TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 11, Novembe
r
2014, pp. 79
5
2
~ 796
2
DOI: 10.115
9
1
/telkomni
ka.
v
12i11.64
80
7952
Re
cei
v
ed
Jul
y
13, 201
3; Revi
sed Septe
m
ber
18, 201
4; Acce
pted
Octob
e
r 8, 20
14
A Chaotic System Based Image Encryption Algorithm
using Plaintext-related Confusion
Yong Zhang
Schoo
l of Softw
a
r
e a
nd Com
m
unic
a
tion En
gin
eeri
ng, Jia
n
g
xi Univ
ersit
y
o
f
F
i
nance an
d
Econom
ics,
Nanc
han
g, P. R. Chin
a
E-mail: zhangyong@jx
u
fe.edu.cn
A
b
st
r
a
ct
A new
plai
ntex
t-related i
m
ag
e
enc
rypti
on sy
stem b
a
se
d o
n
the hy
per-ch
aotic L
o
ren
z
s
ystem i
s
prop
osed i
n
this pap
er. In th
e prop
osed i
m
age e
n
cr
yptio
n
system, the hyper-ch
aotic L
o
ren
z
syste
m
i
s
empl
oyed
to
g
ener
ate s
i
x ps
eud
o-ran
d
o
m
matrix
es, w
her
e, tw
o of th
e
m
use th
e “
a
d
d
-modu
lus
”
op
erati
o
n
s
to carry out th
e pla
i
ntext-u
n
r
e
late
d i
m
a
ge d
i
ffusion, a
nd th
e other fo
ur matrixes are
use
d
to confus
e th
e
plai
ntext-rel
a
te
d imag
e. In the
ima
ge co
nfusi
on, eac
h pixe
l
w
ill sw
ap its location w
i
th an
o
t
her pixe
l, and
the
target l
o
catio
n
is d
e
ter
m
in
ed
by so
me e
l
e
m
ents i
n
th
e p
l
ai
n i
m
a
g
e
a
nd t
he fo
ur
matrix
es. T
he
prop
o
s
e
d
imag
e encry
pti
on syste
m
ca
n resist the c
hose
n
/know
n
plai
ntext attac
ks due to the
appl
icati
on of
the
plaintext-related image
confusion in it. The
simula
tion res
u
lts show that
t
he pr
oposed
enc
ryption syst
em
also h
a
ve the
characters of
fast
encryptio
n/decrypti
on s
pee
d, larg
e ke
y space, strong key sensitiv
ity,
strong pl
ai
ntex
t sensitivity,
go
od statistical
pr
operti
es of cip
her i
m
a
ges, an
d larg
e infor
m
a
t
ion entro
py, etc.
Thus, the proposed system
c
an be us
ed in practical communic
a
tions.
Ke
y
w
ords
:
image encry
ption, hyper-chaotic Loren
z
system
, plai
ntext-r
e
lated confusion,
diffusion, security
ana
lysis
Copy
right
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Image e
n
cryp
tion is
an im
p
o
rtant
resea
r
ch topi
c i
n
im
age info
rmati
on security field. For
image,
due
to its
hug
e
data volu
me
and
stron
g
inform
ation
re
dun
dan
cy
ch
aracte
rs,
the
traditional
text-based data en
cryption
methods, such as
D
ES,
AES and
RSA, etc., are no
t
suitabl
e for its encryption. In
rece
nt decade
s, scie
n
tists have ex
plo
r
ed the ch
aoti
c
system b
a
sed
image e
n
cryp
tion tech
nolo
g
y, and obtai
ned
rich
a
c
hi
evements [1-7]. In these i
m
age e
n
crypt
i
on
system
s, cha
o
tic system
s
are em
ployed
to generate t
he se
cret co
d
e
strea
m
s for
encryption, a
nd
then en
crypt the plain ima
g
e
s into noi
se
-like via circul
arly confu
s
io
n and diffusi
o
n
.
In
s
o
me
imag
e
en
cr
yp
tion
s
y
stem
s [8
-11], thei
r
se
cret
code
streams (i.e.
e
quivalent
se
cret
keys)
are
gen
erate
d
by ite
r
ating
the
cha
o
ti
c
sy
st
em
s wit
h
t
he se
cr
et
keys se
rver as
the
initial value
s
or pa
ramet
e
rs,
and
a
r
e
unrelat
ed
with the pl
ain
image
s. T
h
ese
ma
ke th
em
vulnera
b
le to
the
cho
s
e
n
/kno
wn
plai
ntext atta
cks [
12-1
6
]. They
nee
d to i
n
crease thei
r
ro
und
number of the iterating to impr
ove their capabilities of resist
ing the chosen/known plaintext
attac
k
s
,
thus their
enc
r
ypti
on/decryptio
n spee
ds
are re
du
ce
d
g
r
eatly. To m
a
ke
the im
a
ge
encryption
system fight ag
ainst
the
cho
s
en/
kno
w
n pl
aintext attacks whil
e the e
n
cryptio
n
spe
eds
are hi
gh e
n
o
ugh, some
schol
ars have
prop
osed
pl
aintext-rel
a
te
d image
en
cryption metho
d
s,
whi
c
h can be
divided into the followi
ng two catego
rie
s
:
(1)
Divide th
e plain i
m
ag
e into a
plurality of blocks, and
en
cry
p
t each ima
ge blo
ck
seq
uentially. The secret co
de stream
s of
each
blo
ck a
r
e co-g
ene
rat
ed by the ci
p
her bl
ock of its
previou
s
blo
c
k and the se
cret keys. Th
us, the ci
ph
e
r
block of the current blo
ck is n
o
t only
related
with t
he secret
ke
y, but also in
dire
ctly
relate
d with its pre
v
ious pl
ain bl
ock. After two
roun
ds of o
peratio
ns, ea
ch cip
h
e
r
block is
indire
ctly related with the wh
ole plain im
age.
Therefore, di
fferent plain
image
s co
rre
s
po
nd to
different e
quival
ent se
cret keys, makes t
h
e
enc
ry
pt
ion sy
st
em c
an re
si
st
t
he cho
s
e
n
/
kno
w
n plai
nt
ex
t
at
t
a
cks [
1
7-22]
.
