Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
Vol
.
4
,
No
. 5, Oct
o
ber
2
0
1
4
,
pp
. 75
8~
76
6
I
S
SN
: 208
8-8
7
0
8
7
58
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
Securit
y
of Bi
ometric Dat
a
Using Compress
ed W
a
t
e
rmarking
Technique
R
o
hit
Tha
n
ki*
,
Koma
l B
o
risa
ga
r**
*Faculty
of Technolog
y
&
Engineering
,
C
U Shah University
,
Wadhwan, India
**Department of
Electron
i
cs
and
Communi
cation
Engineering, Atmiy
a Institu
te of
Technolog
y
& Science,
Rajko
t
, India
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
J
u
l 22, 2014
Rev
i
sed
Sep 9, 20
14
Accepted
Sep 20, 2014
This paper
has
focus on biometric da
ta security over open
communication
channel of biom
etric s
y
stem
. Here biometric data is
encod
e
d us
i
ng cs
theor
y
and wavelet bas
e
d embedding technique. Th
e b
i
ometric data is
convert in
to
encoded sparse
measuremen
ts which is generatin
g
using SVD, random seed
and uniform quantization process. Then
th
es
e enc
oded s
p
ars
e
m
eas
urem
ents
are embedding
into the host
color bi
ometric data using wavelet based
watermarking technique.
Th
is proposed techniq
u
e has exp
l
ored
dimension
reduction and co
mputational s
e
curity
pr
ovided by
compressive s
e
nsing. Th
is
proposed technique has also helps to co
mpresse
d and to send se
cret data over
nois
y
communication chann
e
l
of
biometri
c
s
y
s
t
e
m
agains
t var
i
o
u
s
atta
cks
.
The proposed
technique provides more s
ecurity
compare to
ex
isted techniq
u
e
in liter
a
ture due to CS
th
eor
y
. The
nov
elty
of pr
oposed technique is th
at,
waterm
ark infor
m
ation is com
p
ressed
and encoded iris image using C
S
theor
y
and un
ifo
r
m
quantiz
ation
.
Keyword:
Bio
m
e
t
ric Data
C
o
m
p
ressi
ve S
e
nsi
n
g T
h
e
o
ry
Sin
g
u
l
ar Val
u
e Deco
m
p
o
s
itio
n
Discrete Wavelet
Tran
sform
Waterm
arkin
g
Copyright ©
201
4 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
:
R
ohi
t
T
h
an
ki
,
R
e
search
Sc
ho
l
a
r,
Depa
rt
m
e
nt
of
El
ect
ro
ni
cs
an
d C
o
m
m
unicat
i
on E
n
gi
nee
r
i
n
g,
Facu
lty of
Tech
no
log
y
an
d En
g
i
n
eer
i
n
g, C
U
Sh
ah
U
n
i
v
ersity,
Sur
e
n
d
ra
na
gar
– Ahm
e
daba
d Hi
g
h
way
,
W
a
d
h
wa
n
C
i
t
y
, Gu
j
a
rat
,
I
n
di
a.
Em
a
il: ro
h
itth
an
k
i
9
@
g
m
ail.c
o
m
1.
INTRODUCTION
Bio
m
e
t
ric te
mp
late b
a
sed
au
t
h
en
tication
syste
m
is u
s
ed
for hu
m
a
n
id
en
tificatio
n
and
au
t
h
en
tication
in orga
nization in prese
n
t world because
of e
v
ery individual
having unique
biom
etric characteristics [1, 2]. In
20
0
1
, R
a
t
h
a a
n
d i
t
s
resea
r
c
h
t
e
am
are i
d
e
n
t
i
f
i
e
d se
veral
vul
nera
bl
e
poi
nt
s
i
n
bi
om
et
ri
c aut
h
ent
i
cat
i
o
n
sy
st
em
[3]
.
T
h
ere a
r
e
m
o
st
im
port
a
nt
v
u
l
n
e
r
ab
l
e
poi
nt
i
n
bi
om
et
ri
c sy
stem
i
s
t
e
m
p
l
a
t
e
m
odi
fi
cat
i
on at
com
m
uni
cat
i
on cha
n
nel
bet
w
een t
w
o m
odul
es
. A
K
Jai
n
an
d i
t
s
rese
arch t
eam
are pr
op
ose
d
t
h
at
di
gi
t
a
l
wat
e
rm
arki
ng
i
s
one o
f
sol
u
t
i
on f
o
r t
h
i
s
i
s
s
u
e [
4
]
.
They
are al
so p
o
i
n
t
o
u
t
som
e
di
sadvant
a
g
es i
n
bi
om
et
ri
c
sy
st
em
li
ke senso
r
noi
se,
di
f
f
e
rent
vari
at
i
on
i
n
dat
a
ba
se, se
curi
t
y
an
d p
r
i
v
acy
of bi
om
et
r
i
c t
e
m
p
l
a
t
e
. For o
v
e
r
com
e
of t
h
ese di
sad
v
a
n
t
a
ges,
A. R
o
ss a
nd i
t
s
researc
h
t
e
am
are gi
ve
new
b
i
om
et
ri
c sy
st
em
whi
c
h i
s
cal
l
e
d as
m
u
l
t
i
m
o
d
a
l b
i
o
m
etric syste
m
wh
ere two o
r
m
o
re b
i
ometric te
m
p
la
te is u
s
ed
for id
en
tification
an
d
aut
h
e
n
t
i
cat
i
on
of i
ndi
vi
d
u
al
[
5
,
6]
. B
u
t
w
h
e
n
m
u
l
t
i
m
odal
bi
om
et
ri
c sy
st
em
i
s
desi
g
n
f
o
r
l
a
rge
scal
e bi
o
m
et
ri
c
dat
a
, t
h
en
bi
o
m
et
ri
c t
e
m
p
l
a
te can
be
easi
l
y
rec
onst
r
uct
e
d
fo
rm
st
ored
fe
at
ure at
sy
st
em
dat
a
base
by
i
m
post
e
r
an
d
th
is situ
at
io
n
is in
troduced
security o
f
b
i
o
m
etric te
m
p
la
te p
r
o
t
ectio
n
issu
e in
m
u
l
t
i
m
o
d
a
l b
i
o
m
etri
c
syste
m
.
In
few l
a
st
y
ears, m
a
ny
researche
r
s a
r
e
pr
o
pose
d
a
n
d
descri
bed
wa
t
e
rm
arki
ng t
e
c
hni
que
f
o
r
bi
om
et
ri
c dat
a
pr
ot
ect
i
o
n
.
He
r
e
som
e
of t
ech
ni
q
u
es are
re
vi
ewed
w
h
i
c
h i
s
rel
a
t
e
d t
o
pr
o
pos
ed
w
o
r
k
.
A
u
t
h
or
in [7]
descri
be
d im
age watermarking
technique where wa
velet
coefficients
are m
odifi
ed according to bit of
wat
e
rm
ark i
n
f
o
rm
at
i
on.
