Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
3, N
o
. 4
,
A
ugu
st
2013
, pp
. 42
9
~
43
5
I
S
SN
: 208
8-8
7
0
8
4
29
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Iris Image Quality Testin
g and Iri
s
Verifi
cation
Lidong W
a
n
g
Department o
f
A
pplied
Technolo
g
y
,
Mississippi
Valley
S
t
ate University
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
Ma
r 6, 2013
Rev
i
sed
Jun1
6,
20
13
Accepted
Jun 25, 2013
The purpose of
this stud
y
was
to inves
tig
at
e th
e
iris
im
age qu
al
it
y
and ir
is
verification of
ey
es
in brown, h
a
zel, gree
n, and b
l
ue, respectiv
ely
,
and th
e ir
is
image quality
and iris ve
rification under differ
e
nt conditions such as the
changed stand-o
ff distances, th
e
motions
of the h
ead and
e
y
es
,
wi
th glas
s
e
s
,
and without glas
ses. A com
p
arative stud
y
of thre
e
e
y
e
colors in brown, hazel
,
and green was conducted using
a non-
parametr
ic
method based on
the
H
te
st.
The
H
test r
e
sult
s show that there
is no si
gnific
a
nt
differen
ce
in the
iris im
ag
e
quality
of
ey
es in brown, hazel, or gr
een when the lev
e
l of s
i
g
n
ific
ance
is
0.05.
Keyword:
Bio
m
e
t
rics
Iris im
ag
e qu
ality
Iris veri
fication
No
n-
pa
ram
e
t
r
ic m
e
t
hod
H
test
Copyright ©
201
3 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Lid
ong
W
a
ng
,
Depa
rt
m
e
nt
of Ap
pl
i
e
d Tech
n
o
l
o
gy
,
Mississip
p
i
Valley State Un
iversity,
1
400
0 Hw
y,
82
W
e
st,
I
tta Ben
a
, Mississi
p
p
i
38
941
,
U
S
A
.
Em
a
il: lwan
g
2
2
@
st
u
d
e
n
t
s.t
n
tech
.ed
u
1.
INTRODUCTION
Iris recog
n
ition
is a pro
c
ess th
at
analyses the feature
s
(s
uc
h as ri
ngs
, furrows
, and
frec
k
les) that exist
in
th
e co
loured tissu
e su
rround
ing
th
e
pup
il.
Th
ere are little
ag
ing
effects
mad
e
to
iris
p
a
ttern
s after th
e
ag
e
o
f
two. Most eye surge
r
ies rarel
y
affect
th
e iri
s
. Fin
e
iris tex
t
u
r
e can
k
eep
re
m
a
rkably stable over life from age
two
u
n
til d
eat
h
.
Th
erefo
r
e, iris re-enr
o
l
m
e
n
t
is no
t req
u
i
red
and
prev
i
o
u
s
ly reg
i
stered iris d
a
ta can
b
e
used
co
n
tinuo
usly. Iris recog
n
ition can b
e
used in immig
r
ati
o
n
syste
m
s, b
o
rd
er security system
s, n
a
tio
n
a
l iden
tity
cards, ide
n
tity
managem
e
nt a
nd e
-
Governa
n
ce, and aviatio
n security and
access control
for restricte
d
areas at
airports, etc. [1], [2].
Iris rec
o
gnition continues to be
acknowledged as the m
o
st accura
te biometric recognition m
e
thod
avai
l
a
bl
e i
n
t
h
e w
o
rl
d t
oday
(m
ore acc
urat
e t
h
a
n
DN
A
m
a
t
c
hi
ng)
.
Ho
weve
r,
t
h
e
pe
r
f
o
r
m
a
nce
of
t
h
e
i
r
i
s
syste
m
s can
b
e
affected
b
y
iris i
m
ag
es wi
th
p
o
o
r
q
u
a
lity. Iris i
m
ag
e
q
u
a
lity assessmen
t can
b
e
mad
e
b
y
an
alyzin
g
th
e effects o
f
seven
q
u
a
lity
facto
r
s: d
e
fo
cu
s b
l
ur,
m
o
tio
n
b
l
ur, o
f
f-ang
le,
o
ccl
u
s
ion
,
sp
ecu
lar
reflection
,
lightin
g
,
and
p
i
x
e
l
cou
n
t
s
on
t
h
e
p
e
rform
a
n
ce of trad
itio
n
a
l iri
s
reco
gn
itio
n syste
m
. Defo
cus b
l
ur,
m
o
t
i
o
n
b
l
ur, and
o
f
f-ang
le are
th
e fact
o
r
s th
at
m
o
st affect reco
gn
itio
n perform
a
n
ce [3
].
Som
e
of t
h
e m
a
in
param
e
ters that s
p
ecify a
n
im
ag
e system
are its resol
u
tion,
de
pth
of field (DOF),
fi
el
d of
vi
ew,
and e
x
p
o
s
u
re
peri
od
per i
m
age-
fram
e
. The fi
el
d of
vi
ew
det
e
rm
i
n
es t
h
e spat
i
a
l
ext
e
nt
of t
h
e
scene acq
ui
re
d
by
t
h
e sens
or
. Dept
h o
f
fi
el
d
det
e
rm
i
n
es
how far a
plana
r
object can
m
o
ve away from
th
e best
fo
cu
s
po
sitio
n
an
d still b
e
imag
ed withou
t fo
cus errors
. Cu
rren
t iris
recog
n
ition
systems suffer fro
m
l
i
m
i
t
ed
d
e
p
t
h
of
field
,
wh
ich
m
a
k
e
s it so
m
e
wh
at d
i
fficu
lt for an
un
train
e
d
u
s
er t
o
use th
ese
syste
m
s. Trad
ition
a
lly,
the
dept
h
of fi
eld is i
n
crease
d
by reducing the im
ag
e syste
m
aperture, which ad
ve
rs
ely im
pacts the light
cap
turing
p
o
wer
an
d
thu
s
th
e syste
m
sig
n
a
l-to
n
o
i
se
ratio
(SNR) [4
].