(
2
)
D
i
vide the s
e
cr
et key into tw
o levels
:
the
fir
s
t level s
e
cr
et key is the s
y
mmetric
k
e
y
sha
r
ed
by
bo
th co
mmuni
cation p
a
rtie
s;
the second
le
vel se
cret
ke
y is
co
-ge
nerated by th
e fi
rst
level key an
d
the plain im
a
ges. Empl
oy the first level
key to en
cryp
t the plain im
age
s to get t
h
e
interme
d
iate
ciph
er imag
e
s
, an
d the
n
e
m
ploy the
se
con
d
level
ke
y to en
crypt t
he inte
rme
d
i
a
te
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Chaotic S
y
stem
Based Im
age Encryp
tion Algorit
hm
using Plainte
x
t-related
…
(Yong Zha
ng)
7953
ciph
er i
m
ag
e
s
to
get the
final
ciph
er i
m
age
s. Th
u
s
, d
i
fferent plai
n i
m
age
s
co
rre
spond
to different
se
con
d
level keys, an
d the
r
eby co
rresp
ond to di
ffere
nt equivalent
keys. Thi
s
ki
nd of encryption
sy
st
em
can r
e
si
st
t
he cho
s
en/
kn
ow
n plai
nt
ex
t
at
t
a
cks
[
23]
.
The foresaid
two plai
ntext-related
en
cry
p
ti
on sy
stem
s have
hig
h
cryptographi
c
se
curity.
Ho
wever,
be
cau
s
e
plai
ntext-relate
d i
m
age
diffu
si
on i
s
u
s
e
d
i
n
both
of th
em, an
d
ch
aotic
system
s
a
r
e
iterated
to gene
rate ne
w se
cret c
o
d
e
stream
s i
n
their
diffusi
on p
r
o
c
e
ss,
their
encryption
sp
eed
s a
r
e
slo
w
. In o
r
de
r to
improv
e
the encryption sp
eed,
a
ne
w p
l
aintext-rel
a
te
d
image en
cryp
tion system i
s
pro
p
o
s
ed i
n
this pape
r.
The prop
osed system u
s
e
s
the plain
t
ext-
related
confu
s
ion
an
d pl
ai
ntext-unrelate
d
diffusi
on
while n
eed
s
n
o
round
o
perations, th
us i
t
s
encryption sp
eed is in
crea
sed
without lo
ss of security.
The re
mind
er of this pape
r is o
r
ga
nize
d as follo
ws: Section 2 d
e
tails the e
n
c
ryption
scheme of th
e prop
osed system; Sectio
n 3 gives
so
me simulatio
n
results; Section 4 analy
z
e
s
the se
curity p
e
rform
a
n
c
e o
f
the propo
se
d system; Se
ction 5 summ
arie
s the pap
er.
2. Image Encr
y
p
tion Scheme
2.1. Used
Ch
aotic Sy
stem
The hype
r-ch
aotic Lo
ren
z
system i
s
use
d
in this pap
e
r
. Its equat
ion
is as follo
ws:
(1)
Whe
r
e,
a
,
b
,
c
an
d
r
a
r
e t
he pa
ramete
rs of hype
r-ch
aotic Lo
re
nz
system. Whe
n
a
=10,
b
=8/3,
c
=28 and -1.
52<
r
≤
-0.06, Equation (1
) falls
into
th
e cha
o
tic state.
If
r
=-1, Equ
a
t
ion (1
) h
a
s four
Lyapun
ov expone
nts, nam
ely
λ
1
=0.3381
,
λ
2
=0.1586,
λ
3
=0 and
λ
4
=-1
5
.1752.
In Equation
(1), the
ra
n
ge value
s
o
f
the state
variable
s
x
0
,
y
0
,
z
0
an
d
w
0
ar
e
,
r
e
spec
tively,
x
0
ϵ
(-40,4
0
),
y
0
ϵ
(-4
0,40
),
z
0
ϵ
(1,81),
w
0
ϵ
(-250,25
0). {
x
0
,
y
0
,
z
0
,
w
0
}
i
s
part
of the
se
cret
key in th
e p
r
opo
sed
syst
e
m
. Whe
n
di
screte
the Eq
u
a
tion (1)
with
the fou
r
th-o
rder
Ru
nge
-K
utta
method, the step size is 0.0
02.
2.2. Basic Pr
inciple of Encr
y
p
tion S
y
s
t
em
The typical pl
aintext-rel
a
te
d image e
n
crypti
on syste
m
based on
cha
o
tic sy
ste
m
has th
e
stru
cture a
s
sho
w
n
in
Fig
u
re
1. Th
ey i
n
clu
de
seve
ral round
s of
plaintext-un
re
lated
confu
s
i
on
and plaintext
-rel
a
ted diffusion
op
eratio
ns.
T
he
ch
a
o
tic
system
i
s
e
m
ployed
to gen
erate
t
h
e
se
cret code stream
s.
Figure 1. Structure of Typi
cal Image Encryption Syste
m
The
pro
p
o
s
e
d
en
cryption
system
is a
s
sho
w
n
in
Fig
u
re
2.
Differe
nt from
Figu
re 1, in
ou
r
prop
osed
system: (1
) T
he confu
s
ion
is plaintex
t-related,
whil
e the diffusi
on is pl
ainte
x
t-
unrel
ated; (2
) On
e
confu
s
ion
and
two diffusi
on o
peratio
ns wit
hout
roun
ds need
ed,
what
improve
s
th
e
en
cryption/d
e
cryptio
n
sp
eed
gre
a
tly, and
do n
o
t
wea
k
e
n
the
se
curity of t
he
encryption sy
stem.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 79
52 – 796
2
7954
Figure 2. Structure of o
u
r P
r
opo
se
d Imag
e Encryptio
n
System
2.3. Encr
y
p
tion Scheme
Acco
rdi
ng to Figure 2, our prop
osed en
cryption
syst
em co
nsi
s
ts
of four pa
rts,
namely,
se
cret
cod
e
strea
m
gen
erator,
plaintex
t-unrelated di
ffusion I, plai
ntext-relate
d confu
s
io
n, an
d
plaintext-unrelated diffus
i
on II. T
he following four subs
ec
tions
will
detail the proc
edures of t
h
e
foresaid four parts
.
