1
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
2
Biometrics andPattern Recognition
Lab
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
Human Centered Computing Division
Clemson University, Spring 2010
3
About Us
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
4
Biometrics and Pattern Recognition Lab
Established in Summer 2006
Formerly the Image and Video Analysis Lab (IVAL)
In 2008, became part of the Center of Advanced Studies in Identity Sciences (CASIS) with CMU, UNCW, and NC A&T University.
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
5
Biometrics?
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
6
(Bio)(Metrics)
Bio◦Life
Metrics◦To measure
Biometrics:◦The science of identifying or authenticating an individual’s identity based on behavioural or physiological characteristics.
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
7
Biometric CharacteristicsPhysical Characteristics◦Iris◦Retina◦Vein Pattern◦Hand Geometry◦Face◦Fingerprint
Behavioural Characteristics◦Keystroke dynamics◦Signature dynamics◦Voice◦Gait
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
8
Why Biometrics?Eliminate memorization
◦Users don’t have to memorize features of their voice, face, eyes, or fingerprints
Eliminate misplaced tokens◦Users won’t forget to bring fingerprints to work
Can’t be delegated◦Users can’t lend fingers or faces to someone else
Often unique◦Save money and maintain database integrity by
eliminating duplicate enrollments
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
9
Biometric System
Verification (1:1)
Identification (1:N)
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
10
Purpose
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
11
Two main research goalsTo produce:
1. Usable Biometrics It might have 100% performance, but if it isn’t
feasible in the real world, who cares?
2. Unconstrained Biometrics At present, good recognition rates depend on a
lot of variables being just right, or at least consistent
We would like to reduce the dependency or get rid of it altogether
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
12
Constraints
Some common constraints are lighting, non-uniform distance, pose, expression, time lapse, occlusion
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
Typical image used in facial recognition
Unconstrained image
13
Projects
Periocular Region Recognition
Feature Reduction using Computational Intelligence
Aging Effects on Facial Recognition
Effects of Demographics on Facial Recognition
Soft Biometrics
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
14
Periocular Region Recognition
Relaxes image quality (location of iris, focus, blurring) on iris images
Could be used if more of the face is occluded
Currently looking at texture, color, and eye shape
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
15
Feature Reduction using Computational Intelligence
General Regression Neural Network
(GRNN)
Reduce the size of the features to enable faster, more portable biometric applications
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
16
Aging Effects on Facial Recognition
Looking at an image of a person, can we reliably predict◦what age they are?◦what they will look
like in so many years?◦or what they looked
like in the past?
Relaxes time lapse constraint
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
17
Demographics
How do demographics affect recognition?◦Older easier to recognize than younger◦Males easier than females
Why do some algorithms work better on certain populations than others?
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
18
Soft Biometrics
What if we don’t have enough information to identify the person?
We would like to know as much about them as possible: age, gender, ...
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
19
Questions?
Clemson University School of Computing Biometrics and Pattern Recognition Lab Director: Damon Woodard, PhD
Recommended