Iris reco
gn
ition
syste
m
s stil
l n
eed
to
im
p
r
ov
e th
eir accu
r
acy in
env
i
ro
n
m
en
ts ch
aracterized
b
y
un
fa
vo
rabl
e l
i
g
ht
i
n
g
,
l
a
r
g
e st
and
-
of
f di
st
ance
s, an
d m
ovi
ng
sub
j
ect
s [
5
]
.
T
h
e st
an
d-
of
f
di
st
ance i
s
t
h
e
d
i
s
t
a
n
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 3
,
N
o
. 4
,
Aug
u
s
t 2
013
:
42
9
–
43
5
43
0
fro
m
th
e ca
m
e
ra to
th
e su
bj
ect o
r
th
e user
of th
e iris
rec
o
gnition system
.
Researche
r
s m
u
st solve issue
s
such
as cap
turing
eye i
m
ag
es o
f
su
fficien
t
qu
ality in
less
th
an
i
d
eal con
d
ition
s
an
d
accurately lo
calizin
g
the iris’s
spatial extent i
n
poor-quality i
m
ag
es. Recent efforts ha
ve
successfully
designe
d a
nd
developed iris-on-t
h
e-
m
o
v
e
and
iris-at-a-d
i
stan
ce
reco
gn
itio
n syst
e
m
s [5
].
An
iris recogn
itio
n
syste
m
at a d
i
stan
ce
o
f
abou
t th
ree
meters was
d
e
v
e
lop
e
d [6
].
A n
e
w im
ag
e acq
u
i
sition
syst
e
m
called
BIri
s On
t
h
e Move was d
e
v
e
loped
to
redu
ce co
n
s
t
r
ain
t
s in
po
sition
an
d
m
o
tio
n
.
Th
is n
e
w syst
em u
s
es h
i
gh
-resolu
tio
n
cam
eras, v
i
d
e
o
syn
c
h
r
on
ized
strob
e
illu
min
a
tio
n
,
an
d
sp
ecu
l
arity-b
ased i
m
ag
e seg
m
e
n
tatio
n
.
It h
a
s resu
lted
i
n
an
in
creased
cap
t
ure
v
o
l
u
m
e, d
ecreased
acqu
isitio
n
tim
e, in
cr
eased
stan
d-off d
i
stan
ce, and
th
e ab
ility to
acqu
i
re iris im
ag
e
s
fro
m
m
ovi
ng s
u
bject
s [
7
]
.
Th
e
pu
rpo
s
e
o
f
th
is
p
a
p
e
r is t
o
st
u
d
y
:
1
)
iris
i
m
ag
e
qu
ality
an
d
iris v
e
rificatio
n
for fo
ur kin
d
s
of
eyes
(brown, hazel,
gree
n, a
nd
blue); 2) iris im
ag
e quality a
nd iris verification under
di
ffe
re
nt conditions, s
u
ch a
s
th
e ch
ang
e
d
stan
d-
of
f d
i
stances, th
e m
o
tio
n
s
o
f
th
e h
e
ad
an
d eyes, with
or
w
ith
ou
t
g
l
asses;
3)
a no
n-
p
a
ram
e
tric an
alysis b
a
sed on th
e
H
test for
three kinds of
eyes
(bro
wn,
hazel, and
green) to study their
diffe
re
nce.
2.
IRIS
IM
AGE
QU
ALITY A
N
D
IR
IS VER
I
FIC
A
TIO
N
I
N
DIFFE
REN
T
SITU
ATIO
NS
2.1. The Expe
rimental
Me
thod
and
the E
x
perimental
Syste
m
IrisAccess
TM
40
00
, an iris reco
gn
itio
n system
d
e
v
e
lo
p
e
d
by LG Electron
i
cs, was
u
s
ed in
th
is
stud
y
.
The L
G
Iris
A
c
cess
TM
iData EAC So
ft
ware
v 3
.
00
.1
4 was in
stalle
d in the
iris syste
m
. The cam
era iCAM4000
was use
d
to acqui
re the subjects’ iris im
ag
es. The iC
AM
4000 is a two-eye iris ca
mera which incl
udes an
al
i
gnm
ent
i
ndi
cat
or
be
hi
n
d
t
h
e m
i
rro
r an
d
v
o
i
ce p
r
om
pt
s to assi
st
t
h
e
us
er.
It
can
be
us
ed i
n
en
rol
l
m
ent
a
n
d
verification. T
h
e Iris
A
ccess
TM
400
0 sy
st
em
has fi
ve f
u
nct
i
on m
odul
es:
I
r
i
s
Ser
v
er
, Iri
s
E
nr
ol
l
,
Iri
sM
a
n
age
r
,
IrisMon
itor, an
d IrisDB
Admin
.
On
ly admin
i
strato
rs m
a
y lo
g
i
n in Ir
isServ
er
;
I
r
i
sSer
v
e
r
m
u
st b
e
r
unn
i
ng
b
e
fo
re starti
n
g
IrisEnro
ll. Iri
sEnro
ll is u
s
ed
to
en
ro
ll the irises o
f
u
s
ers in
to
th
e syste
m
, an
d
for th
e
i
d
ent
i
f
i
cat
i
on
or
veri
fi
cat
i
o
n
of t
h
e u
s
ers;
Iri
sE
nr
ol
l
m
u
st fi
rst
be
re
gi
st
ered i
n
I
r
i
s
M
a
nage
r.