The se
cret key of the proposed is
K
={
x
0
,
y
0
,
z
0
,
w
0
,
r
1
,
r
2
}, where,
{
x
0
,
y
0
,
z
0
,
w
0
} come
s
from Sectio
n 2.1, and
r
1
a
nd
r
2
are two
8-bit ran
dom
unsig
ned in
t
egers. Assu
me that the plain
image is d
e
n
o
ted by
P
, whose
size is
M
×
N
, an
d gray
scale level is
L
-bit.
2.3.1. Secret
Code Strea
m
Genera
tin
g
Algorithm
The hype
r-ch
aotic Lo
ren
z
sy
stem a
s
shown in Eq. (1), is e
m
plo
y
ed to generate six
pse
udo
-rand
om matrixe
s
, denoted
by
X
,
Y
,
Z
,
W
,
U
and
V
(
a
ll of
size
M
×
N
) by u
s
ing
the
f
o
llowin
g
con
c
ret
e
st
ep
s:
Step 1.
Us
e {
x
0
,
y
0
,
z
0
,
w
0
} in
K
a
s
the in
itial values of
E
quation (1),
iterate Equat
ion (1
)
for
r
1
+
r
2
time
s to bypass the tran
sient
state, and then contin
ue i
t
erating
MN
t
i
mes to get four
pse
udo
-rand
om seq
uen
ce
s, namely, {
x
i
}, {
y
i
}, {
z
i
} and
{
w
i
},
i
=
1,2,…
,
MN
.
Step 2.
Prod
uce
matrixe
s
X
,
Y
,
Z
,
W
,
U
and
V
from t
he sequ
en
ce
s of {
x
i
}, {
y
i
}, {
z
i
} and
{
w
i
},
i
=
1,2,…,
MN
, with the following formulas
:
X
(
k
,
l
)=
floor
(
(
x
(
k
-1)×
N
+
l
+500 mod 1)×10
13
) mod
2
L
(2)
Y
(
k
,
l
)=
floor
(
(
y
(
k
-1)×
N
+
l
+500 mod 1)×10
13
) mod
2
L
(3)
Z
(
k
,
l
)=
(floo
r
(
z
(
k
-1)×
N
+
l
×10
13
) mod
M
)+
1
(4)
W
(
k
,
l
)
=
(flo
or
((
w
(
k
-1)×
N
+
l
+500 mod 1)×1
0
12
) mod
N
)+
1
(5)
U
(
k
,
l
)=
(
f
loor
((
x
(
k
-1)×
N
+
l
+
y
(
k
-1)×
N
+
l
+500 mod
1)×10
12
)
mod
M
)+
1
(6)
V
(
k
,
l
)
=
(flo
or
((
z
(
k
-1)×
N
+
l
+
w
(
k
-1)×
N
+
l
+500 mod
1)×10
12
)
mod
N
)+
1
(7)
Whe
r
e, floor(
t
) retu
rn
s the integer n
o
t greater than
t
, ‘+50
0’ co
nvert
s
the neg
ative state value
s
of
x
,
y
and
w
into the positive
numbe
rs.
2.3.2. Plaintext-unr
ela
t
ed Diffusion
I
Conve
r
t the
plain ima
ge
P
into the matrix
A
by e
m
ploying p
s
e
udo-matrix
X
with the
f
o
llowin
g
st
ep
s:
Step 1.
Let
i
=1,
j
=1.
Step 2.
Tra
n
sform
P
(
i
,
j
) int
o
A
(
i
,
j
) by u
s
i
ng the followi
ng formul
a:
A
(
i
,
j
)=
P
(
i
,
j
)+
X
(
i
,
j
)+
r
1
mod 2
L
(8)
Step 3.
Let
j
=
j
+1.
Step 4.
Tra
n
sform
P
(
i
,
j
) int
o
A
(
i
,
j
) by u
s
i
ng the followi
ng formul
a:
A
(
i
,
j
)=
P
(
i
,
j
)+
A
(
i
,
j
-1)
+
X
(
i
,
j
) mod 2
L
(9)
Step 5.
If
j
<
N
, then goto Step 3.
Els
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Chaotic S
y
stem
Based Im
age Encryp
tion Algorit
hm
using Plainte
x
t-related
…
(Yong Zha
ng)
7955
j
=1,
i
=
i
+1.
If
i
<=
M
, then goto Step 6.
E
l
s
e
goto Step 8.
End
Step 6.
Tra
n
sform
P
(
i
,
j
) int
o
A
(
i
,
j
) by u
s
i
ng the followi
ng formul
a:
A
(
i
,
j
)=
P
(
i
,
j
)+sum(
A
(
i
-1,1 to
N
))
+
X
(
i
,
j
) mo
d 2
L
(10)
Whe
r
e, s
u
m(
t
) retu
rn
s the sum of all the
elements in t
he vector
t
.
Step 7.
Go to
Step 3.
Step 8.
End
。
2.3.3. Plaintext-r
elated Co
nfusion
Confu
s
io
n algorithm is u
s
ed to disru
p
t the pixel
locatio
n
s of the image, so as to
eliminate th
e
correl
ation
be
tween
adj
ace
n
t pixels in t
h
e o
r
iginal
ima
ge. Th
e p
r
op
ose
d
pl
aintex
t-
related
confu
s
ion alg
o
rith
m transfo
rm
s the matrix
A
into the matrix
B
by using the following
st
ep
s:
Step 1.
As f
o
r a given
pi
xel coo
r
din
a
te (
i
,
j
) in the
image
A
, cal
c
ulate the va
lue of
c
o
or
d
i
na
te
(
m
,
n
) by using t
he followi
ng formul
a:
m
=(
U
(
i
,
j
)+
su
m(
A
(
Z
(
i
,
j
), 1 to
N
) mo
d
M
)+1
(11)
n
=(
V
(
i
,
j
)+
sum(
A
(1 to
M
,
W
(
i
,
j
)) mod
N
)+
1
(12)
If
m
=
i
or
Z
(
i
,
j
), or
n
=
j
or
W
(
i
,
j
), o
r
Z
(
i
,
j
)=
i
, or
W
(
i
,
j
)=
j
, then ke
ep
the location
of
A
(
i
,
j
)
unchan
ged; e
l
se, swap the
locatio
n
s of
A
(
i
,
j
) and
A
(
m
,
n
).
