TELKOM
NIKA
, Vol.12, No
.4, Dece
mbe
r
2014, pp. 88
3~8
8
9
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i4.359
883
Re
cei
v
ed Au
gust 20 2
014;
Revi
se
d Oct
ober 2, 20
14;
Accept
ed O
c
tober 25, 20
1
4
Multi Facial Blurring using Improved Hénon Map
Saparudin*
1
,
Ghazali Sulong
2
, Muha
mmed Ahme
d Saleh
2
1
F
a
cult
y
of Co
mputer Scie
nc
e, Sri
w
i
j
a
y
a
Un
iver
sit
y
, In
dral
a
y
a, Sum
a
tera
Selata
n, Indon
esia.
2
Digital Me
dia
and Games C
e
ntre of Exce
lle
nce (MaGIC-X), F
a
cult
y
of Co
mputin
g,
Univers
i
ti T
e
knolo
g
i Mal
a
ysia,
Skudai, Joh
o
r,
8131
0, Mala
ys
ia.
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: Saparu
d
i
n12
04@
ya
ho
o.co
m
1
,
ghaz
ali
@
utms
pace.e
du.m
y
2
,
hasi
bat2
003
@
y
a
h
o
o
.com
3
A
b
st
r
a
ct
Genera
lly, full
encrypti
on is a
ppli
ed o
n
the e
n
tire
i
m
a
ge to
obscur
e
the fa
ces. How
e
ver, it suffers
in
overh
e
a
d
, s
pee
d a
n
d
ti
me.
Altern
atively,
selectiv
e
e
n
cry
p
tion
can
b
e
u
s
ed to
encry
pt
only
the s
ens
iti
v
e
part of the
i
m
a
ge suc
h
as
hu
ma
n faces. T
h
i
s
pap
er
pr
op
os
es a
new
encr
y
ption
al
gorith
m
usin
g e
n
h
a
n
c
e
d
Hen
on c
haotic
map to
conc
e
a
l the
faces. T
h
is tech
ni
que
i
n
volv
es thre
e
steps: face
det
ection,
encrypti
o
n
and d
e
crypti
on
. Experiments
have b
e
e
n
perf
o
rmed to eva
l
u
a
te security su
ch as histogr
a
m
, sensitiv
ity and
statistical an
al
ysis, and resu
l
t
s reveal
that the pro
pose
d
meth
od pr
ovid
es
hig
h
securit
y
w
i
th entropy and
correlati
on cl
os
e to idea
l val
u
e
s
.
Ke
y
w
ords
: s
elective im
age encryption, fa
ce detectio
n
, face blurri
ng, hé
no
n
map, ch
aotic s
ystem
1. Introduc
tion
The p
r
ote
c
tio
n
of private
or pe
rsonal
data
is e
s
se
ntial to provi
de an effici
e
n
t work
environ
ment.
This is be
cau
s
e net
wo
rks have be
come e
s
sen
t
ial to growi
ng multimed
ia
appli
c
ation
s
and the
r
e i
s
a nee
d for d
a
ta se
cu
rity tech
niqu
es to
safeg
uard valuabl
e data
from
unauth
o
ri
zed
acce
ss. T
h
e
confid
entialit
y of data is
provide
d
by encrypti
on,
which ren
d
e
r
s the
data un
re
ada
ble by u
nauth
o
rized
perso
n
s
. Encryption
ca
n be
ap
plied to diffe
ren
t
types of d
a
ta.
Ho
wever, e
a
c
h type h
a
s
its own inh
e
rent ch
ara
c
te
ristics. For
example, ima
g
e
s
rep
r
e
s
ent
a
signifi
cant type of data be
cause
of their
wide
scale u
s
e. Image
s
ha
ve
th
e
i
r
ow
n c
h
ar
ac
te
r
i
s
t
ics
as they co
ntain a larg
e amount of d
a
ta and
nei
g
hbou
ring pix
e
ls have
strong correlati
ons
meanin
g
that
the value
of
each pixel
ca
n be
re
as
o
n
a
b
ly
pre
d
icte
d from
its neig
h
bouri
ng pixels,
thus the re
qui
reme
nt for different en
crypt
i
on met
hod
s. Curre
n
t encryption algo
rith
ms en
crypt th
e
entire i
m
age
inclu
d
ing th
e
backg
rou
nd,
whi
c
h m
a
y b
e
con
s
ide
r
ed
irrelevant o
r
less im
po
rta
n
t.