Iri
sM
a
n
ager i
s
use
d
t
o
m
a
nage t
h
e Users
,
O
p
erat
ors (a
dm
ini
s
t
r
at
o
r
‐
lev
e
l o
n
l
y), Re
m
o
te Un
its, Prog
ram
s
, an
d
Group
s, as
well as Report generation in the syste
m
. IrisM
onitor is used to m
onitor the IrisAccess
TM
400
0
sy
st
em
.
IrisDB
A
d
m
in
i
s
a d
a
tab
a
se admin
i
stratio
n
too
l; th
is t
o
o
l
facilitates an
easie
r m
a
n
i
p
u
l
atio
n
for b
a
cku
p
, imp
o
rt,
create, drop,
upgra
d
e, and manage i
n
th
e IrisServ
e
r d
a
tabase for th
e d
a
t
a
b
a
se ad
m
i
n
i
strato
r. Th
e iris
syste
m
can
b
e
u
s
ed
in
en
ro
llm
en
t an
d
v
e
rificatio
n. En
ro
llm
en
t
is
th
e proce
ss of a
d
ding ne
w records. T
h
e records are
u
s
ed
to
v
a
lid
ate th
e u
s
ers’ id
en
tity d
u
r
ing
the v
e
rificati
o
n
pro
cess. Th
e user can
p
e
rfo
rm
a v
e
rificatio
n
test b
y
click
i
n
g
on
the Veri
ficatio
n
Test bu
tto
n. Th
e system
can also pe
rform
fake eye
detec
tion. T
h
e
fake
eye
d
e
tectio
n
in
creases th
e ti
m
e
req
u
i
red
for en
ro
ll
m
e
n
t
, id
en
t
i
f
i
cat
i
on,
or ve
r
i
fi
cat
i
on, b
u
t
g
r
eat
l
y
enhance
s
t
h
e
security of the
syste
m
. The iri
s
syst
e
m
can prom
pt the user to
prese
n
t
his/her i
r
is to t
h
e
ca
m
e
ra. The
s
y
ste
m
will p
r
o
m
p
t
for an
o
t
h
e
r try if th
e resu
lts o
f
t
h
e i
m
ag
e
p
r
o
c
essin
g
are no
t o
f
g
ood
qu
ality. Ano
t
h
e
r pro
m
p
t
will
also a
ppea
r
i
f
t
h
e im
age was
not
capt
u
re
d
properly
on
the
second try.
There is a
m
a
x
i
mu
m
o
f
three
atte
m
p
ts
for im
age proc
essing. If the
user does
not succeed in t
h
e third attem
p
t, he
or
she will
be aske
d
to begi
n agai
n
b
y
selectin
g the Enro
ll ico
n
fro
m
th
e start men
u
[8
],
[9
].
Iris im
ag
es are d
i
sp
layed
on
t
h
e Main
wind
ow of th
e
sev
e
r
PC; th
e qu
ality o
f
t
h
e IrisCode created
is
displayed i
n
the Processing R
e
sult window a
s
soon a
s
ir
is scan
n
i
n
g
is co
mp
leted
.
Th
e
p
r
ocessin
g
resu
lt is th
e
iris qu
ality score. Th
e
qu
ality
score ra
ng
es
fro
m
0
to
10
0.
Th
e l
o
west
v
a
l
u
e is
fix
e
d
at
0 and
th
e h
i
g
h
e
st v
a
lu
e
is fix
e
d
at 10
0. To
ob
tain
iris i
m
ag
es with
a h
i
gh
er
qu
ality an
d d
e
crease
th
e “False Reject Rate” (FRR: th
e
rejection
rate
of a
n
iris t
h
at shoul
d
be acce
pt
ed),
the
use
r
s
h
oul
d
follow t
h
e
rec
o
mmendations
below [9]:
1)
The
use
r
s
h
o
u
l
d
keep
b
o
t
h
ey
es wi
d
e
ope
n a
n
d
l
o
ok
in
to the rectangu
lar
mirro
r alig
n
i
ng
th
e co
l
o
red
d
o
t
b
e
tween
th
e ey
es un
til th
e au
dio
m
e
ssag
e
of
“W
e
fin
i
sh tak
i
n
g
p
i
ctures
of
yo
ur eyes”
p
l
ays.
2)
Th
e
u
s
er sh
ou
l
d
n
o
t
ro
tate, p
a
n
,
or tilt h
i
s/h
e
r face.
3)
Eye g
l
asses m
u
st
b
e
rem
o
v
e
d b
e
fore en
ro
llmen
t, bu
t m
a
y
b
e
worn
d
u
ring v
e
rificatio
n or id
en
tification
.
4)
Contact lenses
with
patterns
t
h
at cove
r
an
y par
t
of
th
e ir
is can
n
o
t
b
e
w
o
rn
.
2.
2. I
r
i
s
Im
ag
es, Im
ag
e Q
u
a
l
i
t
y Sc
ores
an
d
Iris Verification
for Differ
e
nt
E
y
e Col
o
r
s
Four test s
u
bje
c
ts’ iris im
ages we
re ca
pture
d
using the
Iris
Access
TM
400
0 syste
m
in
th
e
Au
t
o
m
a
ted
Ide
n
t
i
f
i
cat
i
on
Tech
nol
ogy
l
a
b at
M
i
ssi
ssi
pp
i
Val
l
e
y
St
at
e
Uni
v
ersi
t
y
, US
A. T
h
e fo
u
r
t
e
st
sub
j
ect
s had
br
ow
n
,
hazel, green, and
blue eyes, respectively.
None
of them
wore glasse
s. The
stand-off dist
ance from
the
ca
m
e
ra
to
th
e in
d
i
v
i
d
u
al was 2
0
cm
.
Fig
u
re 1
sh
ows th
e fou
r
ind
i
v
i
du
als’ iris imag
es an
d
th
e i
m
ag
e q
u
a
lity
sco
r
es
after th
e en
ro
llmen
t p
r
o
c
ess.