Step 2.
W
hen
th
e
c
o
or
d
i
na
te
(
i
,
j
) t
r
ave
r
se
s th
e who
l
e imag
e
A
i
n
the
scanni
ng o
r
de
r
from left to right and from top to bottom,
repe
at Step 1 to convert th
e matrix
A
into the matrix
B
.
2.3.4. Plaintext-unr
ela
t
ed Diffusion
II
Different f
r
om
the alg
o
rith
m de
scribe
d i
n
Se
ctio
n 2.3
.
2, the plainte
x
t-unrel
ated
diffusion
II in this subsection
carrie
s out the forward diffu
si
on
operation
s
from the la
st pi
xel of the image.
The al
go
rithm
of diffusi
on II
employ
s the
pse
udo
-rand
om matrix
Y
to tran
sfo
r
m t
he mat
r
ix
B
in
to
the matrix
C
by using the f
o
llowin
g
step
s:
Step 1.
Let
i
=
M
,
j
=
N
.
Step 2.
Co
nvert
B
(
i
,
j
) into
C
(
i
,
j
) by u
s
ing
the following
formula:
C
(
i
,
j
)=
B
(
i
,
j
)+
Y
(
i
,
j
)+
r
2
mod 2
L
(13)
Step 3.
Let
j
=
j
-1.
Step 4.
Co
nvert
B
(
i
,
j
) into
C
(
i
,
j
) by u
s
ing
the following
formula:
C
(
i
,
j
)=
B
(
i
,
j
)+
C
(
i
,
j
+1
)+
Y
(
i
,
j
) mod 2
L
(14)
Step 5.
If
j
>1, then goto Step 3.
Els
e
j
=N,
i
=
i
-1.
If
i
>=1, then goto Step 6.
E
l
s
e
goto Step 8.
End
Step 6.
Co
nvert
B
(
i
,
j
) into
C
(
i
,
j
) by u
s
ing
the following
formula:
C
(
i
,
j
)=
B
(
i
,
j
)+sum(
C
(
i
+
1
,1 to
N
))
+
Y
(
i
,
j
) m
o
d 2
L
(15)
Step 7.
Go to
Step 3.
Step 8.
End.
The matrix
C
is the cip
her i
m
age.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 79
52 – 796
2
7956
Acco
rdi
ng to
the d
e
script
i
on of
en
cry
p
ti
on
schem
e, the d
e
cry
p
tion p
r
o
c
e
s
s i
s
the
reverse of en
cryption p
r
o
c
ess.
3. Simulation Resul
t
s
(a)
(b)
(c
)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
(k
)
(l)
(m)
(n)
(o)
Figure 3. Simulation Results. (a)-(c) Plain
image
s of Lena, Baboo
n and Pepp
er, resp
ectively;
(d)
-(f
) Ciph
er
image
s of (a)
-
(
c
), re
sp
ectiv
e
ly
; (g)-
(i) D
e
cry
p
ted im
ag
es of (d
)-
(f), r
e
sp
ectiv
e
ly
; (j)-
(l) Hi
stog
ram
s
of (a)
-
(
c
), re
spe
c
tiv
e
ly
; (m
)-
(o)
Histo
g
ra
ms of (d
)-
(f), r
e
sp
ectiv
e
ly
0
50
10
0
150
200
250
0
300
600
900
120
0
150
0
G
r
ay
s
c
al
e
D
i
st
r
i
but
i
o
n
0
50
100
150
200
250
0
300
0
600
0
900
0
120
00
G
r
ay s
c
al
e
D
i
st
r
i
but
i
o
n
0
50
100
15
0
20
0
250
0
30
0
60
0
90
0
120
0
150
0
G
r
ay s
c
al
e
D
i
st
r
i
but
i
o
n
0
50
100
150
20
0
25
0
0
200
400
600
800
Gr
a
y
s
c
a
l
e
D
i
st
r
i
but
i
o
n
0
50
100
150
20
0
25
0
0
200
400
600
800
Gr
a
y
s
c
a
l
e
D
i
st
r
i
but
i
o
n
0
50
10
0
150
200
25
0
0
200
400
600
800
Gr
a
y
s
c
a
l
e
D
i
st
r
i
but
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Chaotic S
y
stem
Based Im
age Encryp
tion Algorit
hm
using Plainte
x
t-related
…
(Yong Zha
ng)
7957
The
com
put
er u
s
e
d
fo
r sim
u
lation
is
co
nfigured
with the
Intel Duo
Co
re
I5
M460
@2.5
3
G
Hz,
2
G
B
DDR3 RAM, Wind
ows 7
a
nd MAT
L
AB
8.1. We
em
pl
oyed the
p
r
o
posed
system
with
the
se
cret key of
K
={3.31
3
3
, 12.05
46,
4
0
.8879,
-3
4.5
677, 3
5
, 2
01}
, to en
crypt
the
plain ima
g
e
s
of Lena, Ba
boon a
nd Pe
pper
(all
size
of 357×317,
as sho
w
n in
Figure 3a
-3
c,
respe
c
tively) to obtain thei
r ciph
er ima
g
e
s (a
s
sho
w
n in Figs. 3d
-3f, respe
c
tively). Then, we
us
ed the cor
r
e
c
t
secr
et
k
e
y
K
to d
e
c
rypt the
cip
her ima
g
e
s
(as
sh
own i
n
Figu
re 3d
-3
f,
respe
c
tively) to obtain the
recovere
d image
s (a
s
s
how
n in Fig
u
re 3g
-3i, re
spe
c
tiv
e
ly
). The
histog
ram
s
of plain
imag
es Len
a, Bab
o
o
n
an
d Pe
ppe
r (a
s
sh
own in
Figu
re
3a
-3
c, re
spe
c
tively)
are a
s
sho
w
n
in Figure 3j-3l, respe
c
tively. The hi
stog
rams of
ciph
er image
s (a
s
shown in Figu
re
3d-3f, re
sp
ect
i
vely) are a
s
sho
w
n in Fig
u
re 3m
-3o, re
spe
c
tively.