Also, the algo
rithm req
u
ires extra overhe
ad [1].
Selective
en
cryption
i
s
o
ne of
the
ne
w
ways of
secu
ring
cont
ent from
u
n
a
u
thori
z
ed
use
r
s.
In
sel
e
ctive e
n
cryp
tion, the m
a
i
n
go
al i
s
to
redu
ce th
e a
m
ount of
dat
a to b
e
e
n
cry
p
ted
while obtai
nin
g
the requi
re
d level of security. Se
lective encryption
has t
he ad
ditional feature of
pre
s
e
r
ving
so
me of th
e fun
c
tionalitie
s
of
the ori
g
inal
bi
t stre
am. A
common
ap
proach i
s
to
divide
the co
ntent i
n
two p
ublic
and p
r
ivate i
n
sel
e
ctiv
e e
n
cryptio
n
pa
rts, prote
c
ted
part is
mad
e
as
small as possible [2].
This pap
er p
r
esents
a ne
w sele
ctive
e
n
cr
yptio
n
te
chniqu
e ba
se
d
on im
prove
d
Hé
non
cha
o
tic m
ap t
o
ob
scure facial image
s. T
he p
r
o
c
e
ss i
n
volved thre
e
parts nam
ely
face d
e
tectio
n,
encryption a
nd decryptio
n. The re
st o
f
the paper is org
ani
zed i
n
to se
ctions.
Section 2 is the
related
wo
rk
of the study
whi
c
h is foll
o
w
ed
by me
th
odolo
g
y in se
ction 3, while
the re
sults
of the
experim
ents
are expl
aine
d in detail i
n
se
cti
on 4. T
he summa
ry and
con
c
lu
d
i
ng re
marks
are
given in se
cti
on 5
.
2. Related Work
Hén
on
cha
o
tic ma
p [3] ha
s some
ch
aracteri
st
ics,
such
as ergod
icity, rand
om
ness a
n
d
the se
nsitivity of initial co
n
d
it
ions
and
control p
a
ram
e
ters, th
ese
maketh
equ
ali
t
y is improve
d
and
the defect of data red
und
a
n
cy is red
u
ce
d also. In 20
09, [4] prese
n
ted a new i
m
age en
crypt
i
on
algorith
m
ba
sed on
Héno
n’
s ch
aotic
syst
em in or
der t
o
meet the re
quire
ment
s of secure imag
e
transfe
r. Shu
ffling position
s
and
cha
ngi
ng the grey
values of im
age pixel
s
were
combi
n
e
d
to
cha
nge the
relation
ship b
e
twee
n the cipher-ima
ge
and the o
r
igi
nal-im
age. Fi
rst, Arnol
d’s
cat
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 883
– 889
884
map
wa
s u
s
ed to
ch
ange
the p
o
sitio
n
s
of th
e ima
ge pixel
s
. Se
con
d
, the
sh
uffled-ima
ge
wa
s
encrypted pix
e
l by pixelusi
ngHéno
n’s chaotic
syste
m
. The auth
o
rs
im
prove
d
the equation
s
of
Hén
on m
ap b
y
mergin
g the
two e
quatio
n
s
into
one
eq
uation. Thi
s
i
m
prove
d
eq
u
a
tion works li
ke
the origin
al two Héno
n map equ
ation
s
.
Four yea
r
s la
ter, [5] prese
n
ted a novel
image en
cryp
tion techni
qu
e based on a
Héno
n
cha
o
tic map.
Chao
s-ba
se
d image en
cryption
te
ch
nique
s are one of the more p
r
omi
s
in
g
encryption al
gorithm
s. Th
ey pr
ovide v
e
ry efficient
and fast im
a
ge en
cryptio
n
due to th
eir
determi
nisti
c
nonlin
ear sy
stems t
hat
exhibit extrem
e
se
nsitivity
to initial conditi
on a
nd
ran
d
o
m
like be
haviou
r
s.