Th
e iris im
ag
e
acq
u
i
sition
yield
e
d
im
ag
es o
f
th
e irises and th
e su
rro
und
in
g
ey
e
regi
ons
. All of the four im
ages were
used
for the each
i
ndi
vidual veri
fication an
d the verifications
were
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Iris Ima
g
e
Quality Testin
g
a
n
d
Iris Verifica
t
io
n
(Li
d
ong
Wa
ng
)
43
1
successful, although Figure
1 (a) a
n
d Fi
gure
1
(d) a
r
e
i
m
ages with
occlusion a
n
d relatively low
quality
score
s
.
(a)
(b)
(c)
(d)
(a)
African American; Male;
Ey
e co
lor: Brown;
Imag
e with
occlusion
;
Quality
sco
r
e—86.4
(b)
Caucasian; Fem
a
le; Ey
e
colo
r: H
azel; Qu
ality
sco
r
e—98.6
(c)
Caucasian; M
a
le; Ey
e
color: Green; Quality
s
c
or
e—93.4
(d)
Ca
uc
a
s
ia
n;
Fe
m
a
l
e
;
Ey
e
c
o
l
o
r: Bl
ue
; Ima
g
e
wit
h
oc
c
l
usi
on; Qua
lity
sc
ore
—89.
6
Fig
u
re
1
.
Iris imag
es and
im
a
g
e
q
u
a
lity scores fo
r fo
ur
p
e
op
le withou
t g
l
asses
2.
3. I
r
is Im
ages, Im
age Q
u
ality Sc
ores
an
d Iris
Verification
unde
r Ch
an
ged
Conditi
ons
Fig
u
re
2
shows a Ch
in
ese m
a
le’s brown
iris i
m
ag
es
an
d
q
u
a
lity sco
r
es
o
f
t
h
e iris enro
ll
m
e
n
t
with
and
wi
t
h
out
gl
asses. T
h
e
st
an
d-
of
f di
st
a
n
ce fr
om
t
h
e cam
e
ra t
o
t
h
e i
n
di
vi
dual
was 2
0
c
m
. Fi
gure 2 i
n
di
cat
es
th
at th
ere was
a lig
h
t
reflection
du
e to
h
i
s g
l
asses; th
e iris qu
ality sco
r
e d
e
creased. Th
e t
w
o
im
ag
es were u
s
ed
for
verification and eac
h
was
success
f
ul, although Figure
2 (b) is a
n
occlude
d im
age wi
th the light re
fl
ection
an
d a lower
q
u
ality sco
r
e.
(a)
(b)
(a)
Without glasses; Image with
o
cclusion; Quality
score—92.6
(b)
With glasses;
Image with o
cclusion; Quality
sco
r
e—82.2
Fig
u
re
2
.
C
o
m
p
ariso
n
o
f
a
Ch
in
ese m
a
le’s
brown iris im
a
g
es and
im
ag
e q
u
a
lity scores
Glasses, sun
g
l
asses, an
d con
t
act len
s
es can
affect
th
e i
r
is imag
e qu
ality an
d
p
e
rfo
r
m
a
n
ce o
f
th
e iris
sy
st
em
. Previ
o
us resea
r
c
h
de
m
onst
r
at
ed t
h
a
t
t
h
ere were
di
ffe
rent
de
g
r
ad
at
i
ons i
n
per
f
o
r
m
a
nce for
di
f
f
ere
n
t
types of c
o
ntact lenses a
n
d that lenses
producing
larg
er artifacts
o
n
th
e iris
yi
eld
e
d m
o
r
e
deg
r
ad
ed
per
f
o
r
m
a
nce [
10]
.
Th
ree
pe
opl
e
were
t
e
st
ed
whe
n
t
h
ey
wo
re
gl
asses a
n
d
w
h
e
n
t
h
ey
di
d
n
o
t
wea
r
gl
asses
using the Iris
Access
TM
40
00
syste
m
. Th
e stan
d
-
off
d
i
stan
ce was still 2
0
cm
. Tab
l
e 1
lists th
e iris testin
g
resu
lts, in
clud
i
n
g th
e iris im
a
g
e
q
u
a
lity scores of the
en
ro
ll
m
e
n
t
an
d v
e
ri
ficatio
n
ou
tcomes u
n
d
e
r
d
i
fferen
t
co
nd
itio
ns durin
g th
e iris en
ro
ll
m
e
n
t
(with
/
w
ith
ou
t
g
l
a
sses) an
d
t
h
e iris
v
e
rifica
tion
(with
/with
ou
t
g
l
asses).
Because
the ve
rification was conducted
right after the e
n
rollm
e
nt process
(a
lm
ost at the sam
e
tim
e
), the iris
im
age score
d
u
r
i
n
g t
h
e
ve
ri
fi
cat
i
on
was
rega
rded as alm
o
st the sam
e
as th
e
score
duri
ng the enrollm
ent. Table
1
ind
i
cates th
at g
l
asses can
d
ecrease th
e iris i
m
ag
e qu
ality sco
r
es and
su
ng
lasses can lead
to
v
e
rifi
cation
failu
re. Th
e literatu
re [9
] reco
mmen
d
s
t
h
at eye g
l
asses m
u
st b
e
rem
o
v
e
d
b
e
fo
re enro
ll
men
t; ho
wev
e
r,
g
l
asses
d
i
d
no
t affect t
h
e su
ccess i
n
iris v
e
rification
alth
o
ugh
th
e i
r
is im
age score
s
dec
r
eased (se
e
the res
u
lts in
Table
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 3
,
N
o
. 4
,
Aug
u
s
t 2
013
:
42
9
–
43
5
43
2
1). T
h
e
Iris
A
c
cess
TM
4
0
0
0
s
y
st
em
had
bet
t
e
r pe
rf
orm
a
nc
e t
h
an
ex
pect
e
d
a
n
d
as
desc
ri
be
d i
n
t
h
e s
y
st
em
m
a
nual
s
.
