-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 9
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
Multi-Image Encryption Technique Based on
Permutation of Chaotic System Noora Shihab Ahmed
Collage of Science, Department of Computer Science,
Halabja University
Halabja, Kurdistan, Iraq
[email protected]
Abstract-- The key stream generator is the key design issue of
an encryption system. In this work presents an n-array three
key
stream generators (KSI, KSII and KSIII), Based on
permutation
of three chaotic maps (Logistic map, Kent map and tent map).
This work reviews of some image encryption algorithm and
finally investigate three methods for image encryption.
First
algorithm is encryption of original image by using KSI
system.
Second algorithm relies on KSII system to randomly generate
two sequences of numbers by means of selecting proper
factors
along with seed value. Afterwards, the randomly produced
numbers are employed to permute the image by means of
shuffling its rows, columns and pixels sequentially in a
manner
by which first sequence is utilized to shuffle rows while
second
sequence is utilized to shuffle columns. Afterwards masking
process is achieved by means of basic XOR processes between
neighboring rows and columns. This technique employs the
values of the two sequences together to shuffle the pixels.
Third
algorithm relies on KSIII system when the same procedures
returned on image encryption. The results shows that the
proposed algorithm has a high security, speed and gives
perfect
reconstruction of the decrypted image
Index Term-- Chaotic maps, Image encryption, Key stream
generation, Security.
1. INTRODUCTION In the last decays, the very quick evolution in
the dispatch of
digitized images through the World Wide Web and wireless
networks came as a sequence to the interesting progress in
network communication and the processing of digital images
as well. Protecting the data of the transmitted images from
any
sort of unauthorized access takes the much concern of those
who seek for security systems characterized by being trust
worthy, rapid and robust in order to safe keep and send
important confidential database images like military images,
online personal photo album, medical-purposes images, video
conferenceetc. there are substantial characteristics of
digital
images among those, huge information capacity and intense
correlation amongst neighboring pixels. Consequently most
common ciphering techniques such as AES (Advanced
Encryption Standard), IDEA (International Data Encryption
Algorithm), DES (Data Encryption Standard) and so on are
not descent for encrypting digital images in traditional
ways
because of the lack of low-level effectiveness when
ciphering
images [1].
Chaotic algorithms have proven effectiveness and enhanced
performance in image encryption [2-3]. Chaotic algorithms
possess attributes such as ergodicity, responsiveness to
initial
states, pseudo-random action in addition to control factors.
Which are similar to those of cryptography such as diffusion
and confusion as well. Due to its attributes, chaotic
systems
have become a prospective selection when establishing
cryptographical systems. In general, any chaotic schema to
encrypt images involves two phases, one for permutation and
another for diffusion. The reason behind permutation
operation is to lessen the correlation among pixels of an
image. In the other hand, the aim of diffusion process is to
alter gray values of a pixel consecutively with diffusion
actions. Therefor a very little alteration for any pixel can
extend to nearly whole pixels all-over the image. A proper
permutation operation should come up with superior shuffling
result. Moreover, a proper diffusion operation should make
significant amendments over the encrypted image no matter
whether being minor alteration for only single pixel in the
encrypted image.