Aut
h
o
r
i
n
[8]
des
c
ri
be
d
bl
i
n
d
bi
om
et
ri
c wat
e
rm
arki
n
g
sc
he
m
e
fo
r si
g
n
at
u
r
e
usi
n
g
wav
e
let tran
sfo
r
m
wh
ere seco
nd
lev
e
l
d
e
tails wav
e
let co
efficien
ts of
h
o
s
t im
ag
e is
m
o
d
i
fied
acco
r
d
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE Vo
l. 4
,
N
o
. 5
,
O
c
tob
e
r
20
14
:
758
–
7
66
7
59
si
gnat
u
re
bi
t
.
Aut
h
o
r
i
n
[
9
]
pr
o
p
o
s
ed
a
D
W
T an
d
DC
T base
d
wat
e
r
m
arki
n
g
t
ech
n
i
que
fo
r m
u
l
t
i
m
odal
bi
om
et
ri
c dat
a
pr
ot
ect
i
o
n.
In
t
h
i
s
t
ech
ni
q
u
e
,
S
VD i
s
app
lied
o
n
b
i
o
m
etri
c waterm
ark
data an
d th
is
data is
em
beddi
ng i
n
t
o
hi
ghe
r f
r
e
q
uency
wavel
e
t
coef
fi
ci
ent
s
.
Aut
h
o
r
i
n
[1
0
]
descri
be
d r
o
bust
wat
e
rm
arki
n
g
technique to e
nha
nce sec
u
rity of
m
u
ltim
odal biom
e
t
ric
syste
m
. In this proposed techni
que
, first face
feature
t
a
ken a
s
wat
e
r
m
ark
dat
a
w
h
i
c
h i
s
em
bed
d
i
n
g i
n
t
o
fi
n
g
er
pri
n
t
i
m
age usi
n
g
bl
i
n
d
SS
–
Q
I
M
schem
e
.
Aut
h
o
r
i
n
[
11]
descri
be
d a n
e
w aut
h
e
n
t
i
cat
i
on sc
hem
e
based o
n
m
u
l
t
i
modal
bi
om
et
ri
c veri
fi
cat
i
o
n
and
ri
d
g
el
et
t
r
a
n
sf
orm
wat
e
rm
arki
ng t
e
c
h
ni
q
u
e f
o
r face
an
d
i
r
i
s
dat
a
fo
r
D
i
gi
t
a
l
R
i
ght
s M
a
nagem
e
nt
(
D
R
M
).
Aut
h
o
r
i
n
[1
2]
pr
op
ose
d
m
u
l
t
i
m
odal
bi
om
et
ri
c wat
e
rm
arki
ng t
ech
ni
q
u
e
base
d o
n
co
rr
el
at
i
on anal
y
s
i
s
f
o
r
securi
t
y
of
bi
o
m
et
ri
c dat
a
over com
m
uni
cati
on cha
n
nel
of
net
w
o
r
k. I
n
t
h
i
s
pr
o
p
o
s
ed t
echni
que
, aut
h
ors a
r
e
fi
rst
fi
n
d
co
rr
el
at
i
on bet
w
ee
n wat
e
rm
ark
and
ho
st
im
age usi
n
g PLS
and P
S
O a
nd
base
d o
n
t
h
i
s
resul
t
.
B
i
om
et
ri
c dat
a
are em
bedd
i
ng i
n
t
o
h
o
st
im
age. Aut
h
or i
n
[1
3]
pr
op
ose
d
LSB
and
wa
vel
e
t
base
d
waterm
arking
technique
for
secure
face
fe
ature em
bed
in
to
fing
erprin
t
im
ag
e. Au
t
h
o
r
in
[1
4
]
p
r
op
o
s
ed
vari
ous
wat
e
r
m
arki
n
g
t
echni
que
per
f
o
r
m
a
nce for sec
u
ri
t
y
of
user
veri
fi
c
a
t
i
on base
d o
n
fi
nge
rp
ri
nt
an
d faci
al
im
age. The aut
h
ors claim
that
this approac
h
im
proves accuracy of us
e
r
ve
rification and
wa
term
ark detection.
Aut
h
o
r
i
n
[
1
5]
pr
op
ose
d
wat
e
rm
arki
ng t
e
c
hni
que
base
d
on cs t
h
eo
ry
whi
c
h i
n
cl
u
d
e
s
com
p
ressed
sensi
n
g
pr
ocess a
nd c
o
m
p
ressed sensi
ng
reco
ve
ry
pr
ocess. T
h
i
s
t
e
c
hni
que i
s
m
o
re
ro
bu
st
an
d sec
u
re a
g
ai
nst
di
ff
erent
waterm
arking attacks.
The w
o
r
k
p
r
es
ent
i
n
pape
r ari
s
es fr
om
devel
opi
ng
wat
e
rm
arki
ng t
ech
ni
q
u
e
for m
u
l
t
i
m
odal
bi
om
et
ri
c
syste
m
where iris data e
m
beds as a biom
etric data.
Now
a day, iris is a
n
accepted trai
t for worl
dwi
d
e as
i
ndi
vi
dual
a
u
t
h
ent
i
cat
i
on.
In t
h
i
s
pa
per
,
a t
e
c
hni
que i
s
pr
o
p
o
se
d w
h
i
c
h em
bed
s
enc
o
ded s
p
arse m
easure
m
ent
s
o
f
iris d
a
ta as
a waterm
ark
in
first lev
e
l horizon
tal a
nd
ve
rtical wavelet coefficien
ts of r
e
d
ch
ann
e
l of
ho
st
color face im
age. T
h
e host color face
im
age
is decom
posed using si
ngle leve
l discrete wa
velet transform
.
The
enco
de
d s
p
a
r
s
e
m
easurem
en
t
s
of
i
r
i
s
dat
a
i
s
ge
ne
rat
e
d
usi
n
g
Si
n
gul
a
r
Val
u
e
Deco
m
posi
t
i
on (
S
V
D
)
an
d
C
o
m
p
ressi
ve S
e
nsi
n
g T
h
e
o
ry
fram
e
wor
k
.
We ha
ve e
xpl
ore
d
s
p
arse
ness
p
r
o
v
i
d
e
d
by
S
V
D t
o
ge
nerat
e
l
i
nea
r
measurem
ent vector.
W
e
ha
ve
borrowe
d
th
e
id
ea fro
m
[7 an
d 25
]
with
sign
ifican
t m
o
d
i
fi
catio
n
i
n
techniq
u
e
.
