Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

AN IMAGE ENCRYPTION APPROACH USING CHAOS AND

STREAM CIPHER
1
ALIREZA JOLFAEI, 2ABDOLRASOUL MIRGHADRI
1, 2
Faculty and Research Center of Communication and Information Technology, IHU, Tehran, Iran

ABSTRACT

In this paper, a novel image encryption scheme is proposed based on combination of pixel shuffling and
W7. Due to sensitivity to initial conditions, chaotic maps have a good potential for designing dynamic
permutation map. So a chaotic Henon map is used to generate permutation matrix. An external secret key is
used to derive the initial conditions for the chaotic map and W7 secret key. Pixel shuffling is performed via
vertical and horizontal permutation. Shuffling is used to expand diffusion in the image and dissipate the
high correlation among image pixels. In order to evaluate performance, the proposed algorithm was
measured through a series of tests. Experimental results illustrate that the scheme is highly key sensitive
and shows a good resistance against brute-force and statistical attacks.
Keywords: Shuffling, W7, Henon Map, Permutation Matrix, Test.

1. INTRODUCTION characteristics of chaotic functions and the LFSR-

based stream ciphers. According to key's large
Along with the fast progression of data exchange space in the chaotic functions, this method is very
in electronic way, it is important to protect the robust. Finally, this algorithm is very sensitive to
confidentiality of data from unauthorized access. small changes in key so even with the knowledge of
Security breaches may affect user's privacy and the key approximate values; there is no possibility
reputation. So, data encryption is widely used to for the attacker to break the cipher.
confirm security in open networks such as the
The rest of the paper is organized as follows: In
internet. Due to the substantial increase in digital
Section 2, we describe the relation between chaos
data transmission via internet, the security of digital
and cryptography. In Section 3, we briefly
images has become more prominent and attracted
introduce the methods used in this paper. The new
much attention in the digital world today. Also, the
algorithm is proposed in section 4. In section 5, we
extension of multimedia technology in our society
test the efficiency of algorithm by visual inspection.
has promoted digital images to play a more
In section 6, we analyze the security of the
significant role than the traditional texts, which
proposed image cipher and evaluate its performance
demand serious protection of users' privacy for all
through various statistical analysis, key sensitivity
applications. Each type of data has its own features;
analysis, differential analysis, key space analysis,
therefore, different techniques should be used to
etc and compare the results. Finally, some
protect confidential image data from unauthorized
conclusions are given in section 7.
access. Most of the available encryption algorithms
are used for text data. However, due to large data 2. CHAOS AND CRYPTOGRAPHY
size and real time requirement, the algorithms that
are appropriate for textual data may not be suitable Chaos is a phenomenon that occurs in nonlinear
for multimedia data. definable systems sensitive to initial conditions and
has a pseudo-random behavior. Dynamic chaotic
Classical cryptographic algorithms such as RSA,
systems in case of Liapunov exponential equations
DES, and ... are inefficient for image encryption
meet will remain stable in chaos mode. An
due to image inherent features, especially high
important characteristic that has caused this
volume image data. Many researchers proposed
phenomenon to take into consideration for many
different image encryption schemes to overcome
cryptographic systems is being definable despite of
image encryption problems [1, 2, 3, 4]. In this
its pseudo-random behavior. Due to pseudo-random
research we have tried to find a simple, fast and
behavior, the output of the vision system seems
secure algorithm for image encryption using the