In 1989 Mathews was the first who employed chaotic system
to construct a cryptographical algorithm [4]. The literature
has
suggested a large number of schemas to encrypt digital
images
based on chaos. The one-dimensional chaotic system in
addition to two-dimensional have emerged amongst those
schemas to encrypt image. The commonly used one-
dimensional and two-dimensional chaotic systems due to
being simple are Arnold map, Logistic map, standard map,
skew tent map and baker map [5-6]. Recently, several image
encryption techniques which are based on chaotic algorithms
show lack of security and are prone to penetration because
their key spaces and not sufficiently large [7-8] to make
them
hold out against cipher attack such as brute force attack in
contrast to proper encryption schemas which are very
responsive to cipher keys;
it is necessary that both operations, permutation and
diffusion
be characterized with perfect statistical attributes to defeat
any
kind of onslaught such as differential cryptanalysis attack,
chosen-plaintext attack (CPA), known-plaintext attack
(KPA)etcetera. To overpower obstacles like little key space
in addition to lack of security in the construction of both
processes, permutation and diffusion in chaotic systems of
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 10
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
both one and two dimensions, numerous efforts have been
made to enhance cryptosystems based on chaos with taking
into consideration larger key spaces in addition to perfect
diffusion methods. Among these efforts, an algorithm
suggested by Behnia et al. aims to improve security known as
of chaotic encryption based on piecewise nonlinear algorithm
[9]. Lately, Zhang et al. suggested a method to encrypt
image
on basis of skew tent algorithm and permutation-diffusion
construction [10] in which a P-box is generated with
identical
size of the original image besides shuffling whole locations
of
pixels in the image. This method utilizes various key
streams
based on the original image in the diffusion operation.
Therefore this method achieved better security in terms of
repelling chosen-plan attack (CPA).
Zhu et al. [11] suggested a novel permutation technique
concerning bit-level. This method able to perform image
confusion simultaneously. Later Liu and Wang [12] presented
a novel permutation technique concerning bit-level. This
method able to perform image diffusion and confusion
simultaneously. Later, Liu and Wang [12] enhanced the
schema reviewed in [11] in order to achieve color image
encryption. Changing the bits order of image took the much
interest of the authors. In this method all green, blue and
red
color bits of all components are mixed together. During
permutation process, the authors used PWLCM chaotic
algorithm in replacement for Arnold cat algorithm.
Additional
enhancements have been made. For instance using hyper
chaotic systems, coupled map lattice systems (CML) and
multiple chaotic systems based image encryption [13-14].
2. DYNAMIC CHAOTIC SYSTEMS
Chaos can be fined as an obvious fact that appears in nonlinear
known systems responsive to initial states along
with having pseudo random activity. In condition that
chaotic
dynamical systems encounter Lyapunov exponential function
they will continue stabilized in chaos mode. Pseudo random
conduct draw the attention of several cryptographic systems
to
this discernible fact. Pseudo random character helps to make
the clear data of a system appear random to attacker sight.
However, it appears observable to the intended recipient and
possibly be decrypted up to now a number of chaos algorithms
which based on cryptography are proposed. Actually many of
them are utilized in one way or another to encrypt image and
text as well. It is necessary for an encryption system to
have
proper speed to be able to cipher an image of enormous data.
As a matter of fact it is improper to use text encryption
techniques when encrypting an image. Practically, to transfer
a
reasonable amount of data, it demands a wide range sample.
Afterwards this implicates a vast number of keys. As a
consequence, serious management problems would be at keys
delivery stage. Therefore chaotic system came with
solutions.
Amongst its pros is the tactic of simplified key management.
For the reason that, this just demand to guard and assure
the
security of the private key transmission in which parameters
and initial states of chaotic system are included. The
private
key has moderate size. As a consequence, a little memory is
required to keep the private key as well as more insurance
is
available during key transmission. Above of that, the
unauthorized access to keys of short length is remarkably
less
probable as compared with keys of long length throughout
data transmission over the unsecured medium.
2.1 The Logistic mapping
Logistic mapping is a paradigmatic exemplification of
chaotic mapping. In spite of the fact that logistic mapping
is
one dimensional, however the control reaction is quite
perfect.
The following equation represents the logistic formula:
( ) (
) ( )
In the equation, an is denoted to the variable, also is a
denotation to system parameter whereas (0,4], an [0,1]. In case
that 1 < 3 the system takes the act of 'fixed point'. If
= 3 the system startes the transmission phase. When =
3.5699456, the system undertakes a chaotic condition. In
case
= 3.9 the starting value of an is 0.6. In logistic mapping
extent, repeating the process for 200 times with chaotic
ordered collection (sequence values) comes up with the
products shown in figure 1.