The
p
r
o
p
o
sed
wo
rk
al
so
goe
s a st
e
p
fu
rt
he
r
whe
r
e
bi
om
et
ri
c dat
a
i
s
co
m
p
ressed a
n
d
enco
de
d
by
C
S
t
h
e
o
ry
fram
e
wor
k
an
d u
n
i
f
o
r
m
quant
i
zat
i
on res
p
e
c
t
i
v
el
y
before
em
beddi
ng i
n
t
o
h
o
st
m
e
di
um
. The reset
of pap
e
r
or
ga
ni
zed s
u
c
h
t
h
at
sect
i
o
n
2 d
e
scri
bed
basi
c co
nce
p
t
of C
S
t
h
eo
ry
, sect
i
o
n 3
descri
be
d p
r
op
ose
d
wat
e
rm
arki
ng
t
echni
que
an
d s
ect
i
on 4 descri
bed
ex
pe
ri
m
e
n
t
resul
t
s
a
n
d l
a
st
sect
i
on
5
des
c
ri
be
d c
oncl
u
si
on
.
2.
BASI
C
CO
N
C
EPT OF
C
O
MP
RESSI
VE
SENSI
N
G
TH
EORY
Any
si
gnal or im
age
can be reconstructed
from
its
Fourier coe
fficie
n
ts s
u
ccess
f
ully
prove
d by D.
D
ono
ho
and
E. Cand
ès in 2006
[19
,
20
]. Based
o
n
th
is
con
c
ep
t, E. Cand
ès
in
tr
odu
ced n
e
w
sign
al
p
r
o
c
essin
g
th
eory is called
C
o
m
p
ressive or Co
m
p
ressed
Sen
s
i
n
g T
h
eory. Com
p
res
s
ive se
nsing is
a new technology
whe
r
e signal or im
age is
ac
qui
red in a com
p
ressed
format. Any im
age can be acqui
red in a com
p
ressed
fo
rm
at
usi
ng
b
e
l
o
w e
q
uat
i
o
n
1 a
n
d
2:
f
x
(1
)
x
A
y
(2
)
Whe
r
e y is a sparse m
easure
m
en
ts o
f
im
ag
e with
size o
f
M
1 (M
N), A is a
m
easu
r
e
m
en
t
m
a
trix
wh
ich
is
sam
e
for em
bedde
r a
n
d
det
ect
or
si
de
(M
N)
,
is
b
a
sis tran
sfo
r
m
a
tio
n
wh
ich
is app
lied
o
n
im
ag
e, x
is
a sp
a
rse c
o
efficient
with size
of
N
1,
f
i
s
a
n
ori
g
i
n
al
i
m
age.
H
e
re M
i
s
deci
di
ng
fact
or
f
o
r
di
m
e
nsi
onal
re
d
u
ct
i
o
n
and im
age compressi
on rate.
Th
e cs
reco
v
e
ry p
r
o
cess is i
n
v
e
rsi
o
n of cs acq
u
i
sition
.
Wh
ere co
m
p
ressed
d
a
ta are
fed to
so
m
e
n
o
n
-
lin
ear
o
p
tim
iza
tio
n
algorith
m
to
rep
r
od
u
c
ed
th
e co
m
p
lete sig
n
a
l
o
r
im
ag
e.
Wh
en
m
easu
r
em
en
t m
a
trix
A is
sat
i
s
fi
es R
I
P pr
ope
rt
y
and i
n
c
ohe
re
nce pr
o
p
e
rt
y
[21]
t
h
e
n
sparse c
o
ef
fi
ci
ent
s
x
can be ge
t form
measure
m
ents
usi
n
g
bel
o
w e
q
uat
i
o
n
3:
x
A
y
t
s
S
.
.
1
min
(3
)
Whe
r
e
S is indicated CS re
covery al
gorithm like L
1
m
i
nim
i
zat
i
on, B
a
si
s
p
u
rs
ui
t
,
OM
P.
3.
PROP
OSE
D
WATERMARKI
N
G
TE
CHNI
QUE
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Secu
rity of Bi
o
m
etric D
a
ta
U
s
ing
C
o
mpr
e
ss
ed W
a
term
arki
ng
Tec
hni
que
(
R
oh
it Than
ki)
76
0
Thi
s
sect
i
on
d
e
scri
bes t
h
e
pr
op
ose
d
wat
e
r
m
arki
n
g
t
echni
que
base
d o
n
com
p
ressi
ve s
e
nsi
n
g t
h
e
o
ry
and
co
rrel
a
t
i
o
n
pr
ope
rt
i
e
s o
f
Pse
u
d
o
R
a
n
dom
Noi
s
e
(P
N) sequ
en
ce in
wav
e
let domain
.
In th
is
p
r
op
o
s
ed
technique, col
o
r
face im
age is taken as
host m
e
diu
m
a
nd a
p
plied single level disc
rete wavelet transform
(DWT
) on re
d com
p
onent
of
host face i
m
age and
ge
t
HL a
nd L
H
wavelet c
o
efficients for
waterm
ark
em
beddi
ng
. T
h
e i
r
i
s
i
m
age i
s
t
a
ken as
wat
e
r
m
ark m
e
di
um
an
d wh
ich is co
nv
er
ted
i
n
to s
p
arse
m
easurements
usi
n
g cs t
h
e
o
r
y
. The
uni
f
o
rm
qua
nt
i
zat
i
on i
s
use
d
f
o
r e
n
c
odi
ng
spa
r
se
m
easurem
ent
s
of i
r
i
s
i
m
age i
n
t
o
bi
t
0
and
1.
The
s
e e
n
co
de
d s
p
arse
m
easurem
ent
s
of i
r
i
s
i
m
age t
a
ken
as wat
e
r
m
ark i
n
fo
rm
ati
on
whi
c
h i
s
e
m
bed
i
n
t
o
c
o
l
o
r fac
e
im
age. The
pr
o
p
o
s
ed t
e
c
hni
que
i
s
di
vi
ded
i
n
t
o
t
h
ree
part
s l
i
ke wa
t
e
rm
ark pre
p
a
r
at
i
o
n
,
wat
e
rm
ark em
bed
d
i
n
g p
r
oce
d
u
r
e an
d wat
e
rm
ark ext
r
act
i
on & rec
o
nst
r
uct
i
o
n p
r
oce
d
ure
.
The p
r
op
ose
d
wat
e
rm
ark em
bed
d
i
n
g
pr
oce
d
u
r
e an
d e
x
t
r
a
c
t
i
on & rec
o
n
s
t
r
uct
i
o
n p
r
oc
edu
r
e i
s
sh
o
w
n i
n
fi
gu
re 1
and
2
resp
ectiv
ely.
In
fi
g
u
res
d
o
t
t
e
d
box
h
a
v
e
sho
w
n
cs
acq
u
i
sition
proced
ure at emb
e
dd
er sid
e
an
d
cs
reco
nst
r
uct
i
o
n pr
oce
d
u
r
e
at
d
e
t
ect
or
si
d
e
.
3.
1.
W
a
term
a
r
k Prep
ar
ati
o
n
The
waterm
ark prepa
r
ation st
eps a
r
e
descri
bed
below:
Take waterm
ark biom
etric
image
a
n
d
com
p
uter size of im
a
g
e
whic
h is
N
N.
