The Representation of Fingerprint Minutiae as Defects in a Pattern-Formation System

Size: px

Start display at page:

Download "The Representation of Fingerprint Minutiae as Defects in a Pattern-Formation System"

Lynette Miles
5 years ago
Views:

1 The Representation of Fingerprint Minutiae as Defects in a Pattern-Formation System Jonathan Alfson, David Hjelmstad, Lucas Malin, Dominick Ortiz, Wacey Teller

2 Main Idea Main strengths of pattern formation models lie in Level 1 detail! Some Level 2-type defects are observed, but they are not distributed on the fingerprint in a way that is statistically consistent with real fingerprints Thus the ability to insert defects into patterns would allow for the generation of statistically realistic fingerprints We present a mathematical characterization of defects, a basic proof of concept via tests on simple pattern systems, and relate back to the end goal of generating fingerprints.

3 Three Types of Level 2 Characteristics Ending Ridge Bifurcation Dot Citation: Eric Ray, An Overview of Fingerprints for MAT 451

4 Newell-Whitehead-Segel (NWS) Equation Nonlinear Partial Differential Equation Relates Amplitude to spatial components x and y as well as to time t Useful for describing defects i.e. points where the amplitudes are zero Can be analytically solved by making certain assumptions

5 Assumptions Solution to amplitude is a function of phase Thus, the NWS equation applies to phase The steady-state solution for the phase will result in stable patterns in time

6 Field near defects Assuming we are near the core of the defect, we get the following:

7 Field far from defects Assuming we are far from the core of the defect, we get the following:

8 Growing a Defect Place a structure such as the previous two functions into a field of rolls Run a reaction-diffusion model with the above as initial conditions Examine result to see what kind of defect has grown

9 erf function initial conditions Step size N=59

10 erf function result Step size N=59

11 Result There appears to be a defect, most likely a bifurcation with some noise Resolution is too low to allow for proper growth/calculation of defect

12 Determination of steady state roll pattern u 0 = sin 2 (3! y) v 0 = cos 2 (3! y) Observation: Steady state is reached in about.015 seconds

13 Steady state pattern This state is used as the basis for the initial conditions in the experiments that follow

14 Case 1 Enforce a short line of zeros in center for.005 seconds then release u: v:

15 Case 1 After the release: u: v:

16 Case 1 u: Resulting steady state

17 Case 1 v: Resulting steady state

18 Case 2 u: Enforce zero along line from center to one edge for.005 s " Ridge Ending

19 Case 2 v: Enforce zero along line from center to one edge for.005 s " Bifurcation

20 Case 3 u: Same as Case 2, but hold for less time (.0025 s) " Illustrates the importance of timing: By reducing the time held at zero, the discrepancy between the amplitudes before and after the defect occurs is reduced.

21 Case 3 v: Same as 2, but enforce for less time (.0025 s)

22 Case 4!but don t go too far " Holding at zero for too short of a time results in healing of the defect

23 Case 4!but don t go too far

24 Case 5 Use a rectangle as wide as the ridges instead of just a line u: v: " Produces results similar to those of Case 3 when held for a very short time. Holding longer produces an enlarged middle ridge in the steady state

25 Bifurcation on a larger domain

26 Bifurcation on a larger domain Top view with simple filter applied

27 Ridge ending on a larger domain

28 Ridge ending on a larger domain With simple filter applied

29 Dot? " Much more difficult to produce by these methods

30 Defects of a more Natural Look To give the finger prints a more natural look. A controlled randomness was added to the initial grid The randomness was taking the set grid and taking some set percentage of the values than adding it back on.

31 Comparison Non-randomized Bifurcation With randomization

32 Initial grid Without randomization With 50% randomization

33 Conclusion The growth of defects in the steady state of a pattern-formation system has been shown to be possible To implement in the generation of fingerprints, one would need to have the probability distribution of all minutiae. An algorithm could then be developed to distribute the defects randomly, but according to the probabilities via a system of weights.

34 Acknowledgements Bruno Welfert! Use of Schnakenberg pattern formation code Keith Hjelmstad! Assistance with numerical methods in second version of code

ZKTECO COLLEGE- FUNDAMENTAL OF FINGER VEIN RECOGNITION

ZKTECO COLLEGE- FUNDAMENTAL OF FINGER VEIN RECOGNITION What are Finger Veins? Veins are blood vessels which present throughout the body as tubes that carry blood back to the heart. As its name implies,