Iris Recognition

IRIS RECOGNITION AN IMAGE PROCESSING TECHNIQUE

B.venkatesh. [email protected]. & T.Hemasundara rao. [email protected].

ADITYA INSTITUTE OF TECHNOLOGY & MANAGEMENT

III/IV ECE

ABSTRACT:The

pressures

on

today’s

system

administrators to have secure systems are ever increasing. One area where security can be improved is in authentication and identification.

Biometrics provides a

promising solution meeting all demands. Biometric physiological

identification and

utilizes behavioral

characteristics to authenticate a person’s identity which include facial recognition, fingerprints, palm prints, hand geometry, retinal patterns and iris patterns and signature, voice pattern and key stroke dynamics. Many have suffered from high

cost and unsatisfactory error rates. The newer

emerging

technology

IRIS

RECOGNITION, which uses the unique patterns of the human iris, overcomes previous shortcomings and provides positive recognition of an individual without contact or invasion, at extremely high confidence levels. The technology is accurate, easy to use, non-intrusive, and difficult to forge and, despite what many people may think is actually quite a fast

system once initial enrollment has taken

characteristics — a process known as

place.

biometric identification. Various human The recognition principle here is

features have been used as the basis for

the failure of a test of statistical

such

independence on iris phase structure

These include fingerprints,

encoded

by

multi-scale

quadrature

wavelets. This paper, at the outset, throws

measurement

and comparison.

palm prints, hand geometry, vein patterns,

facial

characteristics,

capillary

patterns

in

the

and retina.

light on the technology involved in iris

Additionally, behavioral characteristics

recognition, the algorithms followed by

such as signature dynamics, voice

a statistical analysis highlighting its edge

patterns, and keystroke dynamics have

over

identification

been used for verification. Many of these

systems and its applications in the

require contact or are perceived as

present day world.

This paper is

invasive or intrusive. Others require a

intended for security practitioners who

person to make a final judgment or are

are knowledgeable, but not technically

costly, or suffer from unsatisfactory error

or scientifically oriented.

rates. Iridian Technologies, Inc. has

other

biometric

developed an automated Iris Recognition INTRODUCTION

Technology (IRT) capable of positively

Background

identifying persons without physical

Personal identification has historically been based on what a person possesses (a letter of introduction, perhaps), knows (a secret password), or is (personal recognition). In today’s security industry, possession has become an encoded card, knowledge equates to knowing one’s Personal Identification Number (PIN), and personal recognition has given way to the measurement and comparison of physiological

and

behavioral

contact or real-time human decisionmaking. This technology, for which Iridian Technologies holds many U.S. and international patents, meets this challenge

without

the

drawbacks

exhibited by other mentioned

above.

developing

a

recognition

technologies In

addition

commercially

system

for

to

viable

industrial

application, Iridian Technologies has developed systems for use by the U.S.

government,

correctional

facilities,

nuclear utilities, banks, and now, E-

BASISOFTHE

commerce and computer workstation

T E C H N O L O G Y:-

security. This paper is intended for

Features of the Iris

security

are

The iris is the plainly visible, colored

knowledgeable, but not technically or

ring that surrounds the pupil. It is a

scientifically oriented.

muscular structure that controls the

Is it like retinal scanning?

amount of light entering the eye, with

Iris

practitioners

recognition

who

technology

is

not

intricate details that can be measured,

associated in any way similar to retinal

such as striations, pits, and furrows. The

(capillary) pattern recognition. The iris is

iris is not to be confused with the retina,

the externally visible, colorful, donut-

which lines the inside of the back of the

shaped organ surrounding the pupil of

eye. No two irises are alike. There is no

the eye. The retina is the hemispherical

detailed correlation between the iris

organ behind the cornea, lens, iris, pupil,

patterns of even identical twins, or the

and vitreous humour and is not readily

right and left eye of an individual. The

visible. The Iridian Technologies process

amount of information that can be

captures a video image of the iris,

measured in a single iris ismuch greater

externally visible behind the eye’s

than fingerprints, and the accuracy is

cornea; whereas the retinal scan process

greater

than

scans the fovea, or innermost surface of the eye, the retina. Iris recognition systems truly recognize individuals, doing an exhaustive, one-to-many search of an entire database without benefit of PIN, password, or token. Most other biometric technologies rely on verification, where card or PIN inputs are required to pre-select a single enrolled

file

comparison.

for

a

one-on-one

Uniqueness of the Iris

DNA.