Ap
pl
i
e
d si
ng
ul
ar val
u
e dec
o
m
posi
t
i
on (S
V
D
)
on
wat
e
rm
ark
bi
om
et
ri
c im
age and c
o
n
v
ert
i
n
t
o
U
,
S
and
V m
a
trix
. The
S m
a
trix
v
a
lu
e
with
size
o
f
N
2
1
w
h
i
c
h i
s
t
a
ken
as s
p
a
r
se c
o
ef
fi
ci
ent
s
a
n
d
de
not
e
d
as
x.
Gene
rate m
eas
urem
ent m
a
trix
A
with
size o
f
M
N
2
(M
N)
u
s
ing r
a
ndo
m
seed
w
h
ich
is sam
e
at
em
bedde
r a
n
d
deco
de
r.
Wh
er
e M
i
s
deci
di
n
g
fact
or
f
o
r
co
m
p
ressi
on.
Gene
rate spa
r
s
e
m
easurem
ents
y
of
biom
etric im
age using
measurem
ent matrix
A
and s
p
arse
c
o
e
fficie
n
ts
x
usi
n
g
eq
uat
i
o
n
1 a
n
d
2
.
Th
en
ap
p
lied un
ifo
r
m
q
u
a
n
tizatio
n
with
2
b
it / lev
e
l on
sp
arse m
easu
r
em
en
ts and
en
cod
e
d in
to
W
(0
,
1
)
.
Now t
h
ese e
n
c
ode
d
spa
r
se m
easurem
ents are
use
d
as
sec
u
r
e
wat
e
rm
ark i
n
fo
rm
ati
on.
Fi
gu
re 1.
W
a
t
e
rm
ark
Em
bedd
i
ng Pr
oce
d
u
r
e
3.
2.
W
a
term
a
r
k E
m
bed
d
i
n
g Pr
oced
ure
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE Vo
l. 4
,
N
o
. 5
,
O
c
tob
e
r
20
14
:
758
–
7
66
7
61
The wat
e
rm
ark
em
beddi
n
g
st
e
p
s
a
r
e descri
be
d bel
o
w (ad
o
p
t
e
d fr
om
7,
1
6
,
17
an
d 1
8
):
Take
host col
o
r bi
om
etric image a
nd c
o
m
pute size of M
N
.
Ta
ke re
d
com
pone
nt
o
f
h
o
st
bi
om
et
ric
im
age whe
r
e
w
a
t
e
rm
ark i
n
f
o
r
m
at
i
on i
s
em
bedde
d.
Ap
pl
i
e
d si
n
g
l
e
l
e
vel
di
scret
e
wavel
e
t
t
r
ans
f
o
r
m
on red c
o
m
pone
nt
of
host
bi
om
et
ri
c im
age and g
e
t
diffe
re
nt wa
vel
e
t coefficients
like LL,
HL
, L
H
a
n
d HH.
Gene
rat
e
t
w
o
pn se
q
u
ence
s i
s
gene
rat
e
d
usi
ng
fi
xe
d n
o
i
s
e po
we
r w
h
ere
o
n
e p
n
seq
u
e
n
c
e
i
s
for
bi
t
0 and
anot
her
p
n
se
q
u
ence
i
s
f
o
r
bi
t
1.
Gene
rat
e
wat
e
r
m
arked
bi
om
etri
c i
m
age usi
n
g C
o
x
al
g
o
ri
t
h
m
[22]
eq
uat
i
o
n a
n
d
w
h
i
c
h
i
s
gi
ve
n
bel
o
w.
If e
n
c
ode
d
spa
r
se m
easurem
ent
s
i
s
bi
t
0 t
h
en
1
_
_
1
)
(
1
Sequence
PN
N
R
HL
RW
R
HL
(4
)
2
_
_
1
)
(
1
Sequence
PN
N
R
LH
RW
R
LH
(5
)
Whe
r
e
HL
1R
(RW)
and
LH
1R
(RW
)
i
s
m
odifi
ed H
L
an
d
LH S
u
b
b
a
nd
o
f
re
d com
p
o
n
e
n
t
of
h
o
st
bi
o
m
et
ri
c
im
age,
HL
1R
an
d
LH
1R
is ori
g
in
al
HL
and
LH
Su
b
b
an
d
o
f
re
d c
o
m
pone
nt
o
f
h
o
st
bi
o
m
et
ri
c im
age,
N
is g
a
i
n
factor,
PN
_
S
e
que
nce
_
1
i
s
p
n
se
que
nce
1 f
o
r
bi
t
0
of e
n
c
ode
d s
p
ar
se m
easurem
ent
s
,
PN_Se
quence
_2
is pn
sequ
en
ce 2 fo
r
b
it 1
of
en
code
d s
p
ars
e
m
easurem
ents.
App
lied
i
n
v
e
rse sing
le lev
e
l
discrete wav
e
let
tran
sfo
r
m
to
get waterm
arke
d biom
etric
image.
Fig
u
re 2
.
W
a
term
ark
Ex
tractio
n & Recon
s
tru
c
tio
n Pr
o
c
edur
e
3.
3.
W
a
term
a
r
k E
x
trac
ti
o
n
& Rec
o
ns
truc
ti
on
Proce
dur
e
The wat
e
rm
ark ext
r
act
i
on &
reco
nst
r
uct
i
o
n
st
eps are
desc
ri
be
d bel
o
w
(a
do
pt
ed
fr
om
7, 16
, 1
7
an
d
18
):
Take
waterm
arked bi
om
etric
im
age and c
o
m
pute size of
M
N. Take
r
e
d
c
o
m
pone
nt
of wat
e
rm
arke
d
biom
etric im
ag
e for e
x
traction of e
n
code
d s
p
arse m
easurements.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Secu
rity of Bi
o
m
etric D
a
ta
U
s
ing
C
o
mpr
e
ss
ed W
a
term
arki
ng
Tec
hni
que
(
R
oh
it Than
ki)
76
2
Ap
pl
i
e
d
si
n
g
l
e
l
e
vel
di
scret
e
wa
vel
e
t
t
r
an
s
f
o
r
m
on
wat
e
rm
arked
bi
om
et
ri
c im
age a
n
d
get
di
f
f
ere
n
t
wav
e
let co
efficien
ts lik
e LL,
HL, LH and
HH.
In
itialize d
ecod
e
d sp
arse m
e
a
s
u
r
em
en
ts to
al
l o
n
e
s.
Decode
d_spa
r
s
e
_m
easurem
ent = ones
(1, M
1), whe
r
e M
= size of e
n
coded s
p
arse m
eas
urem
ents
Fi
nd
co
rrel
a
t
i
o
n i
n
H
L
a
n
d
L
H
c
o
m
pone
nt
s
of
wat
e
rm
arke
d
bi
om
et
ri
c im
age.
