Department of Electrical and Computer Engineering

spotlessstareΑσφάλεια

29 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

212 εμφανίσεις

Department of Electrical and Computer Engineering

Physical Biometrics

Matthew Webb

ECE 8741

Department of Electrical and Computer Engineering

Introduction


Identity authentication is used in all aspects
of present day life


ATM, car, house, work, computer


Biometrics are a convenient and secure way
of producing valid ID


Can never forget it or lose it


Department of Electrical and Computer Engineering

Biometrics


What is biometrics?


The study of automated identification by use of
physical or behavioral traits


Growing field due to applicability to many
aspects of everyday life


Very popular since the events of 9
-
11

Department of Electrical and Computer Engineering

Physical vs Behavioral


Physical


Fingerprint


Iris


Ear


Face


Retina


Hands


Behavioral


Signature


Walking gait


Typing patterns



Both


Voice

Department of Electrical and Computer Engineering

Essential Properties of a Biometric


Universal


Everyone should have the characteristic


Uniqueness


No two persons have the same characteristic


Permanence


Characteristic should be unchangeable


Collectability


Characteristic must be measurable

Department of Electrical and Computer Engineering

Biometric System Process Flow

Department of Electrical and Computer Engineering

Pattern Recognition


Description and classification of
measurements taken from physical or mental
processes


Examination of pattern characteristics


Formulation of the recognition system


Important part of any biometric system

Department of Electrical and Computer Engineering

Key Measures of a Biometric System


False Acceptance Rate (FAR)


Rate of imposters being accepted as genuine user


False Rejection Rate (FRR)


Rate of genuine users being rejected


Equal Error Rate (EER)


Point at which the FAR and the FRR are equal

Department of Electrical and Computer Engineering

Fingerprint


Oldest known method of biometrics


Skin has ridges that are non
-
continuous and
form a pattern


Used in criminal databases


Used for security on laptops


Department of Electrical and Computer Engineering

Fingerprint Recognition


Two main types


Minutia matching


Galton features


dots, islands, ponds, spurs, bridges, crossovers, and
bifurcations


Pattern matching


Henry system


left loop, right loop, arch, whorl, and tented arch

Department of Electrical and Computer Engineering

Fingerprint

Department of Electrical and Computer Engineering

Hand Geometry and Palmprints


Palmprints are an extension of fingerprint


Hand shape new technique of identification


Uses features such as finger position, shape,
length, width, and curvature


Used in multimodal systems right now

Department of Electrical and Computer Engineering

Hand Shape Features

Department of Electrical and Computer Engineering

Face


Most widely used and distinctive way
humans recognize each other


Quite popular research subject


Used in casinos, large sporting events, and
galas


Takes many images over time as opposed to
a single capture

Department of Electrical and Computer Engineering

Types of Facial Recognition Systems


Feature based


Finds certain points such as tip of nose, eyes, and
ends of mouth


Creates a geometric map of the face


Global


Uses the entire face image


Eigenface technique and neural networks

Department of Electrical and Computer Engineering

Facial Recognition

Department of Electrical and Computer Engineering

Ears


One of the better biometrics


Does not change with age, facial expression


Does not wear out


Normally found in the same basic spot for all


Immature technique


Used in bimodal systems with facial profile

Department of Electrical and Computer Engineering

Ears

Department of Electrical and Computer Engineering

Iris


Perhaps the most reliable of all biometrics


Completely random


Also an immature technology


Characteristics include rings, furrows,
freckles, striations, crypts, and the iris corona

Department of Electrical and Computer Engineering

Iris Recognition Techniques


Phase
-
based method


Zero
-
crossings representation


Texture analysis


Local intensity variation

Department of Electrical and Computer Engineering

Sample Iris Images

Department of Electrical and Computer Engineering

Conclusion


Biometrics are going to be used for many
things in the near future


Introduced under the assumption that its for
convenience


In reality there for security


Accomplishes both very well