117
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

random in attackers' view, while in receiver's view, The Henon map possesses a strange attractor. For
the system can be defined and decryption is any values of (xi, yi) in S, the sequence of points
possible. Many cryptographic algorithms based on quickly converges to this attractor and remains on it
chaos theory are presented till now [5, 6, 7] and during the iterations that follow.
some of them are somehow employed in way that
3.2. Stream Cipher
are capable of image encryption addition to text
encryption. Image encryption must have special In cryptography, Stream ciphers are an important
features such as suitable speed for image massive class of symmetric encryption algorithms [10].
data ciphering. Text encryption methods are not They encrypt binary digits of a plain-image one at a
suitable for implementation on the image. time, using an encryption transformation which
Practically, we need to transmit a reasonable varies with time. A stream cipher is a symmetric
amount of information, which requires a large key cipher where plain-image bits are combined
sample space and that in turn implies a large with a keystream, typically by XOR operation.
number of keys. The distribution of a large number
of keys is liable to cause horrendous management 3.2.1. W7 Stream Cipher
problems. So, one of the main advantages of The W7 stream cipher is a synchronous
chaotic system's realization is facilitated key symmetric encryption designed for efficient
management approach because this method only hardware implementation at very high data rates
needs to protect and secure transmission of secret [11]. This cipher has been proposed in order to
key (parameters and initial values of chaotic replace A5/1 in GSM security scheme, due to the
system), which has a little volume and therefore not security problems of A5/1 [12]. W7 algorithm
only a little memory is needed to maintain it but supports key lengths of 128-bit and consists of a
also there is more confidence during its transfer. control and a function unit. The function unit is
The unauthorized access to short length keys is responsible for the keystream generation and
significantly less possible than the large length keys contains eight similar cells. Each cell contains three
during data transmission through the insecure LFSRs and one majority function. Fig. 1 shows W7
channel. key stream generator and the detailed design of the
block C2 is shown in Fig. 2. Compared with A5/1,
3. METHODS the key size is increased from 64 bits to 128 bits;
each LFSR sequence is filtered by a cubic Boolean
We used Henon map and W7 stream cipher for
function, 8 parallel identical structures for
constructing our new approach that are described as
outputting one byte instead of one bit. W7 is fast
follows:
and has a cheap and efficient hardware
3.1. Henon Map implementation, so the idea to use W7 for image
encryption purposes might be tempting. This is
The Henon map is a prototypical two-
additionally motivated by the fact that using
dimensional invertible iterated map represented by
existing and tested ciphers instead of designing new
the state equations with a chaotic attractor and is a
ones is advised from a security’s viewpoint.
simplified model of the Poincare map for the
Lorenz equation proposed by Henon in 1976 [8]. 4. THE PROPOSED ALGORITHM
The chaotic Henon mapping has been proposed as a
method of generating pseudo-random sequences In this section, we propose a new image
[9]. The two-dimensional Henon map is defined as encryption scheme that consists of a pixel shuffler
follows: unit and a stream cipher unit. Fig. 3 demonstrates
the block diagram of the proposed algorithm. So
⎧⎪x 2
n +1 = 1 + y n − α x n far, many researchers suggested using combination
⎨ (1) of Pixel scrambling and symmetric encryption [2, 3,
⎪⎩ y n +1 = β x n ,
13, 14]. Pixel scrambling has two important issues
with initial point (x0, y0). The pair (x, y) is the two- that are useful for image ciphering. It not only
dimensional state of the system. When α = 1.4 rearranges the pixel location (diffusion), but also
and β = 0.3 , the system is in chaotic state. Henon changes the value of each pixel (confusion).
Creating confusion in the image before applying the
showed that if the initial point lies in the area S
stream cipher is redundant and not only adds no
defined by the four points (-1.33, 0.42), (1.32,
enhancing property to the system but also increases
0.133), (1.245, -0.14) and (-1.06, -0.5), then the
the computational complexity. Confusion is
subsequent points, (xi, yi) for i ≥ 1 , also lie in S [8].