1.
);
1
_
_
,
1
(
2
()
_
Sequence
PN
R
HL
corr
Horizontal
n
Correlatio
2.
);
2
_
_
,
1
(
2
()
_
Sequence
PN
R
LH
corr
Vertical
n
Correlatio
3.
;
2
/
())
_
()
_
(
)
(
Vertical
n
Correlatio
Horizontal
Corrlation
d
Watermarke
n
Correlatio
Com
p
are the correlation
with m
ean
correlati
on for setting value of
decoded sparse m
easurem
ents.
1.
If(correlation(bit) >
m
ean (correlation))
Decode
d_spa
r
s
e
_m
easurem
ent (bit) =
0;
After
getting
decoded val
u
e of spar
se m
e
asurem
ents and com
p
are thi
s
decoded sparse m
easurem
e
n
ts
with
e
n
co
de
d spars
e
m
easur
em
ents
usin
g SSIM
[7]
fo
r decisio
n
a
b
o
u
t
rec
onst
r
uctio
n of waterm
ark
biom
etric im
ag
e.
It com
p
arison result is greater than m
a
t
c
hing
score value and t
h
en get actual value of s
p
ars
e
m
easurem
ents fr
om
decode
d
values
usi
n
g
u
n
if
orm
qua
ntization
whic
h is
use
d
at em
bed
d
er
side.
After getting d
ecodi
ng value of
s
p
ar
se
m
easurem
ents,
th
en
applied cs
rec
ove
ry
alg
o
rith
m
on this spa
r
se
m
easurem
ents using c
o
rrect
m
easurem
ent m
a
trix
A
w
h
ic
h is
ge
nerated
at em
bedd
er si
de.
After a
pplication
of cs rec
o
ve
ry
algor
it
hm
, e
x
tracted s
p
arse
coefficie
n
ts
x’
of wate
rm
ark biom
etric im
age
is get at
detector
side
.
Applied invers
e singula
r
value deco
m
position (S
VD) on extracted spa
r
se
coefficients, original U and V
matrix value to get
reconstruc
ted wate
rm
ark
biom
etric im
ag
e.
4.
E
X
PERI
MEN
T
AL RES
U
L
T
S
This is
descri
bed
res
u
lts o
f
pr
o
p
o
s
ed
wa
term
arking tec
hni
que
. F
o
r
p
e
rf
orm
a
nce o
f
pr
o
p
o
s
ed
waterm
arking technique
eval
uated usi
n
g
color face
im
ag
e taken from
India
n
face
database
[23] a
n
d
gray
scale iris i
m
age from
CASIA iris data
base
[
24]
whic
h is s
h
o
w
n in
fig
u
re
3. T
h
e size
of
waterm
ark im
age is
12
8
1
2
8
pi
xe
ls an
d
host c
o
l
o
r
im
age is 25
6
25
6 pixels.
(a)
(b
)
Figu
re 3.
(a
) O
r
iginal H
o
st
Fa
ce
Im
ag
e (b)
Original
Waterm
ark Iris Im
age
For generation of
sparse
m
eas
urem
en
ts of iri
s
im
age, singul
ar
va
lue decomposition (SVD)
is
applied
on iris im
age and
get sin
gula
r
(S) m
a
trix value as spa
r
se c
o
efficients
x
with
size o
f
16
384
1.
Meas
ure
m
ent
matrix
A
with size
of 51
2
16
3
84 is
ge
ner
a
ted usi
ng
ra
n
dom
seed. T
h
e
n
ge
ne
rate spa
r
se m
easurem
ents o
f
iris im
age usin
g eq
uatio
n
1 w
ith size of
5
12
1
usin
g
y
512
1
=
A
512
16384
x
16384
1
. These s
p
arse m
easure
m
ents
of
iris im
age are e
n
c
o
d
e
d
usi
n
g
u
n
if
or
m
quantization
with 2 bit /
level.
The
b
its of
enc
o
de
d s
p
ar
se
m
easurem
ents of iris
im
age are 1024
(512
2) which is used as waterm
ark in
f
o
rm
ation is
sho
w
n in
fi
gu
r
e
4.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN:
2
088
-87
08
I
J
ECE Vo
l. 4
,
No. 5
,
Octob
e
r
20
14
:
758
–
7
66
7
63
Figure
4. Enc
o
ded Sparse
Me
asurem
ents of
Iris
Im
age
This enc
o
ded s
p
arse m
easure
m
ents as waterm
ark in
f
o
rm
ation is em
beddin
g
into
HL an
d
HL wa
velet
coefficients of
red c
o
m
pone
nt
of
host
face im
age to ge
t wa
term
arked color face im
age. Daubec
hies wa
velet is
use
d
to
dec
o
m
pos
e H
L
a
n
d
L
H
c
o
ef
ficients
of
re
d c
o
m
pon
ent of
host fac
e
im
age. T
h
e
gain fact
or
val
u
e is set
to 2. T
h
e wate
rm
arked col
o
r
im
age is show
n in fi
gu
re
5 (a
) and extracte
d
sparse m
easure
m
ents of iris im
age
are s
h
o
w
n in
fi
gu
re
5
(b
).
(a)
(b
)
Figure
5. (a
)
Waterm
arked
Face Im
age (b) Ex
tracte
d
Spa
r
se Meas
urem
ents
of
Iris
Im
age
SSIM
[2
6]
is
use
d
f
o
r
fi
nd
s
i
m
i
larity
between e
x
tracted
spars
e
m
easure
m
ents and encode
d s
p
a
r
se
m
easurem
ents of iris im
age for
decision about m
odifi
cation
of
data an
d
reco
nstr
uctio
n
of iris im
age fr
om
spars
e
m
easure
m
ents base
d
o
n
m
a
tching sc
o
r
e. T
h
e
value
of m
a
tching score is set
0.9.
If the sim
ilari
ty
score
is greate
r
tha
n
m
a
tching sc
or
e
then
bi
om
etric data is a
u
the
n
t
i
cate and i
r
is i
m
age rec
onst
r
ucted
fr
om
its spars
e
m
easurem
ents. For
decoding sparse m
easure
m
ents of iris im
age from
extr
acted spa
r
se
m
easurem
ents
use
d
uni
fo
rm
quantization w
h
ich i
s
used at
em
b
e
dde
r side and get actual value of spa
r
se m
easurem
ents
of iri
s
im
age. This actual sparse m
easurem
ents value is used
for reconstruction
of waterm
ark iris i
m
age. If the
sim
ilarity
scor
e is less tha
n
m
a
tching
sc
ore t
h
en
bi
om
etric
data is m
odifie
d
by
im
poster and
un
a
u
the
n
ticated.