In same yea
r
, [1] prese
n
t
ed three di
stinct image
encryption te
chni
que
s for colou
r
image
s. Two
of the thre
e t
e
ch
niqu
es
use a
sele
ctive encryption algorithm and
t
he
first
tech
ni
que
use
s
Regio
n
Based Sel
e
ct
ive Image Encryption. Th
i
s
algorithm e
n
c
rypt
s part of
the bit-stre
a
m
usin
g a well_
proven
ciph
e
r
ing techniq
u
e
. A waterm
ark m
e
ssag
e
is added du
ring this p
r
o
c
ess
along with
a decryption ke
y
that
t
he receiver u
s
e
s
to
decrypt t
he bi
t-stre
am an
d
decomp
r
e
ss t
he
image. In prin
ciple, there is no differen
c
e
betwe
e
n
the origin
al imag
e and the en
crypted imag
e.
Followi
ng tha
t, [6] introduced two differe
nt methods f
o
r sel
e
ctive i
m
age en
crypt
i
on. The
first method d
i
vides an ima
ge into su
b-blocks a
nd
the
n
sele
cted bl
ocks a
r
e ma
n
ually sent to th
e
encryption
proce
s
s. Th
e
seco
nd
me
tho
d
auto
m
atical
ly detect
s
th
e
po
sition
s of
obje
c
ts
and
t
hen
sele
cted
obje
c
ts
are
sent
to the en
cryp
tion pr
ocess.
Morpholo
g
y
techni
que
s a
r
e e
m
ployed
to
detect th
e p
o
s
ition
s
of
obje
c
ts i
n
th
e ima
ges.
The
s
e
two
app
roa
c
h
e
s
we
re
spe
c
i
f
ically devel
o
ped
to encryptm
e
dical and satellite images.
Finally, [7] p
r
opo
se
d an i
m
age en
cryp
tion techni
qu
e that sele
cts facial a
r
ea
s and
encrypts th
e
m
usi
ng
RGB
pixel re
gro
u
p
i
ng of an i
m
a
ge
of m x n
s
i
z
e
. As
a result, it is
diffic
u
lt for
off-the-shelf softwa
r
e
to
resto
r
e
the encrypt
ed i
m
age. T
h
is
techni
que
is useful for
law
enforcem
ent agen
cie
s
to reco
nstruct a f
a
ce fro
m
pict
ure
s
or vide
o
s
relate
d to abuse ca
se
s
.
3. Proposed
Metho
d
First, featu
r
e
extraction
is
perfo
rmed
u
s
ing Vi
ola Jo
ne
s
with be
st
fe
ature sele
ctio
n
usi
n
g
AdaBoo
st alg
o
rithm th
en
a
pplying
Ca
scaded
cl
assifie
r
to
dete
c
t on
e o
r
mo
re
fro
n
t face
s ra
pid
l
y.
Secon
d
, en
crypting and d
e
cryptin
g
hu
man faces.
The p
r
o
c
ess utilizes
Hé
n
on map
but with
some im
prov
ements.
The propo
se
d method im
proved
Hén
o
n
map
usi
n
g
Sine functio
n
on the x=axis and
Tange
nt funct
i
on on the y-a
x
is whi
c
h also includ
es th
e freque
ncy
w to control the Hé
non ch
aotic
sign
al. Th
ese
pa
ramete
rs
use
d
di
sto
r
te
d si
gnal
s fro
m
un
autho
rized u
s
e
r
s to
d
e
crypt
the
sig
nal
due to spe
c
ific ch
aracte
rist
ics
su
ch a
s
n
onpe
ri
odi
city, ergo
dicity, and ra
ndomi
c
it
y.The prop
osed
method u
s
ed
four initial values, four pa
ramete
rs and
two frequ
en
ci
es as give
n by the following
equatio
ns:
)
sin(
1
2
1
2
i
i
i
wx
b
ax
x
(1)
)
tan(
1
2
1
2
n
n
n
wy
b
ay
y
(2)
whe
r
e:
,...