118
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

performed by stream cipher itself through nonlinear image histogram. This means that the
function operation. Pixel location displacement is corresponding statistical information in shuffled
appropriate before applying encryption, because image is as same as the plain-image.
unlike the text data has only two neighbours, each
According to Shannon information theory, secure
pixel in the image is in neighbourhood with eight
encryption systems provide some circumstances on
adjacent pixels. For this reason, each pixel has a lot
information entropy [15]. So that:
of correlation with its adjacent neighbours.
However, it is very important to disturb the high H (P | C ) = H (P ), (2)
correlation among image pixels to increase the
security level of the encrypted images. In order to where C is cipher-image and P is plain-image. In
dissipate the high correlation among pixels, we order to satisfy equation (2), C should not give any
proposed shuffling operation. Pixel shuffler unit information about P. To fulfil this desire cipher-
consists of a permutation map that is applied in two image must be as random as possible. Since a plain-
different directions: vertical and horizontal, to image that has history of uncertainty is distributed
decrease adjacent pixels correlation. A permutation non-uniformly, an ideal cipher-image histogram has
matrix is an identity matrix with the rows and to approximate the uniform balanced distribution.
columns interchanged. It has a single 1 in each row Also each two adjacent pixels should be statistically
and column; all the other elements are 0. The non-correlated. This purpose cannot be achieved
inverse of a permutation matrix is the same as its under a limited number of permutations. Permuted
image cannot resist against statistical and known
transpose, P −1 = PT . So, no extra calculation is
plaintext attacks [16, 17]. So after pixel permutation
needed to compute the reciprocal matrix for
in two directions, the W7 algorithm is deployed. In
decryption. This is a valuable property for
gray scale image, each pixel has a value between 0-
cryptographic purposes that increases algorithm
speed and decreases memory usage. 255. P = {P (i , j ) : i , j = 0,1, 2,..., 255} , where P is
set of pixels and (i , j ) denotes pixel location.
According to shuffling algorithm, a two-
dimensional Henon map is employed as a pseudo- Hence, each pixel's length is 8 bits in GF (2) . Thus,
random number generator to build permutation image matrix is a binary sequence of 8×H×W
matrix. Let's suppose that the plain-image is an length, where H is number of rows and W is number
N×N matrix. The shuffling algorithm is described as of columns. W7 algorithm generates a
follows: pseudorandom cipher bit stream known as
keystream, equal length to shuffled image binary
NoIt ← N sequence. Then these two bit streams are XOR-ed.
For it = 1: NoIt
Shuffled − image ⊕ Keystream = Cipher− image . (3)
(xn+1, yn+1) ← Henon (xn, yn)
V (it) ← xn+iyn For decryption, cipher-image is XOR-ed with
keystream then reverse shuffling operation is
performed via inverse permutation map, i.e.
D (it) ← x n2 + y n2
P −1 = PT .
End For
5. VISUAL TESTING
Pmap ← Permutation map that is generated from D,
Pmap elements ∈ {0, 1}. The algorithm is applied with a 256-level gray
scale TIF image that has the size of 256×256 and
P ← Plain-image visual test is performed. Plain-image is encrypted
For it = 1: NoIt by using the W7 secret key '00001000000008000
100002000010000' (in hexadecimal) and (x0, y0) =
Vertical permutation ← Pmap × P (:, it) (0.4, 0.3) as the initial point of chaotic map. Fig. 5
Horizontal permutation ← P (it, :) × Pmap demonstrates encryption result. By comparing the
plain and cipher-image in Fig. 5, there is no visual
End For information observed in the encrypted image.
Fig.4 shows Shuffling in two directions with
(x0, y0) = (0.4, 0.3) as seed point. It is seen that, as
we expected, the shuffling doesn’t affect the plain-