Fig. 1. Iterative Sequence Value of Logistic Mapping
2.2 Kent mapping
Another type of logistic mapping is Kent mapping. This
map is characterized by short-term anticipated and long-term
unexpected, meanwhile Kent map is very responsive to initial
state. The formula bellow shows Kent equation.
{
( ) ( ) ( )
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 11
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
Where bn is denoted to the variable, also a is a denotation
to
system parameter whereas (0,4], bn [0,1]. In case that a =
0.6 , the starting value of bn is 0.6. In logistic mapping
extent,
repeating the process for 200 times with chaotic ordered
collection (sequence values) comes up with the products
shown in figure 2.
Fig. 2. Iterative Sequence Value of Kent Mapping
2.3 The tent map
Mathematically, the tent map is real-valued formula based on
parameter, and is denoted be f. Tent map equation can be
expressed by:
{ }
The reason behind its naming is the likeness of its graph to
tent shape. By setting the parameter with values from 0 up
to 2, f charts the
( ) {
( )
(3)
Where is a positive real invariable (constant). Setting for
example the parameter =2, the outcome of the function f is
possibly be discernible as the product of the process of
bending the unit duration in twin, thereafter extending the
product duration [0,1/2] to get back the duration [0,1].
Repeating the process, each point proposes the new upcoming
locations as mentioned above, making a sequence cn.
The =2 state of tent mapping is a non-linear transmission of
bit shift mapping and state r=4 of logistic mapping as well.
Fig. 3. Orbits of unit-height tent map
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 12
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
3. THE PROPOSED ALGORITHM
The proposed multi-image encryption algorithm consists of
two stages iterative (multi-level) block permutation and
nonlinear keystream chipper
3.1 The Proposed Non Linear keystream
In the first stage, stream cipher based on permutation
chaotic maps process was depended. Three kinds of generators
processes called keystream I (KSI), keystream II (KSII)
and keystream III (KSIII) are used in proposed system in
shown in the figure 5 and its equivalent description is as
follows:
Step 1
In this step we show a method in construct a system
depending on three chaotic maps: Logistic map (eq1), Kent
map (eq2), and Tent map (eq3). The system producing m-
sequences ai , bi , and ci , keystream bit xi is generated
using
the Boolean function
xi = (ai . bi) + (bi . ci)
This means that xi = ai if bi = 1, xi = ci otherwise, For an
illustration of the system, see figure 4
Fig. 4. Chaotic system
Step 2 We take the permutations of chaotic maps for building
the
systems KSI, KSII, and KSIII
(a) KSI System
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 13
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
(b) KSII System
(c) KSIII System
Fig. 5. (a) KSI System (b) KSII System (c) KSIII System
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 14
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
3.2 Encryption process
The proposed multi-image encryption algorithm can be
summarized in the following steps:
Step 1 Image Encryption Using KSII system
This technique relies on KSII system to randomly generate
two sequences of numbers by means of selecting proper
factors along with seed value. Afterwards, the randomly
produced numbers are employed to permute the image by
means of shuffling its rows, columns and pixels sequentially
in a manner by which first sequence is utilized to shuffle
rows
while second sequence is utilized to shuffle columns.
Afterwards masking process is achieved by means of basic
XOR processes between neighboring rows and columns. This
technique employs the values of the two sequences together
to
shuffle the pixels. The overall procedure can be expressed
by
the following formula:
Cimg = Epixel (Ecolumn (Erow (plainim)))
Whereas: Erow is denoted to encryption as a consequence to
shuffle and mask the rows. Whilst Ecolumn refers to
encryption resulted from column shuffling and masking.
Epixel is denoted to the encryption of shuffling the pixels.
Step 2 Image Encryption Using KSIII system.
This technique relies on KSIII system to randomly generate
two sequences of numbers. One of the sequences is utilized
to
shuffle row while the other is utilized to shuffle column.
