Fingerprint Recognition Through Circular Sampling

1
Fingerprint Recognition Through Circular Sampling

• David Chang and Joseph Hornak
• Rochester Institute of Technology
• Rochester, NY 14623-5604

2
(No Transcript)
3
Introduction

• Useful for personal identification.
• Valuable to criminal investigators and forensic
  scientists.

4
Overview

• Introduction
• Background
• Theory
• Methods
• Results
• Conclusion
• Future Work

5
Background

• Fingerprint Uniqueness
• Obstacles
• AFIS Method

6
Fingerprint Uniqueness

• Major Central pattern
• arch, loop, whorl

7
Fingerprint Uniqueness

• Minor Minutiae
• ridge termination, bifurcation

8
Obstacles

• Rotation
• Displacement
• Missing area
• Image defects

9
AFIS Method

• Image enhancement
• Feature extraction
• Feature mapping
• Classification via flow maps
• Matching
• of minutiae
• Euclidean distances

10
Theory Sampling Process
11
Theory Sampling Process
Fingerprint Image
Concentric Circle Sample
12
Circular Correlation
x
Rotated
x
13
Match Metric Area Ratio

• Average of ratio between the areas of
  corresponding circles in the two samples being
  matched.

14
Match Metric Correlation Fraction

• Average of the max value in the correlated signal
  divided by smaller area of corresponding circles
  in the two samples being matched.

15
Match Metric Angular Density

• Determine mean square error (MSE) among angles
  corresponding to highest magnitude in the
  correlation signal.

MAngular Density 1 - 2(MSE)/?
16
Methods

• Source Images
• 48 Synthetic Fingerprint Images
• 512 x 512 pixels at 1-bit/pixel
• Match Matrices
• 48 x 48 matrix where each column sample is
  matched against each row source sample.
• Done for the 3 metrics.
• Observe effects
• Missing Areas
• Rotation
• Examine displacement effects

17
Results Unchanged Variables
18
Arbitrary Area Removed
19
Results Arbitrary Area Removed
20
Results Rotation Effects
Column images rotated 45?
21
Results Displacement Effects
22
Conclusion

• Of the three metrics, the angular density metric
  proves to be most effective.
• Displacement effects show that a consistent
  selection of the circles center is necessary.

23
Future Work

• Image enhancement
• Test on actual fingerprints
• Observe effects of less circles
• Test on larger database
• Code optimization

24
Thanks!
25
(No Transcript)
26
Fingerprint Chicks

• Materials
• Yellow
• Tempera paint
• Wash tubs
• Large construction paper
• Glue
• Directions
• 1.Each student chooses a construction paper for
  background.
• 2.Have students come up one at a time, to gently
  dip their hands and fingers in the yellow tempera
  paint.
• 3.Each will place their hands and fingers on
  their paper, making a fingerprint.
• 4.Kids decorate their fingerprints to look like a
  new Spring chick.