2
,
1
,
0
,
1
,
0
n
b
a
3.1 Encr
y
p
tion
p
r
oces
s
Input:
dete
c
te
d face
s
Outpu
t
:
en
cry
p
ted face
s
Step1:
gen
erate the se
cret
key by using
equatio
ns 1 a
nd 2.
Step2:
Separate the image c
o
lors
into three co
mp
one
nt matrice
s
Red, Gre
en, an
d Blue.
Step3:
e
n
cry
p
t the
huma
n
face
s u
s
ing
BitXOR op
eration b
a
sed
o
n
the
se
cret
key o
b
taine
d
from step 1 a
nd the col
o
r compon
ent ma
trice
s
a
c
qui
re
d in step 2 a
s
seen in Fig
u
re 1.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Multi Facial B
l
urri
ng u
s
ing I
m
prove
d
He
n
on Map (Sa
p
a
rudi
n)
885
Figure 1. XOR Encryption
After protecti
ng the image
from unauth
o
rized
a
c
cess by encrypting the huma
n
face
s,
the next
step
is to
de
crypt
the
se
cure i
m
age
by
u
s
in
g
de
cryptio
n
algorith
m
whi
c
h wo
rks simi
lar
to the encryption pha
se ex
cept it is don
e in a reverse
manne
r.
4. Experimental Re
sults
and Disc
uss
i
on
This p
a
rt em
ployed two d
i
fferent imag
es. The first image sampl
e
is stan
da
rd
image
databa
se US
C-SIPI (Len
a
)
. The se
con
d
is an in
ternet image fo
r Na
sser foot
ball player
URL
(http://www.al
m
rmaa.
com
)
used to test the propo
sed
method on m
u
lti facial encryption.Hénon
map, imp
r
oved Héno
n m
ap an
d chao
s sy
stem
al
g
o
rithm
s
have
been te
sted
on colo
r im
age
512X51
2 pix
e
ls,The
results of
the
s
e
proce
s
se
s a
r
e
as sh
own
in
Table
3 and
Table
4. He
re
we
can
con
c
lud
e
that the cipher image
s re
sulted fro
m
all the three algorithm
s und
er ch
aotic m
a
p
operation of mode di
splay
a vivid figure of the origina
l
image.
Ho
wever, the
encryptio
n for improve
d
H
énon ma
p gives a bette
r ci
pher. Th
e en
crypte
d
image
s do no
t show any cl
ue of the orig
inal
image be
cau
s
e this m
e
thod are ch
ange
d the value
of pixel throu
gh the eq
uati
ons
(1),
(2)
of prop
os
ed me
thod. As a re
sult, the cip
h
e
r imag
e is m
o
re
cha
o
tic a
nd
random. It
sh
ould b
e
not
e
d
he
re that th
e processin
g
time is ve
ry fast be
ca
use t
h
e
algorith
m
is e
n
crypt j
u
st
on
ly the multi-f
a
ci
al
hum
an
s.Table 1
presents
executio
n time b
e
twe
en
the entire im
age a
nd th
e
are
a
of int
e
re
st in
th
e
image
in th
e en
cryptio
n
and
de
crypt
i
on
respe
c
tively, and the
pro
p
o
se
d metho
d
redu
ce
d the
executio
n time with hi
gh
secu
rity. Secu
rity
analysi
s
such
as hi
stogra
m
, sen
s
itivity, and stati
s
tical pro
pertie
s
were examin
ed by cal
c
ul
a
t
ing
entropy
an
d t
he
co
rrel
a
tion
of two a
d
ja
cent pixel
s
i
n
t
he
ciph
ered i
m
age
a
s
di
scussed
in
the
next
se
ction.And
also
we
hav
e compa
r
ed
the differe
nt
algorith
m
of
sele
ctive en
cryption te
chni
que
based on thei
r re
sult analy
s
is
.