Tabl
e 1.
The e
ffect
s
o
f
gl
asses a
n
d s
u
ngl
asse
s
on
i
r
i
s
im
age (
w
i
t
h
o
u
t
occl
usi
o
n
)
sc
ores
an
d
ve
ri
fi
cat
i
on
out
c
o
m
e
s
N
o
E
t
hnicity
Gender
Eye
colors
E
n
r
o
llm
e
nt Ver
i
fication
Iris i
m
age
scores
Ver
i
fication
outco
m
e
s
1
Af
rican
A
m
e
r
ican
M
a
le Br
own
W
ithout
glasses
W
ithout glasses
97.8
Success
With
sunglasses
With
sunglasses
58.
4
Failur
e
2 Caucasian
Fem
a
le
Blue
W
ithout
glasses
W
ithout glasses
96.6
Success
W
ith glasses
W
ith glasses
88.4
Success
3 Chinese
M
a
le
Br
own
W
ithout
glasses
W
ithout glasses
98.6
Success
With glasses
With
glasses
89.
8
Success
In
pre
v
i
o
us
res
earch
, b
o
t
h
ey
e
and
hea
d
p
o
si
t
i
ons m
u
st
be c
ont
rol
l
e
d
.
Hea
d
o
r
ey
e m
o
t
i
on d
u
r
i
n
g i
r
i
s
scanning can c
a
use im
age blur. The
b
o
d
y
can m
ove i
n
t
h
r
ee dim
e
nsi
ons;
bot
h t
h
e
head
and ey
es can m
ove
inde
pende
n
tly. Increasi
n
g the stability
of
the
body, t
h
e
hea
d
, and the
eyes inc
r
eases the
accurac
y
and
per
f
o
r
m
a
nce i
n
i
r
i
s
scanni
n
g
and i
r
i
s
ve
ri
fi
c
a
t
i
on [3]
,
[5
]. Howev
e
r, th
e resu
lts in
Table 2 indicate that there
were
no m
o
t
i
on-i
n
d
u
ce
d effe
ct
s on i
r
i
s
im
age sco
r
es an
d
th
at th
ere was su
ccess in
iris verificatio
n
wh
en
t
h
e
Chinese m
a
le (brown eyes,
wi
t
h
o
u
t
gl
as
ses)
s
h
o
o
k
hi
s
head
,
no
d
d
ed
,
or
ha
d
ey
e m
o
t
i
on d
u
r
i
n
g t
h
e
en
r
o
l
l
m
ent
and t
h
e ve
ri
fi
c
a
t
i
on. T
h
e st
an
d-
of
f di
st
a
n
ce
fr
om
t
h
e cam
e
ra t
o
t
h
e i
n
di
vi
dual
was al
so
20 cm
. Tabl
e 2 al
so
indicates that the Iris
A
ccess
TM
400
0 sy
st
em
has bet
t
e
r pe
r
f
o
r
m
a
nce t
h
an
expect
ed a
nd
as descri
bed i
n
t
h
e
syste
m
m
a
nual [9].
Tabl
e 2.
The e
ffects
of
m
o
tions
on iris
im
age scores
E
n
r
o
llm
e
nt
No m
o
tion
Head
shaking
Nodding
E
y
e m
o
tion
I
r
i
s i
m
age scor
es
98.
6
98.
6
98.
6
98.
6
Tab
l
e
3
sho
w
s th
e C
h
in
ese male’s
b
r
o
w
n iris im
ag
e
scores
under cha
n
ges
in t
h
e sta
n
d-off
distance
.
Th
e testing
resu
lts d
e
m
o
n
s
trate th
at th
e stand
-
o
f
f
d
i
stance
from
18-25cm
is the im
age capture ra
nge
of t
h
e
IrisAccess
TM
4000 system
and that
distance
ch
ang
e
s with
i
n
th
is rang
e
d
o
not a
ffect i
r
is i
m
age scores.
Tabl
e 3.
The
e
ffects of change
d
sta
n
d-off
distances
on iris im
age sc
ores
Stand-
off di
stances
(c
m
)
17
18
20
22
24
25
26
Iris i
m
age quality
scores
Cannot captur
e
iris i
m
ages
98.
6
98.
6
98.
6
98.
6
98.
6
Cannot captur
e
iris i
m
ages
Man
y
p
r
o
f
essio
n
a
ls typ
i
cally th
o
u
g
h
t
th
at th
e iris cap
tu
re process
was sensitiv
e to
lig
h
ting
co
nd
itio
ns p
r
esen
t in
th
e testin
g
ro
o
m
an
d th
at n
o
d
i
rect
o
r
artificial li
g
h
t
sh
ou
ld
d
i
rectly reflect o
ff th
e
enrollee’s eyes [3]. Howe
ver, after obt
aini
ng iris im
age sc
ores
of the enro
l
l
m
e
nt
when l
i
ght
s we
re o
n
and
of
f
in the la
b, t
h
ere was
alm
o
st no
diffe
re
nce in
iris im
age scores.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Iris Ima
g
e
Quality Testin
g
a
n
d
Iris Verifica
t
io
n
(Li
d
ong
Wa
ng
)
43
3
3.
R
E
SU
LTS AN
D ANA
LY
SIS FO
R T
H
REE EYE COL
O
RS
3.
1.