The cs rec
o
v
e
ry
algo
rithm
like orth
o
g
o
n
al m
a
tching
pu
rs
uit (OM
P
) algo
rithm
[27]
u
s
ed
fo
r
reconstruction of
waterm
ark iris i
m
age from sparse
m
e
a
s
urem
ents. The input data for
OMP algorith
m
is
m
easurem
ent m
a
trix
A
with size
of 512
16
384
, sp
ar
se measu
r
em
ents with size of 512
1 and sparsity level
12
8 a
n
d
out
pu
t of
OM
P
alg
o
r
ithm
is extracted
s
p
arse
coe
fficients
with
size of 16384
1. A
pplied
inve
r
s
e
SV
D on origi
n
al U,
V m
a
trix and ex
tract
ed spa
r
se coe
f
ficients to get
reconstructed waterm
ark
iris
i
m
age
whic
h is s
h
ow
n in
fi
gu
re
6.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN:
208
8-8
7
0
8
Secu
rity of Bi
o
m
etric D
a
ta
U
s
ing
C
o
mpr
e
ss
ed W
a
term
arki
ng
Tec
hni
que
(
R
oh
it Than
ki)
76
4
Figu
re
6.
R
eco
nstr
ucted
Wate
rm
ark Iris
Im
age
usin
g C
S
re
cove
ry
pr
ocess
PSNR,
NCC an
d
BCR ar
e
u
s
ed
fo
r
p
e
rfo
r
m
an
ce m
easu
r
emen
t o
f
pro
posed
techn
i
qu
e.
Th
e PSNR is
used percept
u
al quality
m
eas
ure
betw
een
original color imag
e and wat
e
rm
arked color im
age. The
NCC is
used to
find correlation bet
w
een
orig
i
n
al c
o
lor im
age and
waterm
arked c
o
lor im
age. The NCC
value i
s
nea
r
to
1
is ind
i
cated
th
at
watermar
k
i
ng
techn
i
q
u
e
is m
o
r
e
rob
u
s
t
.
BCR is u
s
ed
to
f
i
n
d
b
i
t co
r
r
ect
r
a
te between
ori
g
inal waterm
ark and e
x
tracted wate
rm
ark. In this paper, PSNR,
NC
C is
calculated betwee
n origi
n
al face
i
m
ag
e an
d
wat
e
r
m
ar
k
e
d
im
ag
e an
d
BCR is calcu
lated
b
e
t
w
een
en
cod
e
d
sp
ar
se m
easu
r
e
m
en
ts an
d
extr
acted
spars
e
m
easure
m
ents value
.
Th
is pr
opo
sed water
m
ar
k
i
n
g
tech
n
i
qu
e is also
tested for comm
on waterm
arking
attacks like
com
p
ression, addition
of different
no
ise, and
geom
etric attacks like crop
ping.
The Table 1
summ
arize
d
the
PSNR,
NCC value betwee
n original co
l
o
r host face im
age
and
waterm
arke
d color host face im
age and SSIM
,
BCR v
a
lu
e between
en
coded
spar
se m
easu
r
em
en
ts and
ex
t
r
acted
sp
ar
se m
easu
r
emen
ts. I
n
wat
e
r
m
ar
k
e
m
beddi
ng
procedure, the two PN sequences are
m
u
ltip
lied with a gain factor
and e
m
bedded in
wavelet
coefficients of
color biom
etric
im
ag
e. Table 2 s
h
ows t
h
at val
u
e
of t
h
e
gain factor does
not affect on
waterm
arked i
m
age an
d e
x
tra
c
ted wate
rm
ark in
f
o
rm
ation.
Table 1. Quality
Measures
of
Proposed Waterm
arking Technique
Results
No Attac
k
JPEG A
ttac
k
Gaussian Noise
Atta
c
k
Salt & P
e
pper
Noise Attac
k
Spec
k
le Attac
k
Cropping
Atta
c
k
Q = 9
0
µ=0,
=0.
001
Densit
y = 0.
005
Variance = 0.
004
NCC
0.
96
0.
98
0.
95
0.
95
0.
95
0.
95
PSNR (dB)
38.
17
38.
82
37.
39
37.
17
37.
41
37.
74
SSIM
0.
99
0.
99
0.
99
0.
99
0.
99
0.
99
BCR
1.
00
1.
00
1.
00
1.
00
1.
00
1.
00
Table
2. E
f
fect of Gain
Factor
on
Pr
opo
sed
Water
m
ar
k
i
ng
Techn
i
qu
e
Gain Fac
t
or
PSNR
(dB)
NCC
SSIM
BCR
1
39.
24
0.
98
0.
99
1.
00
2
38.
17
0.
96
0.
99
1.
00
3
37.
40
0.
94
0.
99
1.
00
4
36.
68
0.
91
0.
99
1.
00
5
36.
19
0.
89
0.
99
1.
00
6
35.
78
0.
86
0.
99
1.
00
7
35.
41
0.
84
0.
99
1.
00
8
35.
14
0.
82
0.
99
1.
00
9
34.
93
0.
81
0.
99
1.
00
10
34.
73
0.
79
0.
99
1.
00
The
waterm
arking technique i
d
ea is
borrowe
d
from
[7 a
n
d
25]
,
the
r
e a
r
e s
i
gnifica
nt m
o
d
i
fication is
taken place in propose
d
technique com
p
are
d
with Inam
dar [7] and Hajja
r
a
[25]. T
h
e com
p
aris
on
of propos
ed
techniq
u
e with
techniq
u
e in [7
, 25]
with d
i
ffere
nt pa
ram
e
ters are sum
m
a
rized in table 3. In the propos
e
d
wo
rk
discu
sse
d he
re, sin
g
le
level ho
rizo
nta
l
and ve
rtical details of re
d
com
pone
nt are
used
fo
r em
beddi
ng;
while in
[7]
se
con
d
lev
e
l dec
o
m
position is
obtaine
d
fr
om
app
r
oxim
a
tion ban
d
,
w
h
ile in [
2
5]
it is obt
ained
fr
om
horizo
n
ta
l details fo
r em
bed
d
in
g.
I
n
the
pr
o
pos
ed
tec
h
nique two PN
sequence a
r
e e
m
bedded at a t
i
m
e
in
single level wa
velet coefficients according to enc
ode
d s
p
arse
m
easurem
e
n
ts of
waterm
a
r
k whic
h is generat
e
d
usin
g C
S
theory
an
d u
n
ifo
r
m
quan
tization. In the proposed techni
que
, waterm
ark biom
etric data is
com
p
ressed b
e
fo
re
em
beddi
ng
i
n
to host m
e
dium
.
In
t
echni
que
[7], large
r
size of
waterm
ark im
age is
degraded
waterm
arked im
age
quality wh
ich
is li
mitat
i
on of this techni
que
.
Where in t
h
e
proposed technique,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN:
2
088
-87
08
I
J
ECE Vo
l. 4
,
No. 5
,
Octob
e
r
20
14
:
758
–
7
66
7
65
larger
size
of
waterm
ark im
age is c
o
m
p
re
ssed
usi
n
g
cs
theo
ry
f
r
am
ewor
k a
n
d em
beds i
n
to
h
o
st m
e
dium
without degrading waterm
arked
im
age
quality.
