinig - - PowerPoint PPT Presentation

Nov 16, 2022 418 likes •658 views

Advanced Signal Processing Seminar Iris Recognition Part I: Patrick Gampp Part II: Andrea Sereinig Advanced Signal Processing Seminar Graz, 12/05/2007 Patrick Gampp Iris Recognition Professor Horst Cerjak, 19.12.2005 1/22 Advanced Signal

slide-1
SLIDE 1

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

1/22

Patrick Gampp Iris Recognition

Iris Recognition

Part I: Patrick Gampp Part II: Andrea Sereinig Advanced Signal Processing Seminar Graz, 12/05/2007

slide-2
SLIDE 2

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

2/22

Patrick Gampp Iris Recognition

Content

  • Part I:

– Basics – The Daugman Iris Recognition System

  • Part II:

– Wildes Iris Recognition – Iris on the Move – Iris Recognition Systems

slide-3
SLIDE 3

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

3/22

Patrick Gampp Iris Recognition

Part I: Content

  • Motivation
  • Anatomy of the Human Iris
  • Image Aquisition
  • Iris Localization
  • Wavelet Transformation
  • Iris Feature Encoding by 2D Wavelets Demodulation
  • Pattern Matching
  • Recognizing Irises Regardless of Size, Position and

Orientation

  • Decision Environment
  • Countermeasures Against Subterfuge
  • Performance, Execution Speed
slide-4
SLIDE 4

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

4/22

Patrick Gampp Iris Recognition

Motivation

  • Noninvasive
  • Covert evaluation possible
  • Iris patterns variability is enormous
  • Iris is well protected from environment
  • Iris pattern is stable over time
  • Insensitive to angle of illumination
  • Insensitive to changes in viewing
slide-5
SLIDE 5

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

5/22

Patrick Gampp Iris Recognition

Anatomy of the Human Iris (1)

[Wildes 1997] [Wildes 1997]

slide-6
SLIDE 6

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

6/22

Patrick Gampp Iris Recognition

Anatomy of the Human Iris (2)

  • Iris patterns largely complete by eighth month
  • Pigmentation accretion into adolescence
  • Average pupil size increases until adolescence
  • Advanced age: slight depigmentation and shrinking of

pupillary opening

  • Patterns are stable with age
  • General structure genetically determined, but

minutiae dependend from initial conditions

  • NIR wavelengths reveal deeper stromal features
slide-7
SLIDE 7

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

7/22

Patrick Gampp Iris Recognition

Image Aquisition (1)

  • Need sufficient resolution, sharpness, contrast
  • Iris is relatively small
  • Iris is dark object
  • Human operators are sensitive about their eyes
  • Well-centered image while remaining noninvasive
  • Eliminate artifacts like reflectations, aberrations
slide-8
SLIDE 8

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

8/22

Patrick Gampp Iris Recognition

Image Aquisition (2)

  • LED point light source
  • NIR illumination 700nm- 900nm
  • Sensitive CCD camera
  • 100- 200 pixels monochromatic image
  • Distance 15- 46cm
  • Video rate capture
  • Self- positioning of object
  • Focus assessment: maximizing high-frequency power
  • f 2D Fourier spectrum
slide-9
SLIDE 9

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

9/22

Patrick Gampp Iris Recognition

Iris Localization

  • 1. Cirular path of contour integration for pupil edge
  • 2. Circular path for limbic boundaries
  • 3. Arcuate path with fitted splines for eyelid boundaries

∫

∗

) y , x , r ( σ ) y , x , r (

ds r π 2 ) y , x ( I r ∂ ∂ ) r ( G max

⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − −

=

2 2 σ 2 ) r r ( σ

e π 2 σ 1 ) r ( G

[Daugman Website] function smoothing Gaussian )....., r ( G s Coordinate Center )....., y , x ( Radius ....., r Image )....., y , x ( I

σ
slide-10
SLIDE 10

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

10/22

Patrick Gampp Iris Recognition

Why (Gabor) Wavelets?