Da
ta
a
nd
D
e
s
cript
iv
e St
at
is
t
i
cs
f
o
r
Iris
Imag
e Qua
lity
Sco
res
In add
itio
n to th
e
fou
r
su
bj
ect
s tested
, twen
ty eig
h
t
add
itio
n
a
l stud
en
ts at
th
e
u
n
i
v
e
rsity
were inv
ited
to
p
a
rticip
ate i
n
iris enro
llm
e
n
t and
v
e
rificatio
n
tests i
n
Novem
b
er,
2012 t
o
st
udy t
h
e
differe
n
ce i
n
iris i
m
age
quality am
ong
three types of eyes in
brown, hazel, and gre
e
n. T
h
ese stude
nts were
18-25 y
ears old. African
Am
erican students we
re dominant at the university
;
C
a
ucasi
a
n st
u
d
ent
s
were a m
i
nori
t
y
gro
u
p
. M
o
st
of t
h
e
African Am
erican
stude
n
ts had brow
n ey
es
. Am
on
g t
h
e
28
st
u
d
ent
s
, 1
6
st
u
d
e
n
t
s
ha
d
br
o
w
n
ey
es;
seve
n
stude
nts ha
d hazel eyes; and
five stude
n
ts had green eyes.
Although t
h
e five and seve
n
meet the require
m
e
nt
[
1
1
]
fo
r th
e
H
t
e
st, it was e
x
pe
cted to
find m
o
re st
ude
n
ts with hazel
ey
es
or green eyes.
Tab
l
e
4
sh
ow
s a br
eakdo
wn
in
ethn
icity, gen
d
e
r
,
ey
e c
o
l
o
r, and i
r
is image sc
ores
. T
h
e Stand-off
di
st
ance
was
2
0
cm
. No
ne o
f
t
h
e st
u
d
ent
s
w
o
re
gl
asses
d
u
r
i
ng t
h
e i
r
i
s
en
r
o
l
l
m
e
nt
and i
r
i
s
veri
fi
cat
i
on.
Al
l
of
the ve
rifications we
re s
u
ccess
f
ul.
Tabl
e
5 s
h
ows
t
h
e m
ean
and
t
h
e st
a
nda
r
d
devi
at
i
o
n
of
t
h
e i
r
i
s
i
m
age
score
s
f
o
r t
h
e
s
t
ude
nt
s
wi
t
h
ey
es i
n
br
o
w
n
,
ha
zel
, an
d
gree
n.
Tabl
e 4.
Dem
ogra
phi
cs
of
t
h
e st
ude
nt
s
part
i
c
i
p
at
ed
i
n
i
r
is scanning te
sts and thei
r iri
s
im
age scores
No
E
t
hnicity
Gender
E
y
e color
I
r
i
s scor
e
1 Afr
i
can
Am
er
ican
M
a
le
Br
own
99.
0
2 Afr
i
can
Am
er
ican
M
a
le
Br
own
89.
6
3 Afr
i
can
Am
er
ican
M
a
le
Br
own
98.
6
4 Afr
i
can
Am
er
ican
M
a
le
Br
own
97.
9
5 Afr
i
can
Am
er
ican
M
a
le
Br
own
99.
0
6 Afr
i
can
Am
er
ican
M
a
le
Br
own
98.
6
7 Afr
i
can
Am
er
ican
M
a
le
Br
own
98.
6
8 Afr
i
can
Am
er
ican
M
a
le
Br
own
89.
6
9 Afr
i
can
Am
er
ican
Fem
a
le
Br
own
89.
6
10
Afr
i
can
Am
er
ican
Fem
a
le
Br
own
83.
5
11
Afr
i
can
Am
er
ican
Fem
a
le
Br
own
98.
6
12
Afr
i
can
Am
er
ican
Fem
a
le
Br
own
98.
6
13
Caucasian
M
a
le
Br
own
98.
6
14
I
ndian
M
a
le
Br
own
98.
6
15
I
ndian
M
a
le
Br
own
97.
9
16
Chinese
M
a
le
Br
own
98.
6
17 Af
rican
A
m
e
r
ican
Male
Hazel
98.6
18 Af
rican
A
m
e
r
ican
Male
Hazel
97.9
19 Af
rican
A
m
e
r
ican
Male
Hazel
98.4
20 Af
rican
A
m
e
r
ican
Fe
m
a
le
Hazel
89.7
21 Af
rican
A
m
e
r
ican
Fe
m
a
le
Hazel
98.8
22 Af
rican
A
m
e
r
ican
Fe
m
a
le
Hazel
99.0
23 Caucasian
Fe
m
a
le
Hazel
98.6
24 Caucasian
Male
Green
98.6
25 Caucasian
Male
Green
99.0
26 Caucasian
Male
Green
98.4
27 Caucasian
Fe
m
a
le
Green
97.9
28 Caucasian
Fe
m
a
le
Green
89.8
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
JECE
Vo
l. 3
,
N
o
. 4
,
Aug
u
s
t 2
013
:
42
9
–
43
5
43
4
Tabl
e 5.
The m
ean
and
st
an
dar
d
devi
a
t
i
on
o
f
i
r
i
s
i
m
age sc
ore
s
f
o
r t
h
e st
ude
nt
s i
n
t
h
ree
ki
nds
ey
e
col
o
rs
Eye
co
lo
r
Brown
Hazel
Green
95.
93
97.
29
96.
74
4.
89
3.
36
3.
90
3.2.
Non-parametric Anal
ys
is
for
the Im
age Quality
of
Three E
ye C
o
l
o
rs
Table
5 shows there is di
fference in the image
qua
lity of
the three eye
co
lors (brown, hazel, a
nd
gree
n)
. P
r
o
f
ess
i
onal
s
ar
e co
nc
erne
d a
b
o
u
t
w
h
et
he
r or
not t
h
ere is a si
gni
ficant differe
n
c
e
. The
H
test,
a n
on-
param
e
t
r
i
c
m
e
tho
d
,
was
use
d
t
o
co
n
duct
a c
o
m
p
arat
i
v
e st
u
d
y
am
ong t
h
e t
h
ree ey
e c
o
l
o
rs
. T
h
e
H
test is also
called
th
e
Krusk
a
l-Wallis test [11
]
. It is a rank-su
m
te
st th
at is u
s
ed
to
test th
e
n
u
l
l h
ypo
th
esis t
h
at
k
in
d
e
p
e
nd
en
t ran
d
o
m
sa
m
p
le
s co
m
e
fro
m p
opu
latio
n
s
with
ap
prox
i
m
atel
y id
en
tical
m
ean
s ag
ai
n
s
t the
altern
ativ
e h
y
p
o
t
h
e
sis th
at th
e m
ean
s o
f
th
e pop
u
l
atio
ns
are n
o
t
al
l
equal
.