The iris is unique because of the chaotic

or eye, is one of the most carefully

morphogenesis of that organ. To quote

protected organs in one’s body. In this

Dr. John Daugman (Reference 1), “An

environment,

advantage

with

deleterious effects of aging, the features

fingerprints is the chaotic morphogenesis

of the iris remain stable and fixed from

of its minutiae. The iris texture has

about one year of age until death.

chaotic dimension because its details

Natural Protection from Artifice

depend

The

the

on

iris

initial

shares

conditions

in

human

and

eye

not

has

subject

to

physiological

embryonic genetic expression; yet, the

properties that can be exploited to

limitation of partial genetic penetrance

impede use of images and artificial

(beyond expression of form, function,

devices to spoof the system. As a matter

color and general textural quality),

of policy, Iridian Technologies, Inc. does

ensures that even identical twins have

not discuss the details of these properties

uncorrelated iris minutiae. Thus the

or specific countermeasures in the open

uniqueness of every iris, including the

media

pair

possessed

parallels

the

by

one

uniqueness

individual, of

every

PROCESSING

fingerprint regardless of whether there is

Image Acquisition

a common genome.” Given this, the

Optical

statistical probability that two irises

optimized for specific iris recognition

would be exactly the same is estimated

applications allow image acquisition at

at 1 in 1072.

distances from 3.5" to nearly one meter.

Stability of the Iris

In its simplest functional configuration,

Notwithstanding its delicate nature, the

an optical unit acquires multiple images

iris is protected behind eyelid, cornea,

of the presented iris through a simple

aqueous

frequently

lens, a monochrome CCD camera, and a

eyeglasses or contact lenses (which have

frame grabbing board. Multiple low

negligible effect on the recognition

level LEDs operating in the 720 to 850

process). An iris is not normally

nanometer range provide illumination

contaminated with foreign material, and

over the focal range of the camera. User

human instinct being what it is, the iris,

alignment in the X, Y, and Z axes is

humor,

and

platforms

designed

and

aided by a combination of mirrors, audible verbal direction, and in some cases, zoom or auto-focus lenses. In some models, a range finder, initiates the process automatically when a subject approaches within 18" of the unit. Iris Definition The image that best meets the focus and detail clarity requirements of the system is then analyzed to locate the limbus (the outer boundary of the iris that meets the white sclera of the eye), the nominal pupillary boundary, and the center of the pupil (see Figure 1). The precise location of the circular iris has now been defined and

processing

can

take

place.

Field Optimization The system quickly defines the suitable usable area of the iris for feature extraction

and

analysis.

Using

algorithms that exclude areas covered by eyelids, reflection,

deep etc.,

shadow, it

specular

concentrates

its

feature-extraction power in those areas

most visible (Figure 2). A dynamic

produce what is known as Iris Code,

feature of the system automatically

which is a 512-byte record.

adjusts the width of the pupillary boundary-to-limbus zone in real time to maximize the amount of iris analyzed, given varying ratios of pupil to iris sizes. Elimination of marginal areas has little negative impact on the analysis process. In actual practice, excellent enrollments

Image Analysis

and subsequent recognitions are obtained

The features of the iris are then analyzed

with 40% or less of the iris available for

and digitized into a 512 byte (4096 bit)

analysis.

IrisCode

Feature

locations

and

dimensions are defined using a polar

record, half of which describes the

coordinate (as opposed to a Cartesian, X,

features, half of which controls the

Y) system.

comparison process. During enrollment,

Image analysis Storing and comparing images

this IrisCode record is stored in the

2-D Gabor wavelets filter and map segments of the iris into hundreds of vectors(phasors).The 2-D Gabor phaser is simply the “what” and “where “ of the image. The phase demodulation process used to encode iris patterns. Local regions of an iris are projected onto quadrature

2d

gabor

wavelets,

generating complex-valued coefficients whose real imaginary parts specify the coordinates of a phasor is quantised to one of the four quadrants, setting two bits of phase information is used to

database for future comparison. During a recognition attempt, when an iris is presented at a recognition point, the same process is repeated; however the resulting IrisCode record is not stored, but is compared to every file in the database.

zero to that pair comparison. If two bits are different, it assigns a value of one to that pair comparison. After all pairs are compared, the number of disagreeing bit-pairs is divided by the total number of bit-pair comparisons resulting in a two digit quantitative expression of how different the two IrisCode records are. A Hamming Distance of .10 means that two IrisCode records differed by 10%. Hamming Distance Calculation Comparison of IrisCode records includes calculation of a Hamming Distance (HD), as a measure of variation between the IrisCode record from the presented iris and each IrisCode record in the database. Each useable pair of the 2048 available pairs of bits is compared

Recognition or Rejection

(Figure 3), and a value assigned using

Millions of IrisCode record comparisons

exclusive-OR logic. (The total 2048

have

pairs are seldom compared in their

Distributions (Probability Densities) in

entirety,

field

Figure 4. Here, the mean imposter

optimization process described above.)

Hamming Distance is near 0.5, which is

Bit #1 from the presented IrisCode

what one would expect in a truly random

record is compared to bit #1 from the

relationship. Conversely, the mean value

reference IrisCode record, bit #2 from

for authentics is 0.08 (or 8%). Another

the

key

because

presented

of

IrisCode

the

record

is

defined

aspect

the

of

the

Frequency

Frequency

compared to bit #2 from the reference

Distributions is the very small Standard

IrisCode record, and so on. If two bits

Deviation, resulting in very tightly

are alike, the system assigns a value of

clustered HD scores for both impostors

and authentics. These two “spikey”

seconds).

distributions appear, visually at least, to be

discrete

(totally

separate).