119
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

6. SECURITY ANALYSIS uniform. Relatively uniform distribution in cipher-

image histogram points out good quality of method.
A good encryption procedure should be robust
6.3. Information Entropy
against all kinds of cryptanalytic, brute-force and
statistical attacks. In this section, the performance Information theory is the mathematical theory of
of the proposed image encryption scheme is data communication and storage founded in 1949
analyzed in detail. by Shannon [20]. Information entropy is defined to
express the degree of uncertainties in the system. It
6.1. Key Space Analysis is well known that the entropy H(m) of a message
It is well known that a large key space is very source m can be calculated as:
important for an encryption algorithm to repel the
2 N −1 1
brute-force attack. Since the algorithm has a 128-bit
128
H (m ) = ∑ i =0 P (m i ) log 2
P (m i )
, (5)
key, the key space size is 2 . Furthermore, if we
consider the two seed points of shuffler as part of where P ( m i ) represents the probability of symbol
the key, the key space size will be even larger. This
m i and the entropy is expressed in bits. Let us
means if the precision is 10−14 , then the key space
suppose that the source emits 28 symbols with
can be 2128 ×1014×2 2221 3.4 ×1066 . Apparently, equal probability, i.e., m = {m1 , m 2 ,..., m 28 } . Truly
the key space is large enough to resist all kinds of
brute-force attacks. random source entropy is equal to 8. Actually,
given that a practical information source seldom
6.2. Histogram Analysis generates random messages, in general its entropy
To prevent the leakage of information to value is smaller than the ideal one. However, when
attackers, it is important to ensure that encrypted the messages are encrypted, their entropy should
and original images do not have any statistical ideally be 8. If the output of such a cipher emits
similarities. The histogram analysis clarifies how symbols with entropy less than 8, there exists
pixels in an image are distributed by plotting the certain degree of predictability, which threatens its
number of pixels at each intensity level. Fig.5 security. Let us consider the cipher-image in Fig.
shows histogram analysis on test image using 5(b), the number of occurrence of each gray level is
proposed algorithm. The histogram of original recorded and the probability of occurrence is
image contains large sharp rises followed by sharp computed. The entropy is as follows:
declines as shown in Fig. 5(c) and the histogram of 1
∑ i =0 P (m i ) log2 P (m i ) = 7.9904 ≈ 8. (6)
255
the encrypted image as shown in Fig. 5(d) has H (m ) =
uniform distribution which is significantly different
from original image and has no statistical similarity The value obtained is very close to the theoretical
in appearance. Therefore, the proposed algorithm value of 8. This means that information leakage in
does not provide any clue for statistical attack. The the encryption process is negligible and the
encrypted image histogram, approximated by a encryption system is secure upon the entropy
uniform distribution, is quite different from plain- attack.
image histogram. The uniformity caused by the
proposed encryption scheme is justified by the chi- 6.4. Correlation Coefficient Analysis
square test [18, 19] as follows: There is a very good correlation between
256 adjacent pixels in the image data [21]. Equation (7)
(v k − 256) 2
χ2 = ∑ 256
, (4) is used to study the correlation between two
adjacent pixels in horizontal, vertical and diagonal
k =1
orientations.
where k is the number of gray levels (256), vk is the
∑ j =1(x j × y j ) − ∑ j =1x j × ∑ j =1 y j
N N N
observed occurrence frequencies of each gray level N (7)
(0–255), and the expected occurrence frequency of Cr = ,
(N ∑ x 2j − (∑ x j )2 ) × (N ∑ y 2j − (∑ y j )2 )
N N N N
each gray level is 256. With a significance level of j =1 j =1 j =1 j =1
2
0.05, it is found that χtest < χ 2 (255, 0.05) ,
where x and y are intensity values of two
implying that the null hypothesis is not rejected and neighbouring pixels in the image and N is the
the distribution of the encrypted histogram is number of adjacent pixels selected from the image
to calculate the correlation. 1000 pairs of two