T
h
e m
a
jor a
dva
nt
ag
e of
no
n
-
param
e
t
r
i
c
m
e
t
h
o
d
s i
s
t
h
at
t
h
ey
d
o
not
r
e
qui
re s
p
eci
fi
c assum
p
t
i
ons
(s
uch
as
n
o
r
m
al
di
st
ri
but
i
o
n
o
r
app
r
oxi
m
a
t
e
n
o
rm
al
di
st
ri
but
i
o
n
)
ab
out
t
h
e
sam
p
l
e
d pop
ul
at
i
ons. T
h
ere
f
o
r
e, n
o
n
-
p
aram
et
ri
c
m
e
t
hods c
a
n be
use
d
un
der
m
o
re gene
ral
c
o
n
d
i
t
i
ons
.
Th
e
d
a
ta of the sam
p
les are rank
ed
jo
in
tl
y fro
m
lo
w to h
i
gh
as thoug
h
t
h
ey con
s
titu
te a sin
g
l
e
sam
p
l
e
. If
i
s
t
h
e sum
of t
h
e ra
n
k
s
assi
gne
d t
o
t
h
e
val
u
es
of t
h
e
i
th sam
p
le an
d
, the
H
test is
based
on
th
e fo
llo
wi
n
g
statistic:
(1
)
If eac
h sam
p
le has at least five obse
rvations and t
h
e calc
u
lated
H
is
g
r
eater th
an
o
r
eq
u
a
l to
[1
1]
f
o
r
de
gr
ees of
f
r
eed
o
m
,
t
h
e n
u
l
l
hy
pot
hesi
s s
h
o
u
l
d
be
re
ject
ed
and t
h
e al
t
e
r
n
a
t
i
v
e hy
p
o
t
h
esi
s
shoul
d
be acce
pted.
is the leve
l of significance. T
h
e followi
ng
null hy
pot
hesis is form
ulated:
Th
ere is no
statistical
ly sig
n
i
fican
t d
i
fferen
ce in
th
e iris i
m
ag
e qu
ality sco
r
es
of th
e t
h
ree k
i
n
d
s of
eyes in
brown,
hazel, a
n
d gree
n.
The
outcom
e is: the
hypothes
is is accepte
d
or re
jecte
d
at .
I
n
t
h
is stud
y,
k
= 3;
= 16;
= 7;
= 5;
an
d
= 28
. A
r
ra
n
g
i
n
g t
h
e dat
a
i
n
Ta
bl
e 4
joi
n
t
l
y
according to si
ze and assi
gni
ng the data the
ranks 1,
2,
3,
…, and 28; thus,
= 210.5,
= 119.5, a
n
d
= 7
6
.
Su
bstitu
tin
g
th
ese v
a
l
u
es in
to
form
u
l
a (1
),
H
was ob
tain
ed
and
H =
1.1
5
.
i
s
gi
ven
i
n
TAB
L
E
IV [
1
1]
.
= 5.
99
1 f
o
r
and
deg
r
ees
of f
r
ee
dom
. Si
nce t
h
e
cal
cul
a
t
e
d
H =
1
.
15
is less
than 5.991, t
h
e
null hy
pot
hesi
s m
u
st be acce
pted;
the
r
e
is
no significa
n
t
diffe
re
nce i
n
t
h
e iris
im
age
quality
score
s
of eyes
in
brown,
hazel, and
green when the
leve
l of significa
nce
is 0.05.
4.
CO
NCL
USI
O
N
Iris im
ages and
quality score
s
for
four
kinds of ey
es
(brown, hazel, gree
n,
a
n
d blue)
wi
thout
glasses
were ca
ptured
thro
ugh the Iri
s
Access
TM
4000
system
an
d
th
e ir
is v
e
r
i
f
i
cat
io
n
s
f
o
r
th
e
f
o
u
r
k
i
nd
s
o
f
eyes w
e
r
e
successful. The
iris
verificati
ons with glasses
were s
till s
u
ccessful although
glasses ca
n dec
r
ease iris i
m
age
q
u
a
lity scores
d
u
e
to
ligh
t
reflectio
n
.
Sun
g
l
a
sses can
also lead
to v
e
rificatio
n failure.
There
we
re
no
m
o
t
i
on-i
n
d
u
c
e
d ef
fect
s
on t
h
e i
r
i
s
im
age scores
and the
success i
n
iris
verification
whe
n
t
h
e C
h
i
n
ese m
a
l
e
(br
o
w
n
ey
es, wi
t
h
o
u
t
gl
asses
)
had s
o
m
e
head sha
k
i
n
g, n
o
ddi
ng
, an
d ey
e
m
o
t
i
on
during the enrollm
e
nt and
ve
rification. T
h
e
IrisAccess
TM
4000 system
can ca
pture ir
is i
m
ag
es if th
e stan
d-o
ff
distance ra
nge
s from
18-25c
m
.