(a)
(b)
Figure 2. Test Image: (a) L
ena (b
) Footb
a
ll
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 883
– 889
886
Figure 3. Partial encryption
of
Lena: (a)
origin
al (b
)en
c
rypted im
ag
e (c)de
c
rypte
d
image
Figure 4. Partial encryption
of five
fa
ces:
(a)o
rigin
a
l im
age (b
)e
ncrypted image
(c
)de
c
ry
pte
d
image
Table 1. Execution time for all image
s an
d facial pa
rt of the image
Executi
on Time
Image
Encr
y
p
t and
decr
y
pt
all image
Encr
y
p
t and
decr
y
pt
part of image
1.493323 second
s
0.483053 second
s
4.1 Securit
y
Analy
s
is
The
se
curity
of the p
r
o
posed te
chni
que
wa
s an
alyzed
usi
n
g
stand
ard hi
stogram
pro
c
ed
ures in
cludi
ng sen
s
itivity
for initia
l value, entrop
y
, and adjace
n
t correlatio
n.
Histo
g
ram Analy
s
is
: The
first security evaluation
of the pro
p
o
s
ed
method h
a
s
been te
sted
b
y
comp
ari
ng th
e histog
ram
s
of encrypte
d
image an
d
plain imag
e. The re
sult
s o
f
this evaluation
are illustrated in Figure
5. T
he proposed method has
shown
m
o
re di
stri
buted frequency
and
more
unreco
gnizable
en
crypted ima
g
e
s
, goo
d stat
i
s
tical
pro
p
e
r
ties a
nd mo
re rob
u
st
aga
inst
s
t
atis
tical attac
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Multi Facial B
l
urri
ng u
s
ing I
m
prove
d
He
n
on Map (Sa
p
a
rudi
n)
887
Figure 5. Hist
ogra
m
analy
s
is of Lena
(a
)o
riginal im
age
(b)en
c
rypted i
m
age
Sensitiv
it
y
:
Another tes
t
with
respec
t t
o
secret
k
e
y is
the key sens
itivity tes
t
that indic
a
tes
how
much a
n
en
crypted im
a
ge is sen
s
itive toward
s
the cha
nge
in the key. For a se
cure
cryptosystem
, a decryptio
n algorithm
will not decry
p
t cipher im
age correctly, even if there
is a
one bit diffe
rence bet
wee
n
key. It mea
n
s that la
rg
e
key sen
s
itivity is re
quired
for hig
h
ly se
cure
crypto
system
s. An ideal i
m
age e
n
cryp
tion sh
ould
b
e
se
nsitive
with respe
c
t to the se
cret key
su
ch th
at a
si
ngle
bit chan
ge in
the
key sh
ould
p
r
od
uce
a
comple
tely different
encrypted
im
ag
e
[8].
The exp
e
rim
ental results
of this resea
r
ch
d
e
mo
nstrated that the
encryption
a
l
gorithm
wa
s very sen
s
itive to the secret
key be
cause it depen
ds on fou
r
initial values. Any small cha
n
ge
in any value,
even if the ra
te of ch
ange
wa
s 1 bi
t m
a
y result in fa
ulty decryptio
n. The diffe
re
n
t
analysi
s
of faulty tests is summari
ze
d in
Table 2.
Table 2. Fa
ce
decryptio
n error
with
faulty initial value for Len
a imag
e
Initial
v
a
lue
Fault decr
y
p
ti
o
n
ima
g
e
1
Initial value fault
decr
y
ption w
i
th x
axes
2
Initial value fault
decr
y
ption w
i
th y
axes
3
Initial value fault
decr
y
ption
w
i
th b
o
th x,
y
axese
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 4, Dece
mb
er 201
4: 883
– 889
888
Entrop
y
:
Informatio
n den
sity, or entro
py, is a meth
od
for me
asu
r
ing u
n
certai
nty in a seri
e
s
of
numbe
rs or b
y
tes. In techn
i
cal term
s, en
tropy me
a
s
u
r
es the level o
f
difficulty or the probability
of
indep
ende
ntly predictin
g e
a
ch n
u
mbe
r
i
n
the se
ries.
In orde
r to an
alyze the entropy
of image encryption, e
a
ch pixe
l
colo
r is re
pre
s
e
n
ted by 8
bits. If entrop
y
is evaluate
d
in the afore
m
enti
one
d ca
se, the entro
py obtained i
s
8. In gene
ral,
the entro
py value of the source
i
s
sm
a
ller than the
ideal value.