  • Advantage over Fourier Transform representing

functions with discontinuities and sharp peaks

  • Compact representation of images, eg. jpeg2000
  • Simple cells in visual cortex can be modelled
  • „Quantum principle“ in information

good time- vs. frequency- resolution trade-off

[Ulrich Günther 2001]

slide-11
SLIDE 11

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

11/22

Patrick Gampp Iris Recognition

Complex 2D Gabor Wavelets

  • Mother function:
  • Generating function:

[ ]

⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − + − − − + − −

=

2 2 2 2

β ) y y ( α ) x x ( π ) y y ( v ) x x ( u i π 2

e e ) y , x ( Ψ ) ' y , ' x ( Ψ 2 ) y , x ( Ψ

m 2 θ mpq −

=

[ ] [ ] q

) θ sin( y ) θ sin( x 2 ' y p ) θ sin( y ) θ cos( x 2 ' x

m m

− + − = − + =

− −

parameters shift position; in n Translatio q....., p, parameter Dilation m....., envelope

  • f

rotation Discrete ....., θ carrier sinusoidal

  • f

variables frequency Spatial ....., v , u envelope

  • f

Length and Width , ..... β , α variables space Visual y....., x, envelope gaussian

  • f

peak

  • f

Location ....., y , x

[Daugman 1988]

slide-12
SLIDE 12

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

12/22

Patrick Gampp Iris Recognition

Iris Pattern Encoding (1)

⎪ ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ ⎨ ⎧ ⎪ ⎭ ⎪ ⎬ ⎫ ⎪ ⎩ ⎪ ⎨ ⎧ ≥ ⎪ ⎭ ⎪ ⎬ ⎫ ⎪ ⎩ ⎪ ⎨ ⎧ =

∫∫ ∫∫

− − − − − − − − − − − −

) , ( Re , ) , ( Re , 1

2 2 2 2 2 2 2 2

) ( ) ( ) ( ) ( ) ( ) ( Re

p

ρ φ β φ θ α ρ φ θ ω ρ φ β φ θ α ρ φ θ ω

φ ρ ρ φ ρ φ ρ ρ φ ρ d d e e e I if d d e e e I if h

r i r i

⎪ ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ ⎨ ⎧ ⎪ ⎭ ⎪ ⎬ ⎫ ⎪ ⎩ ⎪ ⎨ ⎧ ≥ ⎪ ⎭ ⎪ ⎬ ⎫ ⎪ ⎩ ⎪ ⎨ ⎧ =

∫∫ ∫∫

− − − − − − − − − − − −

) , ( Im , ) , ( Im , 1

2 2 2 2 2 2 2 2

) ( ) ( ) ( ) ( ) ( ) ( Im

p

ρ φ β φ θ α ρ φ θ ω ρ φ β φ θ α ρ φ θ ω

φ ρ ρ φ ρ φ ρ ρ φ ρ d d e e e I if d d e e e I if h

r i r i

[Daugman Website]

region Iris ....., θ , r β to proportion inverse

  • ctaves

3 spanning frequency; Wavelet ....., ω 1.2mm to 0.15mm from range fold

  • 8

; parameters size wavelet scale

  • Multi

....., β , α system coordinate polar in image iris Raw )....., φ , ρ I(

slide-13
SLIDE 13

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

13/22

Patrick Gampp Iris Recognition

Iris Pattern Encoding (2)

  • Cyclic phase- quadrant code: Gray code
  • Coarse phase quadrant quantization:

Ignore imaging contrast, illumination, camera gain

  • Many wavelet sizes, frequencies, orientations

256 Byte per iris

  • Masking bits: ignore obscured data
slide-14
SLIDE 14

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

14/22

Patrick Gampp Iris Recognition

Pattern Matching

  • Test of statistical independence
  • Fractional HD (Hamming Distance)
  • Obscured data: set mask bit ‚0‘
  • XOR, AND in parallel wordlength- size chunks

single machine instruction

  • 300MHz CPU: 100 000 iris comparisons per second

B mask A mask B mask A mask ) B code A code ( HD I I I ⊗ =

slide-15
SLIDE 15

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

15/22

Patrick Gampp Iris Recognition

Distribution of Hamming Distances

  • 9 million comparisons of different irises
  • 4258 iris images acquired in UK, USA, Japan, Korea
  • Bernoulli trial, but correlation between „coin tosses“
  • Only small subsets of bits are mutually independent
  • Same distribution for genetically identical irises

[Daugman 2004]