Distance ch
an
g
e
s with
in
t
h
is rang
e do
no
t a
ffect iris image scores
. There was
al
m
o
st n
o
d
i
fferen
c
e in iris imag
e scores
wh
en
t
h
e ligh
t
s we
re on
a
nd w
h
e
n
t
h
e l
i
ght
s we
re of
f i
n
t
h
e l
a
b. The
IrisAccess
TM
4
0
0
0
sy
st
em
has
bet
t
e
r
per
f
o
rm
ance t
h
a
n
ex
pe
ct
ed an
d a
s
des
c
ri
be
d i
n
t
h
e
sy
st
em
m
a
nual
.
Acco
r
d
i
n
g t
o
t
h
e
resul
t
s
o
b
t
a
i
n
ed
fr
om
t
h
e
no
n
-
pa
ram
e
t
r
i
c
m
e
t
hod
based
o
n
t
h
e
H
test, at th
e
0.05
level of signifi
cance, the
r
e is no si
gnifica
nt diffe
re
nce
in the iris i
m
age quality of
eyes i
n
brown, hazel
, and
gree
n.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Iris Ima
g
e
Quality Testin
g
a
n
d
Iris Verifica
t
io
n
(Li
d
ong
Wa
ng
)
43
5
REFERE
NC
ES
[1]
RM Bolle, JH C
onnell, S Pankan
ti, NK Ratha,
and AW Senior.
Guide to
Biome
t
rics
.
Springer-V
erlag,
New York,
2004.
[2]
JD Woodwar, NM Orlans and
P
T Higgins.
B
i
ometr
i
cs
. California: McGraw-Hill/Osborne. 2003.
[3]
ND Kalka, J Zu
o, NA Schmid, and B Cukic.
“I
mage Quality
Assessment for Iris Biometri
c”
.
P
r
oceed
ings
of S
P
I
E
Biometric Techn
o
log
y
for Human Identification I
II. vol. 6202
, pp
.
445-452,
Orlan
do, FL, USA, 20
06.
[4]
R Naray
a
nswamy
, PEX Silv
eira, H Setty
,
VP Pa
uca, and JVD Gracht.
“Extend
e
d depth-o
f
-field
iris recognitio
n
sy
ste
m
for a wor
k
station
e
n
vironme
n
t”
. Proceedings of the SPIE
Biometric Te
chn
o
log
y
for Human Identification
I
I
.
vol. 5779
, pp
. 41
-50, Orlando, FL, USA, 2005.
[5]
A Ross. “
Iris Re
cognition
:
the
Pa
th Forward”
.
Co
mputer
. pp
. 30-3
5
, Februar
y
201
0.
[6]
W Dong, Z Sun, T Tan.
“A d
e
s
i
gn of iris reco
gniti
on system at a distance”.
Proceedings of
the 2009 Chin
e
s
e
Conference on
Pattern Recogniti
on (CCPR 2009)
, Nanjing, China. November 4-6
,
2009, pp
. 1-5
.
[7]
JR Matey
,
O Naroditsky
,
K Hanna,
R
Kolczy
nsk
i
, DJ LoI
acono
,
S Mangru, M
Tinker,
TM Zappia,
and WY Zhao
.
“Iris on the Mo
ve: Acqu
isition
of Images for Iri
s
Recognition in
Less Con
s
trained Environments”.
P
r
oceedings
of
the I
EEE. vo
l. 9
4
, No. 11,
pp.19
36-1947, Novem
b
er 2006.
[8]
LG El
ectron
i
cs
-
Iris
T
echno
log
y
Divis
i
on,
Iris
A
c
ces
s
TM
4000 Har
d
ware Manu
al,
New Jersey
, US
A, 2008.
[9]
LG Ele
c
troni
cs
- Iris
Techno
log
y
Divis
i
on
, Iris
A
cces
s
TM
Software User Manual,
Vers
ion 3.
00,
New Jersey
, USA,
Decem
ber13,
20
07.
[10]
SE Baker, A H
e
ntz, KW Bowy
er
, and PJ Fl
ynn. “Deg
radat
i
o
n
of Iris Recog
n
ition Perform
ance Due to No
n-
Cos
m
etic P
r
es
cri
p
tion Conta
c
t
Le
ns
es
”.
Computer
Vision and Image Understanding
. vol. 114
, no.
9, pp. 1030-104
4
,
September, 2010
.
[11]
JE Freund and B
M
Perles.
Statistics: A First Course
. (8
th
Ed
.), Pearson Prentice Hall, New Jersey
, 2
004.
BI
O
G
R
A
P
HY
OF
A
U
T
HO
R
Dr. Lidong Wan
g
is the Dir
ector
of the Automa
ted Identification
Techno
log
y
(AI
T
) Program
and an Assistant Professor in th
e Department
of
Applied Techn
o
log
y
at Mississ
ippi Valley
State Univ
ersity
,
USA. He had conducted r
e
sear
ch
at
the Univ
ersity
of South Caro
lina, Ohio
State Univ
ersity
, and Mississippi
State Univ
ersity
;
and condu
cted projects supported b
y
th
e
Department of
Defense (DOD), the Nation
a
l Scien
ce Foundation (NSF), and the National
Aeronauti
c
s
and S
p
ace Adm
i
nis
t
ration (NAS
A) be
fore he m
oved to M
i
s
s
i
s
s
i
ppi
Vall
e
y
S
t
a
t
e
Univers
i
t
y
in 20
07. His
curr
ent
res
earch
int
e
res
t
s
includ
e: b
i
om
etri
cs
and r
a
dio
freque
n
c
y
identif
ication
(R
FID). He has
pu
blished ov
er 40
papers in
var
i
ou
s journals.
Dr. Wang has b
een invited to review papers
b
y
over 10 professional journals. He has also
been invited b
y
four professional journals to ac
t as their guest ed
itor. He has bee
n
the Editor
-
in-Chief of th
e I
n
terna
tiona
l Journal of Autom
a
te
d Identifi
c
a
tion
Techno
log
y
(IJAIT) for fiv
e
ye
ar
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.