Same
as illustrated in technique step 1, pixel shuffling is
achieved
by employing the two sequences. After accomplishment of
row and column shuffle process, masking procedure is
achieved by doing basic XOR processes between neighboring
rows and columns. The overall process can be expressed by
the following formula:
Cimg = Epixel (Ecolumn (Erow (plainim)))
Whereas: Erow is denoted to encryption as a consequence to
shuffle the rows in addition to masking. Whilst Ecolumn
refers to encryption resulted from column shuffling and
masking. Epixel is denoted to the encryption of shuffling
the
pixels.
Flowchart of multi-image encryption algorithm is shows in
figure 6
Fig. 6. multi-image encryption process
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 15
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
4. EXPERIMENTAL RESULTS AND SECURITY ANALYSIS
Simulation results and performance analysis of the proposed
multi-image encryption scheme as provided in this section.
The original image Lina_color is shown below in part (a) of
figure 7. This RGB image is encrypted by our proposed
algorithm. The encrypted result of proposed algorithm is
shown in the part (b) of figure given below. The encryption
is
done using KSI. Part (c) of the figure is the output of
reverse
encryption using KSII. Part (d) in the same figure encrypted
using KSIII
(a) (b) (c) (d)
Fig. 7 (a). Original image of picture Lina_color (b) Encrypted
image by KSI system (c) Encrypted image by using KSII system (d)
Encrypted image by using
KSIII system
Figure 8(a) Original image and 8(b) shows histogram of original
image, 8(c) shows encryption image using the proposed KSII
system, 8(d) shows the histogram of encryption image, 8(e) shows
the Multi-image encryption using KSIII and 8(f) shows
histogram of the Multi-image encryption.
(a) (b) (c) (d)
(e) (f)
Fig. 8. (a) Original image of picture (b) Histogram of original
image (c) encryption image using KSII system (d) Histogram of
encryption image using KSII
system (e) encryption image using KSIII system (f) Histogram of
encryption image using KSIII system
From analysis of figure 8, it is clearly reflected that
histogram
of original image, encrypted image and Multi-encryption
image are entirely different. Statistical analysis of
histogram
cannot give any information about original image.
Entropy Analysis
The entropy H of a symbol source S can be calculated by
following equation.
Where p(si) represents the probability of symbol si and the
entropy is expressed in bits. If the source S emits 28
symbols
with equal probability, i.e. S = { s1, s2, . . . , s256}, then
the
result of entropy is H(S) = 8, which corresponds to a true
random source and represents the ideal value of entropy for
message source S. Information entropy of an encrypted image
can show the distribution of color value.
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 16
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
Table I
Entropy of original image and image encryption
Original
image
Level One
Encryption
Level Two
Encryption
Mulita level
encrypted image
Using KSI Using KSII
Using KSIII Row Column
Entropy 7.2718 7.2712 7.2596 7.29159 7.2915
Correlation
Table 2 shows the correlation coefficient in the horizontal
direction and vertical direction to Level 1, Level 2 and Level 3
image
encryption after testing by Visual Basic. The experimental
results show that the correlation of neighboring points are very
small.
At the same time the chart of the pixels correlation of
plaintext and cipher text plaintext shows, encryption adjacent
pixel gray has
no relevance after encryption. The secrecy of cipher text is
very good.
Table II
Correlation coefficient of Multi-Image Encryption
Level One
Encryption Level Two Encryption
Mulita level
Encryption
Using KSI Using KSII
Using KSIII Row Column
Vertical 0.9637 0.8795 0.8761 0.8389
Horizontal 0.9735 0.9882 0.8921 0.8604
Fig. 9. Correlations of Adjacent Pixels in (a) the Plain Image;
(b) the one level image encryption.