Ho
wever,
wh
en imag
es a
r
e
encrypted fo
r a source, it
s
entropy
sh
oul
d be
8 bits i
d
eally. In ca
se
if entropy i
s
l
e
ss than
8 bit
s
,
then the
r
e
e
x
ists a
certai
n de
gre
e
of
predi
ctabilit
y. For a
crypt
o
syste
m
to
resi
st the
entropy
attacks, the e
n
tropy of the crypto
system
shoul
d be cl
ose to the ide
a
l value [9], [10].
The
ent
ropy evaluation of
the
propo
se
d method
is
7.9995 i
n
tab
l
e 3, its cl
ose
to ideal
entropy.
Table 3. Entropy analysi
s
betwe
en auth
o
rs
E
n
t
r
o
p
y
A
n
a
l
ys
i
s
f
o
r
L
e
n
a
i
m
a
g
e
Authors Entrop
y
values
[15] 7.9993
[1] 7.9936
Proposed Metho
d
7.9995
Correla
tion coefficien
t
:
Correl
ation d
e
termin
es the
relatio
n
ship
betwe
en t
w
o
variable
s
. In
othe
r
words,
co
rrel
a
tion is
a me
asu
r
e that
co
mputes th
e d
egre
e
of si
mi
larity betwe
e
n
two vari
abl
es.
The co
rrelation coeffici
ent
is a useful
m
easure
to j
u
d
ge e
n
cryption
quality of
an
y crypto
syste
m
.
An image
cry
p
tosyste
m
is
said to
be g
o
od, if encry
pti
on alg
o
rithm
hide
s all attri
butes
of a pl
ain
image, a
nd
e
n
crypte
d im
a
ge i
s
totally random
an
d h
i
ghly un
co
rrel
a
ted [11]. If e
n
crypte
d im
a
ge
and pl
ain ima
ge a
r
e
compl
e
tely different
then t
hei
r co
rre
sp
ondi
ng
correlation
co
e
fficient mu
st be
very low, or v
e
ry clo
s
e to
zero. If the co
rrelati
on
co
efficient is
equ
al
to one, then
the two imag
es
are ide
n
tical
and they are
in perfect co
rrel
a
tion.
In case of pe
rfect correlation (the correlatio
n
coeffici
ent is
equal to 1
)
, an encryption
pro
c
e
ss com
p
letely
false becau
se
the encrypted
im
age
is
sam
e
a
s
t
he pl
ain i
m
a
ge. When
th
e correlatio
n
co
efficient
i
s
-1 th
en
en
crypted
ima
g
e
is
negative of original (pl
a
in) i
m
age. In sh
o
r
t, the co
rrela
t
ion coeffici
en
t between a
n
image an
d it is
1, the correl
ation co
effici
ent betwe
en
an
image
and totally unrelate
d
ima
ge is zero,
and
correl
ation
coefficient
bet
wee
n
a
n
ima
ge a
nd it
s
n
egative i
s
-1
[12]-[16]. Th
e security of
the
prop
osed te
chniqu
e as
e
v
aluated by
examin
ing th
e co
rrelation
coeffici
ents depen
ding
on
vertical, ho
ri
zontal, an
d diago
nal of the adja
c
ent
pixels in p
l
ain image
and en
crypted
image.Ta
ble
4 presents th
e pe
rform
a
n
c
e compa
r
ison
of the p
r
op
o
s
ed
metho
d
with the
previous
works of these para
m
eters [1],[16].
Table 4. Co
rrelation Coefficient value
s
b
e
twee
n Autho
r
s
Correlati
on C
o
e
fficie
n
t
A
n
al
y
s
is
for Le
na i
m
age
Author
Adjacent pixel w
i
th Diagonal
R G
B
[16] 0.02428
0.04041
0.011095
Proposed
Method
0.0038
0.0031
0.0019
Author
Adjacent pixel w
i
th horizontal and
vertical
Horizontal Vertical
[1] -0.0495
0.0697
Proposed
Method
-0.0376
0.0535
5. Conclusio
n
This pap
er
p
r
esented
a
fa
cial en
cryptio
n
techniq
ue and pro
p
o
s
e
d
a ne
w colo
r
ima
g
e
encryption
al
gorithm
ba
se
d on
imp
r
ov
ed
Hén
on chaotic
map.