Mathematically, however, the adjacent slopes cross, and at that point, a nominal Accept/Reject decision threshold can be established.

At

Hamming

Distance

(.342), the probability of a False Reject is approximately the same as the probability of a False Accept. That setting reflects a conscious decision that an IrisCode differing by more than .342 (34.2%) from the codes with which it is compared, is likely to be part of the Imposter

Frequency

Distribution

because it is approaching a random relationship with the compared code. Sophisticated

algorithms

have

been

developed to adjust automatically that threshold based on number of bit-pairs available for comparison and the size of

PE R FO R MAN C E

database. Obviously, rejection of an

Crossover (Equal) Error Rate (CER)

individual occurs only after rejection of

At present, a crossover error rate is

all stored files (normally less than two

estimated based on the point at which the Impostor and Authentic distributions cross.

The

Iridian

Technologies’

(theoretical) crossover error rate is .0000008, or 1 in 1,200,000 (Note Table 1). This would equate to a Hamming Distance of just under .342. This is an extremely conservative statement about

an

error

rate,

because

in

many

population.

Although

published

comparisons over time, recognition HDs

Crossover Error Rates are “worst case”,

average 0.08, where the probability of a

the Hamming Distance for the average

False Accept is 1 in 1048.

authentic comparison in the authentic

Recognition Speed

Frequency Distribution of Figure 4,

The speed of the recognition process is

equates to imposter odds of 1 in 1048.

determined

by

many

interacting

variables, the speed of the processor and size of the database being the two most obvious. Databases containing millions of IrisCode records have been searched

Iris Recognition is suitable for many applications. Here are some of the benefits: •

in seconds. Faster processors have kept recognition speeds in the 2–3 second

•

range, even with large databases. •

Confidence The promise of the iris recognition technology described herein has been

•

Smallest outlier population: there are very few people who cant use this technology, for instance blind people. Unparalled stability: human iris patterns are stable from about one year in age till death. Unique design facilities &superior management of large databases: Unmatched search speed is

established based on millions of file

limited not by database size but

comparisons

by hardware selected for server

accuracy.

with

System

extraordinary

problems

management.

almost

exclusively reflect lighting, focus or

•

High level user acceptance.

image

•

No more need for ID Cards or

degradation

issues.

Having

Passwords

analyzed formally 12 separate databases

So there is no need to worry

totaling more than a million scores in accordance with recognized principles of

about

statistical decision theory, one can say

damaged or stolen or passwords

with 99% confidence that the values

being forgotten.

established for the Hamming Distances

•

ID

cards

being

lost,

Easily accommodates growing

of authentics and impostors in Figure 4

numbers

are within 9% of the whole (infinite)

All that is required to add users is

of

users

•

to register their iris patterns. This flexibility makes iris recognition

to scan from a distance.

ideal for large facilities. •

The iris is a very small organ

•

Time and cost savings are large

It is a moving target and can be obscured by objects such as the eyelid and eye lashes.

There is no need to issue or

•

•

The camera here needs to

manage ID cards and passwords.

have the correct amount of

Once

illumination.

users

are

registered,

The

system

additional time and costs are

linked by the camera is

reduced to near zero.

currently

Border Control & Immigration

images in a monochrome

Iris recognition is already being

format.

used

in

numerous

only

capturing

airports

worldwide for government and

CONCLUSION

transportation applications such

Highly

as

restricted

recognition is feasible today using the

access, and simplified passenger

iris of the human eye. This unique and

travel. Your iris is literally your

complex

passport at border crossings in

dimensions (measures) of variation than

countries like Canada,. Where

any other biometric feature currently in

iris recognition supports border

use, remains stable throughout a lifetime

control, enrolled travelers can

and is readily available for sampling in a

bypass

and

non-intrusive way. And has the speed

immigration lines, while security

required minimizing user frustration

officials can more efficiently

when accessing company systems. The

concentrate

process uses simple and non-threatening

border

control,

lengthy

on

unknown travelers.

customs

examining

accurate,

organ,

positive

which

personal

has

more

video technology to take images of the iris, digitize the features, and create a

Drawbacks

512-byte code, which is then compared

There are few drawbacks.

against an entire database in less than two seconds. Recognitions can then be

used to control access and entry, to provide recognition information to an existing entry control system or for any other

purpose

where

positive

identification is needed. Recent testing, under

U.S.

conditions,

Government

controlled

in

real-world

three

environments, and in a variety of operational applications have proven the practicality

and

feasibility

of

the

extremely accurate iris recognition for any

function

requiring

positive

recognition. References 1. http://www.sans.org/rr/whitepapers/a uthentication/132.php 2. http://www.computer.org/itpro/home page/jan_feb/security3.htm 3. http://www.rycom.ca/solutions/pdfs/i ridian/iris_whtppr.pdf 4. http://www.cl.cam.ac.uk/users /jgd1000/history.html 5. http://www.stanford.edu//