It can b
e
se
en from
Figu
re 3
that: (1
) The
ciph
er i
m
age
s (as shown in Fi
gu
re 3
d
-3f,
respe
c
tively)
are
noi
se
-like
imag
es with
out any
le
a
k
a
ge of
visual
i
n
formatio
n; (2) T
h
e
de
cry
p
ted
image
s
(a
s
shown in
Fig
u
r
e
3g-3i, respectively) ar
e
identi
c
al
to t
he o
r
igi
nal
pl
ain im
age
s
(as
s
h
ow
n in
F
i
gu
r
e
3
a
-
3
c
,
res
p
ec
tive
ly)
;
(3
)
Th
e
cip
her image
s have
flat histo
g
ra
ms
(a
s
sho
w
n in
Figure 3m-3o
)
, which is wa
y different from those of pl
ain image
s (a
s sh
own in Figure 3j
-3l
)
.
4. Securit
y
A
n
aly
s
is
For th
e im
a
ge e
n
cryptio
n
sy
stem, th
e commo
n u
s
ed
security
evaluatio
n
rules a
r
e
encryption/de
cryption
spe
e
d
, key spa
c
e,
statistical
ch
ara
c
ters of ci
pher ima
g
e
s
, key sen
s
itivity,
plain imag
e
sen
s
itivity (i.e. resi
sting
the diffe
ren
t
ial at
t
a
ck
s),
resi
st
ing t
h
e cho
s
e
n
/
k
n
o
wn
plaintext attacks an
d information entrop
y
, etc.
4.1. Encr
y
p
tion and De
cr
y
p
tion Spee
d
Without lo
ss of generalit
y, we only consi
der the
encryption/de
cr
yption
spe
ed of the
image
with
th
e si
ze
of
357
×31
7
.
When
we
use
t
he propo
sed sy
ste
m
to encrypt the plain imag
es
wit
h
t
h
e
se
cr
et
key
K
={
3.3
133, 1
2
.054
6
,
40.887
9, -3
4.5677,
35,
2
01} (id
entical to
the key used
in Sectio
n 3), the time co
nsum
ed fo
r
gene
rating th
e se
cret cod
e
stream
s in
Section
2.3.1 i
s
T
1
≈
0.259
85
s, and th
e tim
e
co
nsume
d
for diffusi
on-c
onfu
s
io
n-diff
usio
n op
erati
ons i
n
Se
ctio
ns
2.3.2-2.3.4 i
s
T
2
≈
0.09
126
s. So the time
for en
crypting
one i
m
age
is
T
1
+
T
2
≈
0.351
11s. Wh
en we
use
the
prop
ose
d
system
to en
crypt
L
pieces of im
a
ges, th
e
aver
age tim
e
fo
r
encrypting
ea
ch
image is
T
1
/
L
+
T
2
, becau
se
the algorithm
steps fo
r gen
erat
ing
the se
cret code stre
ams
only
ne
e
d
to be
executed o
n
ce. If
L
=100
0, the
avera
ge time
for en
cryptin
g
ea
ch
imag
e i
s
T
1
/
L
+
T
2
≈
0.0
9
152
s. Note t
hat
T
1
i
s
slig
htly affected
by
r
1
a
nd
r
2
. When
r
1
=
r
2
=255,
T
1
t
a
ke
s it
s
maximum of about 0.26
12
5s, and
whe
n
r
1
=
r
2
=0,
T
1
take
s its minim
u
m f about 0.2593
0s.
Table 1 list
s
the encryptio
n/dec
ryption time of our propo
sed
syste
m
and that of [17]. It
can b
e
se
en
from Table
1 that for on
e image
en
crypting, the prop
osed sy
stem ha
s sl
o
w
er
encryption/de
cryption spe
ed
than
the
system in [17]; while for 100
0 pie
c
e
s
of imag
es
encrypting, the prop
osed system ha
s ab
out
triple sp
e
ed than the system in [17].
Table 1. Co
m
pari
s
on
Re
sul
t
s of Encrypti
on/De
cryptio
n
Time (s)
Encry
p
tion scheme
Time for one
plain image
Aver
age time for
1000 plain image
s
Encry
p
tion Decry
p
tion
Encry
p
tion Decry
p
tion
Proposed
0.35111
0.34277
0.09152
0.08318
Ref. [17]
0.23725
0.26294
0.23725
0.26294
4.2. Ke
y
Spa
c
e
The
se
cret
key of the p
r
o
posed
syste
m
is
K
={
x
0
,
y
0
,
z
0
,
w
0
,
r
1
,
r
2
}, where, the rang
e
values of
x
0
,
y
0
,
z
0
and
w
0
are
sep
a
rate
ly
x
0
ϵ
(-40,40
),
y
0
ϵ
(-40,40
),
z
0
ϵ
(1,81),
w
0
ϵ
(-2
50,25
0); th
e
step size
of
x
0
,
y
0
or
z
0
is
10
-13
, and the
step
si
ze of
w
0
is
10
-12
;
r
1
and
r
2
are in
tegers rang
e
in
[0,255] with
the ste
p
size of 1. So
the
ke
y
spa
c
e si
ze
of th
e propo
se
d
system i
s
ab
out
1.6777
×1
0
64
,
equivalent to the length of key being 2
1
3
bits. If we use the co
mpu
t
er in Section
3
to perfo
rm th
e exhau
stive
attack, it will t
a
ke
us
an av
erag
e of ab
o
u
t 9.1177
×1
0
55
years to
crack
the encryptio
n system. Th
erefo
r
e, the p
r
opo
se
d syst
em can
re
sist
the brute-fo
rce attacks.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 79
52 – 796
2
7958
4.3. Statistic
a
l Chara
c
te
r
s
of Ciphe
r Images
Gene
rally, we ca
n inve
stigate the e
n
cryption
syste
m
again
s
t the
statistical attacks f
r
om
two a
s
p
e
ct
s:
One i
s
wheth
e
r th
e hi
stog
rams of
ci
ph
e
r
ima
g
e
s
are
flat;
the other is whethe
r the
adja
c
ent pixel
s
in the cip
h
e
r
image
s hav
e stron
g
co
rrelation
s
. With
out loss of ge
nerality, we take
the ciphe
r im
age
s sho
w
n i
n
Figure 3
d
-3f as exampl
es. From Fig
u
re 3m
-3o, we can see th
at
these
cipher i
m
ages have f
l
at histogram
s
. So the following
will ma
i
n
ly discuss the
correlations of
adja
c
ent pixel
s
in the cip
h
e
r
image
s.