The en
crypti
on
p
r
o
c
e
s
s wa
s
applie
d to
all
RGB
chann
els. T
h
e
s
e
chann
els go
throu
gh th
e t
w
o
dimen
s
io
nal
Hén
on
chaotic
map to pro
d
u
c
e a rand
om
bit-stream. In orde
r
to prod
uce the e
n
cry
p
ted image
bi
tXOR ope
rati
on
is exe
c
uted
b
e
twee
n the
random
key a
nd the
orig
i
n
al pixel value
s
of all
ch
an
nels
(RGB).
The
prop
osed sch
e
me ha
s mult
i-dime
nsi
onal
keys to re
si
st all possible b
r
ute-fo
rce types of attack.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Multi Facial B
l
urri
ng u
s
ing I
m
prove
d
He
n
on Map (Sa
p
a
rudi
n)
889
The effe
ctiveness of the
prop
osed m
e
thod was
fi
rstly evaluated
by se
cu
rity analysi
s
,
whi
c
h covere
d histog
ram,
sen
s
itivity,
entropy
, an
d
correlatio
n
analysi
s
. Hi
stogram a
naly
s
is
sho
w
s that t
he hi
stog
ram
of ci
phe
r im
age i
s
flat
or uniformly di
stribute
d
,
so
the alg
o
rithm
i
s
se
cure from
freque
nt anal
ysis attack a
nd the ac
hie
v
ed re
sults
were very promisin
g, the best
r
e
sults on this
dataset
to d
a
te.
As well,
the sele
ctive encry
ption
ap
proa
ch
redu
ces th
e
overh
e
ad
of en
cryptin
g
non-se
nsitive
area
s a
nd
en
han
ced
t
he
e
x
ecution
time
. This stu
d
y a
l
so i
n
vestig
ated
the stre
ngths
of the propo
sed algo
rithm i
n
different
environments
suc
h
as
multi fac
i
al encryption.
Referen
ces
[1]
Kaur, R. Comparativ
e Anal
ysis and Impl
e
m
entatio
n of Image Encr
ypt
i
on Alg
o
rithms.
International
Journ
a
l of Co
mputer Scie
nce
& Netw
ork Security.
2003; 1
3
(
12): 1-10.
[2]
Pisarchik, A., F
l
ores-Carm
o
n
,
N. and Carpi
o
-Val
adez
, M. Encr
yptio
n
and
Decr
yption of
imag
es
w
i
t
h
Chaotic Map Lattices.
Chaos:
An Interdiscip
l
i
nary Jour
nal of
Nonli
n
e
a
r Scie
nce.
200
6; 16(
3): 33-38.
[3]
Ren, W
., Kang, C., Li, Y.,
Gong, L. Chaotic
Imm
une Gen
e
tic H
y
brid A
l
g
o
rithms an
d Its Aplicati
on.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
ng.
2013; 1
1
(2): 9
75-9
84.
[4]
W
e
i-bin, C., Xin,
Z
.
Image
Encryptio
n
Al
g
o
rith
m base
d
on Hén
o
n
Ch
aotic
Syste
m
. IASP
2009.
Internatio
na
l C
onfere
n
ce o
n
Image An
al
ysis
and Si
gn
al Pro
c
essin
g
,
T
a
izhou. 200
9: 94-9
7
.
[5]
Rames
h
Kum
a
r
y
a
dava, DBK. s., SK. Sinha, KK.
Pan
d
e
y
. A Ne
w
Appro
a
ch of
Colo
ur Imag
e
Encr
yptio
n
Bas
ed on
Hén
on l
i
k
e Cha
o
tic Ma
p.
Journa
l of Informatio
n
En
gin
eeri
ng a
nd
Appl
icatio
ns
.
201
3; 3(6): 1-1
0
.
[6]
Pand
ura
nga,
H., NaveenK
u
m
ar, S. Selective Im
age Enc
r
y
p
ti
on for Me
dical
and S
a
te
llite Imag
es.
Internatio
na
l Journ
a
l of Eng
i
n
eeri
ng Scie
nce
& T
e
chnolo
g
y.