0
50
100
150
200
250
050100150200250
0
50
100
150
200
250
300
0100200300
Pix
el g
ray
valu
e lo
cati
on
(x+1
,y)
Pix
el g
ray
valu
e lo
cati
on
(x+1
,y)
a. Pixel gray value location(x,y) b. Pixel gray value
location(x,y)
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 17
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
Fig. 10. Correlations of Adjacent Pixels in (a) the second level
encryption (Row); (b) the second level encryption (Column)
Fig. 11. Correlations of Adjacent Pixels of the third level
encryption.
5. CONCLUSION
In this work, a multi-image encryption scheme based on
permutation at chaotic maps is proposed. The chaotic
systems principle has a large key space and its
implementation is quite simple. The algorithm is based on
the concept of shuffling the pixels in the image. The
experiments results and analysis show that the proposed
multi-image encryption system has a very large key space,
high sensitivity to secret keys has low correlation
coefficients close to ideal value 0, good entropy value.
REFERENCES [1] Schneier, B., Cryptography: Theory and Practice,
CRC Press,
Boca Raton, 1995.
[2] Fridrich, J., Symmetric ciphers based on two-dimensional
chaotic maps, International Journal of Bifurcation and Chaos,
1998, 8: 12591284.
[3] Chen, G. R., Mao, Y. B., Chui, C. K., A symmetric image
encryption scheme based on 3D chaotic cat maps. Chaos,
Solitons & Fractals, 2004, 21: 749-761.
[4] Matthews, R., On the derivation of a chaotic encryption
algorithm, Cryptologia, 1989, 8: 29-41.
[5] Guan, Z.-H., Huang, F., Guan, W., Chaos-based image
encryption algorithm, Physics Letters A, 2005, 346: 153-157.
[6] Kocarev, L., Chaos-based cryptography: a brief overview,
IEEE Circuits and Systems Magazine, 2001, 1: 6-21.
[7] Li, S., Zheng, X., Cryptanalysis of a chaotic image
encryption method, in: Proc. IEEE Int. Symposium on Circuits
and
Systems, vol. II, 2002, pp. 708-711.
[8] Rhouma, R., Solak, E., Belghith, S., Cryptanalysis of a new
substitution-diffusion based image cipher,Commun. Nonlinear
Sci. Numer. Simulat., 2010, 15: 1887-1892.
0
50
100
150
200
250
300
0501001502002503000
50
100
150
200
250
050100150200250
0
50
100
150
200
250
300
050100150200250300
Pix
el g
ray
valu
e lo
cati
on
(x,y
+1)
Pix
el g
ray
valu
e lo
cati
on
(x,y
+1)
a. Pixel gray value location(x,y) b. Pixel gray value
location(x,y)
Pix
el g
ray
valu
e lo
cati
on
(x+1
,y+
1)
Pixel gray value location(x,y)
-
International Journal of Video&Image Processing and Network
Security IJVIPNS-IJENS Vol:16 No:01 18
160801-3939-IJVIPNS-IJENS February 2016 IJENS
I J E N S
[9] Behnia, S., Akhshani, A., Ahadpour, S., Mahmodi H., Akhavan
A., A fast chaotic encryption scheme based on piecewise nonlinear
chaotic maps, Physics Letters A, 2007, 366: 391-396.
[10] Zhang, G. J., Liu, Q., A novel image encryption method
based on total shuffling scheme. Opt. Commun. 2011, 284:
2775-2780.
[11] Zhu, Z. L., Zhang, W., Wong, K. W., Yu H., A chaos-based
symmetric image encryption scheme using a bit-level permutation,
Information Sci., 2010, 181: 1171-1186.
[12] Liu, H., Wang, X., Color image encryption using spatial
bit-level permutation and high-dimension chaotic system, Opt.
Commun. 2011, 284: 3895-3903.
[13] Gao, T., Chen, Z., A new image encryption algorithm based
on hyper-chaos, Physics Letters A, 2008, 372: 394-400.
[14] Ye, R., Zhou W., A Chaos-based Image Encryption Scheme
Using 3D Skew Tent Map and Coupled Map Lattice, I. J.
Computer Network and Information Security, 2012, 1: 38-44.