Assu
ming th
at we rand
o
m
ly choo
se
N
pairs of adjacent pixels, denoted b
y
(
x
i
,
y
i
),
i
=
1
,2,…,
N
(L
abelin
g
x
={
x
i
},
y
={
y
i
}
)
. The
n
the correlat
ion coefficie
n
t
r
xy
betwe
en
x
and
y
can
be
cal
c
ulate
d
usi
ng the followi
ng formul
as:
,
(
1
6
)
co
v
,
∑
(17)
∑
(18)
∑
(19)
Whe
r
e, cov
(
x
,
y
) re
prese
n
t
s
the
cova
ria
n
ce
of vecto
r
s
x
a
nd
y
,
D
(
x
) rep
r
esents the varia
n
ce
of
x
,
E
(
x
) re
pre
s
ents the mea
n
value of
x
, and
N
d
enote
s
the length o
f
x
.
Table 2. Co
rrelation Coefficient
s
of the Plain and Ci
p
her Imag
es
Horizontal Vertical
Diagonal
Fig. 3a
Fig. 3d
Fig. 3b
Fig. 3e
Fig. 3c
Fig. 3f
Lena
0.952573
-0.004649
0.971203
-0.051114
0.920942
-0.016763
Baboon
0.884423
-0.023753
0.760920
-0.008060
0.722701
-0.036221
Pepper
0.981947
-0.003522
0.980936
0.013633
0.958665
0.007701
(a)
(b)
(c
)
(d)
(e)
(f)
Figure 4. Re
sults of Co
rrel
a
tion Analysi
s
; (a)-(c) Co
rrelation
s
in ho
rizo
ntal direct
ion for the pla
i
n
image
s of Le
na, Baboon a
nd Peppe
r, re
spe
c
tively; (d)-
(f) Co
rrelati
ons in h
o
ri
zo
ntal dire
ction
for
the ciph
er im
age
s of Lena,
Baboon an
d Peppe
r, resp
ectively
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
0
50
100
150
200
250
300
0
50
100
150
200
250
300
P
i
x
e
l
g
r
a
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
)
P
i
xe
l
gra
y
va
l
u
e
on l
o
c
a
t
i
on(
x
,
y
+1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Chaotic S
y
stem
Based Im
age Encryp
tion Algorit
hm
using Plainte
x
t-related
…
(Yong Zha
ng)
7959
Her
e
, we
tak
e
N
=20
00, a
nd calculate t
he correlatio
n co
efficient
s of adja
c
ent
pixels of
plain imag
es
(as
sh
own i
n
Figure 3
a
-3c, resp
ec
tively
) and
ciph
er i
m
age
s (a
s shown in Figu
re
3d-3f,
re
spe
c
tively) in the
hori
z
ontal,
ve
rtical
an
d di
a
gonal
directio
ns. T
he
cal
c
u
l
ated
re
sults
are
listed in
Tabl
e 2. Me
an
wh
ile, the correl
ations i
n
the
hori
z
o
n
tal di
rectio
n fo
r Fi
gure
3a
-3f a
r
e
illustrated in Figure 4.
As appa
rent from Tabl
e 2 and Figu
re 4,
the adjace
n
t pixels in plain
images h
a
ve
strong
correl
ation
s
,
and th
eir correlation
co
efficient
s a
r
e
cl
o
s
e to
1;
while
the a
d
ja
cent
pixels in
cip
her
image
s hardl
y have correl
ations, which
is clos
e to 0. These d
e
m
onst
r
ate th
at the propo
sed
system
can fi
ght again
s
t the statistical attacks.
4.4. Ke
y
Sen
s
itivit
y
Analy
s
is
A
s
f
o
r t
he se
cret
key
K
1
={
x
0
,
y
0
,
z
0
,
w
0
,
r
1
,
r
2
}, any element of {
x
0
,
y
0
,
z
0
} cha
n
ges its
value by 10
-13
, or
w
0
chan
g
e
s its value b
y
10
-12
, or any element of {
r
1
,
r
2
} chan
ge
s its value by 1
,
to get the n
e
w
secret
key
denote
d
by
K
2
.
The sensiti
v
ities of the
secr
et key will
be analyzed i
n
the followin
g
two a
s
pe
cts:
(1) Use
the
prop
osed
sy
stem with
the
se
cret
keys
of
K
1
and
K
2
to e
n
crypt t
he pl
ain
image
P
to
o
b
tain two
cip
her imag
es,
d
enoted
by
C
1
and
C
2
, res
p
ec
tively. Compare the images
of
C
1
and
C
2
to obtain two ind
e
xes,
denote
d
by
Diff
1
and
Diff
2
resp
ectivel
y
, by using the
following formulas
:
∑∑
|
Sign
,
,
|
100%
(20)
∑∑
|
,
,
|
100%
(21)
Whe
r
e, Sign
(·) i
s
the si
gn
function.
M
and
N
re
pre
s
ent the h
e
i
ght and
widt
h of the ima
ge,
respe
c
tively (herei
nafter th
e same m
ean
ing).
For two
ra
ndom n
o
ise
image
s,
the theoretical
values of
Diff
1
and
Diff
2
are
255/25
6
≈
9
9
.6
094% and 2
1
845/65
536
≈
3
3
.3328%, re
spectively [17].
(2) Use
the prop
osed system
, encrypt the plain im
age
P
1
wit
h
se
cret
key
K
1
to get
ciph
er im
age
C
, then d
e
crypt the ima
ge
C
w
i
th sec
r
et
key
K
2
to get the d
e
crypte
d ima
ge,
denote
d
by
P
2
. C
o
mp
ar
e ima
g
e
s
P
1
a
nd
P
2
to
obt
ain two in
dex
es, d
enote
d
by
Diff
3
an
d
Diff
4
respe
c
tively,
by using the f
o
llowin
g
form
ulas:
∑∑
|
Sign
,
,
|
100%
(22)
∑∑
|
,
,
|
100%
(23)
If
P
1
takes
the plain im
age
s of Len
a, Baboon
a
nd Peppe
r (as sho
w
n in
Figure 3
a
-3
c,
respe
c
tively), and
P
2
is a rand
om
noise imag
e, the theoreti
c
al value
s
o
f
Diff
3
are all
255/25
6
≈
9
9
.6
094%, and t
he theo
retica
l values of
Diff
4
are ab
o
u
t 28.6105%,
28.2020% a
n
d
30.826
5%, re
spe
c
tively [17].