) m N ( m

) p 1 ( p m N ) m ( f

−

− ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ =

freedom

  • f

Degrees 249 N trials) Bernoulli correlated (for N ) p 1 ( p σ2 = ⇒ − =

slide-16
SLIDE 16

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

16/22

Patrick Gampp Iris Recognition

Homogeneous Rubber Sheet Model

  • Problem: Pupil size change
  • Solution:

( ) [ ] [ ]

π 2 , θ 1 , r ) θ , r ( I ) θ , r ( y ), θ , r ( x I ∈ ∈ →

) θ ( ry ) θ ( y ) r 1 ( ) θ , r ( y ) θ ( rx ) θ ( x ) r 1 ( ) θ , r ( x

s p s p

+ − = + − =

( )

( )

points boundary Pupillary ....., ) θ ( y ), θ ( x points boundary Limbus ....., ) θ ( y ), θ ( x

p p s s
slide-17
SLIDE 17

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

17/22

Patrick Gampp Iris Recognition

Cyclic Scrolling

  • Problem: Iris orientation depends upon head tilt etc
  • Solution: Compare iris code at many orientations
  • Preserve only best match

[ ] [ ]

1 n n n n n

(x) F 1 ) x ( nf ) x ( F dx d ) x ( f (x) F 1 1 ) x ( F

−

− = = − − =

[ ]

ns

  • rientatio

relative n at t tests independen n , ..... ) x ( F 1 match false getting not

  • f

y Probabilit , ..... ) x ( F 1 match false a getting

  • f

y Probabilit , .... . dx ) x ( f ) x ( F Criterion Acceptance HD , x..... CDF , )..... x ( F n

  • rientatio

1 in comparison for PDF , .......... f

n x

− − = ∫ [Daugman 2004]

slide-18
SLIDE 18

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

18/22

Patrick Gampp Iris Recognition

Uniqueness of Failing the Test of Statistical Independence

  • Cumulative binomial distribution function
  • False match probabilities:

even poor match (HD=0.3) provides compelling evidence of identity

∑

=

=

x x n n

) x ( f ) x ( F

[Daugman 2004]

) n π 2 ln( 2 1 n ) n ln( n

e ! n factorials high for ion Approximat s Stirling'

+ −

≈

slide-19
SLIDE 19

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

19/22

Patrick Gampp Iris Recognition

Decision Environment

  • 7070 different pairs of same-eye images
  • Overlap between distributions: error rate
  • Non-ideal: different distances, cameras, lighting
  • Decidability d´: Separation of distributions

2 σ σ µ µ ' d

2 2 2 1 2 1

+ − = [Daugman 2004] [Daugman 2004]

0.327

slide-20
SLIDE 20

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

20/22

Patrick Gampp Iris Recognition

Countermeasures Against Subterfuge

  • Fake iris, photograph, videotape, glass eye….
  • Observe pupil diameter caused by random light levels:

250ms constriction, 400ms dilation

  • Steady- state oscillation: 0,5Hz (Hippus)
  • Track eyelid movements
  • Ocular reflection by random light at various positions

due to moist cornea

  • Spectral signature of living tissue in NIR light
  • Vein patterns silhouettes in NIR light
slide-21
SLIDE 21

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

21/22

Patrick Gampp Iris Recognition

Speed Performance

  • 300 Mhz Sun Workstation
  • 100 000 iris comparisons per second
  • Large database: division into units of 100 000

[Daugman 2004]

slide-22
SLIDE 22

Advanced Signal Processing Seminar

Professor Horst Cerjak, 19.12.2005

22/22

Patrick Gampp Iris Recognition

References

  • [Daugman 2004]: How Iris Recognition Works
  • [Wildes 1997]: Iris Recognition: An Emerging Biometric

Technology

  • [Daugman]: Recognizing Persons by their Iris Patterns
  • [Daugman 2003]: Demodulation by Complex- Valued Wavelets for

Stochastic Pattern Recognition

  • [Daugman 1988]: Complete Discrete 2-D Gabor Transforms by

Neural Networks for Image Analysis and Compression

  • [Tai Sing Lee 1996]: Image Representation Using 2D Gabor

Wavelets

  • [Daugman Website]: http://www.cl.cam.ac.uk/~jgd1000/
  • [Ulrich Günther 2001]: Advanced Computing in NMR Spectroscopy