Table
3. C
o
m
p
ariso
n
of
Pr
o
p
o
se
d tech
niq
u
e
with
I
n
am
dar techni
que
[
7
]
a
n
d
Ha
jjara
tech
niq
u
e
[2
5]
Para
m
e
ters
Proposed
Techni
que
Ina
m
dar
Tec
hniq
u
e [7]
Hajjara Tech
niq
u
e [25]
Wavelet
Deco
m
p
osition
First level
Horizontal & Vertical
Details of Red com
ponent
Second level of Appr
oxim
a
tion
Details
Second level of Hor
i
zontal
Details
PSNR Range
34 to 40 dB
30 to 38 dB
3 to 5 dB
No. of P
N
Se
quen
ce
T
w
o PN
Sequences ar
e e
m
bedded at
a
ti
m
e
in Horizontal and
Vertical
Details bands
Three P
N
Sequenc
es are
em
bedded at a ti
m
e
in all
details bands
Only One
PN S
e
quence is
em
bedded in either in bands
Water
m
a
r
k
E
n
coded Spar
se Measur
em
ents of I
r
is
Im
a
g
e
Signatur
e
Im
age
L
ogo
Co
m
p
u
t
ational
Security Achieve
d
Due to Co
m
p
r
e
ssive Sensing T
h
eor
y
plus Secr
et Key
Secret Ke
y
Secret Ke
y
Co
m
p
ression of
Wa
ter
m
a
r
k
Da
ta
Due to Co
m
p
r
e
ssive Sensing T
h
eor
y
No such scope
No such scope
Authen
tica
tio
n
through Te
m
p
lat
e
Matching
Featur
e of Reconstr
ucted ir
is i
m
age
extracted and
m
a
tc
hed
Featur
e of Recovered
Signatures are ext
r
acted and
ma
t
c
h
e
d
No such scope
5.
CO
NCL
USI
O
N
This pa
pe
r p
r
o
pos
ed a
ne
w b
i
om
etric water
m
arkin
g
tech
nique
usi
ng
wa
v
e
lets and C
S
t
h
eo
ry
. T
h
e
pr
o
pose
d
tec
h
niq
u
e c
o
m
b
ines the
field
of
biom
etric
waterm
arking a
n
d
com
p
ressive s
e
nsin
g.
The
p
r
op
ose
d
techniq
u
e is r
o
b
u
st agai
nst Gaus
sian, S
p
e
c
kle an
d Salt & pe
ppe
r n
o
is
e, C
r
o
p
p
in
g and J
P
EG c
o
m
p
ressi
o
n
.
This technique is not robust
against hist
ogram
equaliza
tion and
filter attacks. This
t
echni
que is
used to
biom
etric data pr
otection
o
v
e
r
com
m
unication c
h
an
nel
betw
een tw
o m
o
d
u
les o
f
biom
etric sy
stem
.
REFERE
NC
ES
[1]
A Jain and
A K
u
mar. “Biometric
Recognition:
An Overview”.
Second Gen
e
ration Biometri
cs:
The Ethical, Leg
a
l
and Socia
l Context,
E. Mordin
i a
nd D. Tzovaras (
E
ds.)
, Springer
.
2012: 49 –
79.
[2]
A Jain, K Nandakumar and A Nagar.
“
B
iom
e
tr
ic Tem
p
l
a
te Se
c
u
rit
y
”
.
EURASI
P
Journal on Advances in Sign
al
Processing, Sp
ecial Issue on
Ad
vanced
Signal Pro
cessing and Pattern Recogn
iti
on Methods for
Biometrics
. 200
8: 1
– 17.
[3]
N Ratha, J Connell and R Bolle. “Enha
ncing Security
and Privacy
in Biom
etric Based Authen
tication S
y
stems”.
IBM Systems Jo
urnal
. 2001; 40(
3): 614 –
634.
[4]
A Jain and U Uludag. “H
iding Biometric Data”.
I
EEE Transactio
ns on Pattern A
nalysis and Mac
h
ine Inte
llig
enc
e
.
2009; 25(11): 14
94 – 1498
.
[5]
A Jain, A Ross
and S Prabhakar. “
A
n Introduction to Bi
om
etric
Recognit
i
on”
. IEEE Tr
ansact
io
ns on Circuits and
S
y
stems for Vid
e
o Technolog
y
,
Specia
l
Issue on
Image and
Video Based
Biometrics
. 2004
; 14(1)
: 4 –
20.
[6]
A Jain, A Ross
and S Pankanti.
“
B
iom
e
tric
s: A Tool for Information Secur
i
ty
”.
IEEE T
r
ansactio
ns on Information
Forensics and S
ecurity
. 2006; 1(
2): 125 –
143.
[7]
V Inamdar, P
Rege and
M Ar
y
a
. “Offline Handwritten
Sign
ature b
a
sed Blind Biometric
Watermarking and
Authentication Techn
i
que
usin
g
Biorthogonal Wavelet
Transform”.
International Journal of Computer
Applica
tions
. 20
10; 11(1): 19 –
2
7
.
[8]
G Langelaar, I
Sety
awan and R
Lagnedi
jk. “
W
a
term
arking of
Digital Im
age
and Video Data – A State of Ar
t
Review”.
I
EEE Signal Processin
g
Magazine
. 200
0: 20 –
46.
[9]
N Chaudhar
y
, D
Singh and
D Hussain. “Enhan
c
i
ng Securi
t
y
of
Multim
odal Bio
m
etric Auth
enti
cat
ion S
y
st
em
b
y
Im
plem
enting W
a
term
arking Utili
zing DW
T and DCT”.
IOSR Journal of Comp
uter Engineerin
g
. 2013; 15(1): 6 –
11.
[10]
C Li, B Ma,
Y Wang and
Z Zh
ang.
“Sparse Reconstru
c
tion Based
Wate
rmarking for
Secure B
i
ometr
i
c
Authenti
cat
ion”
.
Biome
t
ric
Re
co
gnition
. 2011; 7
908: 244 –
251.
[11]
S E
d
wa
rd,
S S
u
ma
nthi a
nd R Ranihemamalini. “
Person Authenti
cation Usi
ng Multimodal Biometri
cs with
Watermarking
”.
Proceedings of
2011 Internat
ion
a
l Conferen
ce
o
n
Signal Processing, Com
m
unication, Com
puting
and Networking
Techno
logi
es
(I
CSCCN). 2011:
100 – 104
.
[12]
M Qi, Y Lu, N Du, Y Zhang,
C Wang and J
Kong. “A
Novel Image Hiding Approach Based
on Correlation
Anal
y
s
is for
Sec
u
re Multim
odal
Biom
etrics”
.