2013; 5(2): 3
2
-
43.
[7]
Kester, QA. A
Cr
yptogr
aph
ic
Image E
n
cr
ypti
on T
e
chni
qu
e f
o
r F
a
ci
al-b
lurri
ng
of Imag
es.
International
Journ
a
l of Adv
ance
d
T
e
chn
o
l
ogy an
d Eng
i
n
eeri
ng Res
earc
h
(IJAT
E
R).
2013: ar
Xiv: 13
07
.6409.
[8]
H. Liu, Z
.
Z
hu,
H. Jiang, B. Wang.
A nove
l
i
m
a
ge e
n
crypti
on al
gorith
m
b
a
sed o
n
impro
v
ed 3d ch
aoti
c
cat m
a
p
.T
he 9
th
IEEE International Confer
en
ce forYoung Computer Sc
ient
ists (ICYCS 2008). 2008:
301
6–
302
1.
[9]
R. Ena
y
at
ifar. Image
encr
y
pt
i
on v
i
a
lo
gistic
map fu
nctio
n
a
nd
hea
p tre
e
.
Int. J. Phys. Sci.
201
1; 6(
2)
:
221-
233.
[10]
Z
.
Han, W
.
F
e
ng, L. Hui,
L.
Da Ha
i, L. Ch
ou.
A new
i
m
a
ge e
n
cryptio
n
alg
o
rith
m b
a
se
d on ch
aos
system
. Proce
edi
ngs of IEEE Internation
a
l
Confer
ence
oni
n Rob
o
tics, Intelli
ge
nt S
y
stem
s and Sig
n
a
l
Processi
ng. 20
03; 2: 778
–78
2
.
[11]
I. Elashr
y, O. A
llah,
A. Ab
bas,
S. El-Ra
bai
e,
F
.
El-Samie.
H
o
momor
phic
i
m
age
encr
y
pt
i
on.
Jo
urn
a
l of
Electron
ic Ima
g
in
g.
200
9; 18: 0330
02.
[12]
H. Elkamc
houc
hi, M. Mak
a
r.
Measuri
n
g
e
n
c
r
yption
qu
al
ity
for bit
m
a
p
i
m
a
ges
encrypt
ed
w
i
th rijn
dae
l
and k
a
mk
ar bl
ock ciph
ers
. Procee
din
g
s of
the 22
nd
. IEEE National Radio
Scienc
e Conference
(NRSC’ 2
0
0
5
). 200
5: 277
–2
84
.
[13]
N. El-F
isha
w
y
and O. Z
a
i
d
. Qualit
y
of e
n
cr
yp
tion me
asur
em
ent of b
i
tmap
i
m
ages
w
i
th rc
6
,
mrc6, and
rijnd
a
e
l
block c
i
ph
er alg
o
rithm
s
.
Internationa
l Journ
a
l of Net
w
ork Security.
200
7; 5(3): 241
–25
1.
[14]
Abug
hars
a
, AB
., Hasan
Bas
a
ri, ASB., Alman
gush,
H. A N
o
vel Ima
ge E
n
cr
ypti
on
usi
n
g
an
Integrati
o
n
T
e
chnique of B
l
ocks Rotation
based on
the
Magic c
u
be and
the AES Algorithm.
Intern
ation
a
l J
ourn
a
l
of Computer S
c
ienc
e Issues (IJCSI).
2012; 9(4): 1-12.
[15]
Massou
d
i, A., Lefebvr
e
, F
., De Vle
e
sch
ou
w
e
r, C.,
Macq, B., Quisquater,
JJ. Overvie
w
on Se
lectiv
e
Encr
yptio
n
of Image an
d
Vi
deo: Ch
all
eng
es
a
n
d
Persp
e
ctives.
EURA
SIP Journal on Infor
m
ation
Security.
200
8:
1792
90.
[16]
W
ang,
X., Z
hao, J., Liu,
H. A Ne
w
I
m
age E
n
cr
ypti
on En
nc
ypti
on
base
d
on
C
haos.
Optics
Co
mmun
icati
o
ns.
2012; 2
85(
5): 562-5
66.
Evaluation Warning : The document was created with Spire.PDF for Python.