Here, ta
ke th
e plai
n ima
g
e
s
of Le
na,
Baboon
an
d
Peppe
r (as shown in
Figu
re 3
a
-3c,
respe
c
tively) as exam
ple
s
, and do th
e trials for
100 ti
mes fo
r ea
ch
image to te
st the se
cret
key
sen
s
itivities.
Then
list th
e
cal
c
ul
ated
a
v
erage
value
s
of
Diff
1
,
Dif
f
2
,
Diff
3
and
Diff
4
in Table 3
(wh
e
re, the
value
s
in
pa
re
ntheses a
r
e t
heoretical
val
ues for ea
ch
index). In
ea
ch t
r
ial, u
s
e
t
he
following formulas
to generate
the ran
dom se
cret key of
K
1
={
x
0
,
y
0
,
z
0
,
w
0
,
r
1
,
r
2
}:
x
0
=-40
+8
0×
ra
nd
(24)
y
0
=-40
+8
0×
ra
nd
(25)
z
0
=1
+8
0×
r
and
(26)
w
0
=-25
0+5
0
0
×
ra
nd
(27)
r
1
=(f
l
o
o
r
(
rand
×10
000
0)) mod 256
(28
)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 11, Novem
ber 20
14: 79
52 – 796
2
7960
r
2
=(f
l
o
o
r
(
rand
×10
000
0)) mod 256
(29
)
Whe
r
e,
ran
d
is a
MATL
AB function
use
d
to
gen
erat
e
the
0-1
unifo
rmly di
stribute
d
ran
dom
numbe
rs, and
floor(
t
) return
s the intege
r less than or e
qual to
t
.
Table 3. Re
sults of Key Sensitivity Test
s
Diff
1
(99.60
94%
)
Diff
2
(33.33
28%
)
Diff
3
(99.60
94%
)
Diff
4
Lena
99.6096%
33.3390%
99.6109%
28.6173%
(28.6
1
05%)
Baboon
99.6053%
33.3368%
99.6072%
28.1998%
(28.2
0
20%)
Pepper
99.6109%
33.3262%
99.6116%
30.8287%
(30.8
2
65%)
As can
be se
en from Ta
ble 3, the calcu
l
ated value
s
of
Diff
1
,
Diff
2
,
Diff
3
and
Diff
4
are very
clo
s
e to their
theoreti
c
al va
lues, indi
catin
g
that
the pro
posed en
cryp
tion system i
s
very sen
s
itive
to the secret
keys. Th
ese also
sho
w
th
a
t
each key in the key sp
ace is valid.
4.5. Resistin
g the Differ
e
n
tial Attac
k
NPCR (n
umb
e
r of pixels
chang
e rate
)
and UA
CI (u
nified averag
e cha
ngin
g
i
n
tensity)
are u
s
u
a
lly used to m
e
a
s
ure the
abilit
y of encry
pti
on sy
stem to
fight again
s
t the different
ial
attacks [3]. Assume th
at the plai
n ima
ges
P
1
a
nd
P
2
are id
entica
l
except that
P
2
(
i
,
j
)=(
P
1
(
i
,
j
)+
1)
mod 256 for
a certai
n coo
r
dinate (
i
,
j
). We u
s
e the prop
osed en
cryption
syst
em to encryp
t
the
plain ima
g
e
s
P
1
and
P
2
wit
h
the
same
secret key to
get two
cip
h
e
r
imag
es,
de
noted by
C
1
and
C
2
, respe
c
tively. Then the definition
s
of NP
CR and
UACI can be e
x
presse
d as f
o
llows:
NPCR
∑∑
|
Sign
,
,
|
100%
(30)
UA
C
I
∑∑
|
,
,
|
100%
(31)
For two ra
ndom noi
se
images, the theoret
ical values
of NPCR
and UA
CI are
255/25
6
≈
9
9
.6
094% and 2
5
7
/768
≈
33.46
3
5
%, resp
ectiv
e
ly [17].
Here, con
s
id
er the plain
image
s of Lean, Baboo
n
and Peppe
r all with the size of
128
×12
8
, 25
6×2
56, 3
5
7
×
317
and
512
×51
2
. Fo
r e
a
ch
imag
e,
do 1
00 tri
a
ls to calculate
the
averag
e valu
es
of NP
CR
and
UACI, a
nd the
n
li
st
the results i
n
Table
4. Note that a
ra
nd
om
se
cret key
is gene
rated
by Equation
(24)-(2
9) for e
n
crypting each image.
Table 4. Re
sults of Plaintext Sensitivity
Test
s
Plain image
size
NPCR (9
9.6094
%)
UACI (33.4
635%
)
Lena
Baboon
Pepper
Lena
Baboon
Pepper
128×128
99.6095%
99.6044%
99.6089%
33.4528%
33.4863%
33.4122%
256×256
99.6101%
99.6112%
99.6077%
33.4546%
33.4316%
33.4894%
357×317
99.6101%
99.6082%
99.6103%
33.4927%
33.4551%
33.4979%
512×512
99.6108%
99.6099%
99.6090%
33.4679%
33.4755%
33.4807%
As can
be
se
en fro
m
Ta
bl
e 4, the te
st
values
of NP
CR and
UA
CI are ve
ry cl
o
s
e to
the
theoreti
c
al va
lues
of NPCR and
UACI, in
dicatin
g
that the propo
se
d
system
ca
n fight agai
nst th
e
differential attack
s
effec
t
ively.
4.6. Resistin
g Chos
en/kn
o
w
n
Plain
t
ex
t Attac
k
s
For different
plain
im
age
s in
a ce
rtain
i
m
age en
crypt
i
on
system, if
the e
quivale
nt se
cret
keys o
r
pa
rt of the equival
ent se
cret ke
ys ar
e
kept u
n
ch
ang
ed, the attacker
ca
n crypt-analy
z
e
the equivale
n
t
keys of the encrypt
ion
system via ch
o
o
sin
g
or a
c
q
u
iring m
u
lti-p
a
irs
of plain
and
ciph
er ima
g
e
s
[
12-1
6
]
.