Jo
urnal of Networ
k and Computer
Applications
. 2
010; 33(3): 247
–
257.
[13]
M Vasta, R Singh and A Noore, M Houc
k and
K Morris. “Rob
ust Biometric
I
m
age Watermar
king for Fingerp
rint
and Face
Template Protection
”
.
I
E
ICE Ele
c
troni
c
s
Express
. 2006
; 3(2): 23
– 28
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN:
208
8-8
7
0
8
Secu
rity of Bi
o
m
etric D
a
ta
U
s
ing
C
o
mpr
e
ss
ed W
a
term
arki
ng
Tec
hni
que
(
R
oh
it Than
ki)
76
6
[14]
D Moon, K Tachae, S Jung, Y
Chung, K Moon, D Ahn a
nd S
Kim. “Performa
n
ce Ev
alu
a
tion
of Watermarkin
g
Techn
i
ques
for
S
ecure M
u
lt
im
odal Biom
etr
i
c S
y
s
t
em
s
”
.
In Co
mputational In
t
e
llig
ence and S
ecurit
y
, Spring
er
Berlin
Heidelber
g
. 2005: 635 –
6
42.
[15]
F Tiesheng,
L Guiqiang
, D Chuny
i
and W Danhua. “A Di
gital Im
age Watermark
i
ng Method Based on the Theo
r
y
of Compressed
Sensing”.
In
tern
ational Journal Automation
and Control
Engin
e
ering
. 2013; 2(2): 56 –
61.
[16]
P Redd
y
,
V Pras
ad and D Rao. “
R
obust Digital
Wa
termarking o
f
Color Images under Noise Attacks”.
Internat
ion
a
l
Journal of Recent Tr
ends in
Eng
i
neering
. 2009;
1(1): 334 –
338.
[17]
Chris Shoemaker. “Hidden B
its:
A Survey
of
Techniques for Dig
i
tal
W
a
term
arkin
g
”.
Ind
e
pend
ent
Study EER-290
,
Prof Rudko, Spring. 2002
.
[18]
G Langelaar, I
Sety
awan and R
Lagnedi
jk. “
W
a
term
arking of
Digital Im
age
and Video Data – A State of Ar
t
Review”.
I
EEE Signal Processin
g
Magazine
. 200
0: 20-46.
[19]
D Donoho. “Compressed Sensing”.
I
E
E
E
T
r
ans
.
Infor
m
. T
h
eor
y
.
2006; 52(4): 128
9-1306.
[20]
E Cand
ès. “Compressive Sampling”. Proc
eed
ing
s
of the I
n
tern
ational Congr
ess
of Mathematician
s
, Madrid, Spain
.
2006.
[21]
E Cand
ès and
J
Romberg. “L1-
Magic: Recov
e
r
y
of
Sparse Sign
als via Conv
ex P
r
ogramming”. 2
005.
[22]
I Cox, J Kili
an, T Sham
oon an
d F Leight
on. “
S
ecure
Spread
Spectrum
Water
m
arking for Multim
edia”.
IE
EE
Transactions on
Image Processing
. 1997; 6(12): 1
673 – 1687
.
[23]
V Jain, A Mukherjee. “The In
d
i
an Face Database”. h
ttp://vis-www.cs.umass.edu/~v
idit/Ind
iaFaceDatabase. 2002
.
[24]
For Iris Databas
e
: h
ttp://www.sinobi
ometrics.co
m
/caisair
is.html
[25]
S Hajjara, M Abdallah and A
Hudaib.
“Digital Image Wa
termarking Using Localized
B
i
ortho
gonal Wavelets
”.
European Journ
a
l of
Sc
ie
nt
if
ic
Re
se
arc
h
. 2009; 2
6
(4): 594 –
608.
[26]
Z W
a
ng, A Bovik, H Sheikh and E Sim
oncelli
. “
I
m
a
ge Qualit
y Assessm
ent from
Error Visibilit
y to Structur
a
l
Sim
ilarit
y
”
,
IEEE Transaction o
n
Image
Processing
, vo
l. 13, no.
4, April 2004.
[27]
J Tropp and A
Gilbert, “Signal Recover
y
from Random
Measur
ements via Ort
hogonal Match
i
n
g
Pursuit”.
IEEE
Transactions on
Information Theory
. 2007; 53(12
): 4655 –
4666.
BIOGRAP
HI
ES OF
AUTH
ORS
M
r
.
Rohit M Thanki
has
rece
i
v
ed B.
E.
degre
e
in El
ec
tronics
and Communication Engin
eering
from Saurashtra University
, Rajk
ot and M.E. in
Communication
Engineering fro
m Sardar Patel
Univers
i
t
y
, Val
l
abh Vid
y
anag
a
r
. He is
curre
ntly
pursuing
his Ph.D. in Electronics and
Communication
Engineering fro
m C U Shah Universi
ty
, Wa
dhwan Ci
ty
, Guj
a
ra
t,
India.
His
are
a
of research is to Design Watermarking Algorithm
for Biometric Data Prot
ection
.
He has guided
more than 15 UG students in their project work.
He has published 11 research p
a
pers in various
high impact factor intern
ational journals. He has
presented
7
research p
a
pers in
v
a
rious national
and intern
ation
a
l confer
ences.
He has publishe
d books titled C
o
m
p
arative
anal
y
s
is of digit
a
l
watermarking techniques and Desi
gn of Operatio
nal Transcondu
ctance Amplifier with Lambert
Publishing House, German
y
.
H
i
s area of
inte
r
e
st is Digita
l
W
a
term
arking,
Im
age & Signal
Proc
e
ssing,
Compre
ssive
Se
nsing
,
Pattern
Recogn
ition,
and Digital VLSI Design.
Dr. Komal R Borisagar
receiv
e
d B.E. degr
ee in Electroni
cs an
d Communicatio
n from C. U.
Shah Engineering Colleg
e
, Saurashtra Univer
si
ty
, Rajkot, Gujarat, Ind
i
a in
2002 and M.E.
degree in Com
m
unication S
y
st
em
Engineering
from
Changa Institute of Te
ch
nolog
y, Gujara
t
University
, and
Ahmedabad in
2008. In
2012
,
she r
e
ceived
her doctoral degree from
the
Department of
Electronics
and
C
o
mmunication Engineer
ing, JJT
Un
ive
r
sity
, Ra
jastha
n.
She
ha
s
teaching
experience of ov
er 10
y
e
ars. She is wo
rking a
s
Assista
n
t Profe
ssor a
t
Elec
tronic
s
&
Communication Department, Atmiy
a
Institu
te of
Technolog
y
an
d Sc
ience, Rajk
ot. Her areas of
inter
e
s
t
ar
e wir
e
l
e
s
s
com
m
unicati
on, s
p
ee
ch
pro
c
essing and signal
& image processing.
Evaluation Warning : The document was created with Spire.PDF for Python.