I
n
orde
r t
o
re
sis
t
t
he chos
en/
kno
w
n pl
aint
ex
t
at
t
a
cks,
it
is nec
es
sa
ry
t
hat
different pl
ain
image
s
co
rre
s
po
nd to
different e
quiva
le
nt se
cret keys in th
e e
n
cryption
system.
In
our p
r
op
ose
d
image en
cryption
syst
em, althoug
h the diffusi
on is plai
nte
x
t-unrel
ated,
th
e
confu
s
io
n is plaintext-rel
a
ted,
so tha
t
the different plain ima
ges
corre
s
p
ond to different
equivalent
ke
ys. Therefo
r
e
,
the propo
se
d system
ca
n
resi
st the ch
ose
n
/kn
o
wn plaintext attacks.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Chaotic S
y
stem
Based Im
age Encryp
tion Algorit
hm
using Plainte
x
t-related
…
(Yong Zha
ng)
7961
4.7. Information Entrop
y
The i
n
form
ation e
n
tro
p
y re
flects t
he
un
certaint
y of th
e
imag
e info
rm
ation. Th
e bi
g
ger the
informatio
n entropy is, the
more un
ce
rtain the
imag
e information
is, and the more uni
ntelli
gible
the image is.
For the
L
-l
evel grayscal
e image, den
ote the em
erge
nce p
r
ob
abilit
y of gray value
i
by
p
(
m
i
), and then the information entrop
y
can be
expressed a
s
the
followin
g
formula:
∑
log
(32)
For an 8
-
bit random n
o
ise grayscal
e im
age, the theo
retical valu
e of information
entropy is 8.
Her
e
,
L
=2
56.
We use the plain imag
es of
Lena,
Baboon an
d Peppe
r all of size
128
×12
8
, 25
6×2
56, 3
5
7
×
317
and
512
×51
2
a
nd th
e
i
r corre
s
po
ndi
ng
ciph
er i
m
age
s to
cal
c
u
l
ate
the informati
on entro
py value
s
, and list the result
s i
n
Table 5. Without loss of generality, the
se
cret
key
K
use
d
is fixed
on {3.3
133,
1
2
.0546, 4
0
.8
879,
-34.56
7
7
, 35, 20
1}
(same a
s
the
ke
y
use
d
in Secti
on 3).
Table 5. Re
sults of Inform
ation Entropy
Tests
Image size
Entrop
y of Lena
Entrop
y of Babo
on
Entrop
y of Pepp
er
Plain
Cipher
Plain
Cipher
Plain
Cipher
128×128
7.34907
7.99680
7.25776
7.99646
7.58134
7.99617
256×256
7.36839
7.99903
7.35535
7.99919
7.57459
7.99915
357×317
7.37914
7.99823
4.35025
7.99855
7.56647
7.99840
512×512
7.38398
7.99929
7.45159
7.99932
7.57076
7.99922
As ca
n be se
en from Fig
u
re 5, the informati
on entro
py values of
plain imag
es
are
way
different from
the theoreti
c
al value; whil
e the in
forma
t
ion entropy
values of ci
p
her ima
g
e
s
a
r
e
clo
s
e to
the
theoretical
val
ue. Th
ese d
e
m
onst
r
at
e th
at the
ciph
er
image
s
have
no
inform
ation
leakage. Con
s
eq
uently, the prop
osed system ca
n re
si
st the vario
u
s informatio
n entropy atta
cks.
5. Conclusio
n
This p
ape
r propo
se
s a ne
w plaintext-re
lat
ed imag
e encryption
system, whi
c
h i
n
clu
d
e
s
plaintext-un
re
lated imag
e diffusion a
n
d
plaintex
t-rel
a
ted imag
e confu
s
io
n. For multi-im
ag
e
s
encryption, th
e propo
sed
image e
n
crypt
i
on sy
stem
i
s
much faste
r
than the tradi
tional plai
nte
x
t-
related e
n
cry
p
tion system
on encryption sp
eed d
u
e
to no rou
nd ope
ration
s nee
ded in
its
encryption p
r
ocess. Be
cause t
he
confusi
on i
s
plaintext-rel
a
ted, differe
nt plain im
age
s
corre
s
p
ond t
o
different e
q
u
ivalent key
s
, so the
p
r
op
ose
d
sy
stem
can
re
sist th
e ch
osen/kn
o
w
n
plaintext atta
cks. Fi
nally, the
simulatio
n
re
sult
s
sho
w
that the
pro
posed
syste
m
po
sse
s
ses the
cha
r
a
c
ters of
huge
key sp
ace,
stron
g
key sen
s
itivit
y, stron
g
plai
ntext sen
s
itivity, good stati
s
tical
prop
ertie
s
of the ciph
er im
age
s, large i
n
form
atio
n e
n
tropy, etc. T
herefo
r
e, the
prop
osed im
age
encryption sy
stem can be
applie
d in act
ual com
m
uni
cation
s.
Ackn
o
w
l
e
dg
ements
This work wa
s fully suppo
rted by the Natu
ral Sci
e
n
c
e Found
ation
of Jiangxi Province
(Grant No. 20
122BAB201
0
36).
Referen
ces
[1]
F
r
idrich J. S
y
m
m
etric ciphers
base
d
on t
w
o-
dime
nsio
nal ch
aotic maps.
Int J Bifurc Chaos
. 1998; 8(6):
125
9-12
84.
[2]
Lia
n
SG, Sun
JS, W
ang Z
Q. A block c
i
p
h
e
r
base
d
o
n
a
suitab
le us
e of
the cha
o
tic st
and
ard ma
p.
Cha
o
s Solito
n
s
Fractals
. 2005
; 26(1): 117-1
2
9
.
[3]
Che
n
GR, Mao YB, Chui CK. A s
y
mmetric i
m
age e
n
cr
ypti
on schem
e ba
sed on 3
D
cha
o
tic cat maps.
Cha
o
s Solito
n
s
Fractals
. 2004
; 21(3): 749-7
6
1
.
[4]
Ye GD. Image scrambli
ng e
n
c
r
y
p
ti
on a
l
gor
ith
m
of pixe
l bit b
a
sed o
n
cha
o
s
map.
Pattern Recognition
Letters
. 201
0; 31(5): 34
7-3
5
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.