120
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

adjacent pixels are selected randomly from image to Another measure, UACI, is defined by the
test correlation. Correlation test image is depicted following formula:
in Fig. 5(a). Fig. 6 shows the correlation 1 | C1 (i , j ) − C 2 (i , j ) |
distribution of two adjacent pixels in the plain- UACI =
W ×H
× ∑i , j [ 255
] × 100%. (11)
image and cipher-image. It is observed that
neighbouring pixels in the plain-image are Tests have been performed on the proposed
correlated too much, while there is a little scheme on a 256-level gray scale image of size
correlation between neighbouring pixels in the 256×256 shown in Fig. 5(a). The MAE experiment
encrypted image. Results for correlation result is shown in table 2. It is illustrated that there
coefficients are shown in table 1. is a fluctuation between MAE of row and column-
row permuted image. The MAE of Column-row
6.5. Differential analysis permuted image is about 9 percent more than MAE
In general, a desirable property for an encrypted of row permuted image while the MAE of
image is being sensitive to the small changes in encrypted image is 21 percent more than MAE of
plain-image (e.g., modifying only one pixel). shuffled image. The larger the MAE value, the
Opponent can create a small change in the input better the encryption security. The NPCR and
image to observe changes in the result. By this UACI test results are shown in table 3. Results
method, the meaningful relationship between obtained from NPCR show that the encryption
original image and encrypted image can be found. scheme's sensitivity to small changes in the input
If one small change in the plain-image can cause a image is under 0.01%. The UACI estimation result
significant change in the cipher-image, with respect shows that the rate influence due to one pixel
to diffusion and confusion, then the differential change is very low. The results demonstrate that a
attack actually loses its efficiency and becomes swiftly change in the original image will result in a
practically useless. Three common measures were negligible change in the ciphered image. The reason
used for differential analysis: MAE, NPCR and lies in the mode of operation. The function of
UACI [22, 23]. MAE is mean absolute error. NPCR shuffler and W7 operates in counter mode and does
means the number of pixels change rate of ciphered not mangle plaintext in a complicated way.
image while one pixel of plain-image is changed. 6.6. Sensitivity analysis
UACI which is the unified average changing
intensity, measures the average intensity of the An ideal image encryption procedure should be
differences between the plain-image and ciphered sensitive with the secret key. It means that the
image. change of a single bit in the secret key should
produce a completely different cipher-image. Fig. 7
Let C (i, j) and P (i, j) be the gray level of the shows key sensitivity test result. It can be observed
pixels at the ith row and jth column of a W×H that the decryption with a slightly different key
cipher and plain-image, respectively. The MAE (different secret key or initial values) fails
between these two images is defined in completely. Therefore, the proposed image
W H encryption scheme is highly key sensitive.
1
MAE =
W ×H ∑∑ | C (i , j ) − P (i , j ) |. (8)
6.7. Image Sequence Randomness Test
j =1 i =1
To ensure the security of a cryptosystem the
Consider two cipher-images, C1 and C2, whose cipher must have some properties such as good
corresponding plain-images have only one pixel distribution, long period, high complexity and
difference. The NPCR of these two images is efficiency. In particular, the outputs of a
defined in cryptosystem must be unpredictable in the absence

NPCR =
∑ i , j D (i , j ) ×100%, (9)
of knowledge of the inputs. Recently, the NIST
designed a set of different statistical tests to justify
W ×H randomness of binary sequences produced by either
hardware or software based cryptographic random
where D (i , j ) is defined as or pseudorandom number generators. These tests
focus on a variety of different types of non-
⎧0, if C1 (i , j ) = C 2 (i , j ),
⎪ randomness that could exist in a sequence. The
D (i , j ) = ⎨ (10)
mathematical description of each test can be found
⎪⎩1, if C1 (i , j ) ≠ C 2 (i , j ).
at [24]. So, we used the NIST test suite in order to
test the randomness of the surveyed algorithms. In

121
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

all tests if the computed P-value is < 0.01, then REFRENCES:

conclude that the sequence is non-random.
Otherwise, conclude that the sequence is random. [1] M. Sharma and M.K. Kowar, “Image
Fig. 8 illustrates that the image sequence Encryption Techniques Using Chaotic
encrypted by the proposed scheme has no defect Schemes: a Review,” International Journal of
and passes all the statistical tests with high P- Engineering Science and Technology, vol. 2,
values. no. 6, 2010, pp. 2359–2363.
[2] A. Jolfaei and A. Mirghadri, “An Applied
Imagery Encryption Algorithm Based on
Shuffling and Baker's Map,” Proceedings of
7. CONCLUSION
the 2010 International Conference on Artificial
Intelligence and Pattern Recognition (AIPR-
In this paper, an encryption scheme based on
10), Florida, USA, 2010, pp. 279–285.
combination of a chaotic map and stream cipher is
[3] A. Jolfaei and A. Mirghadri, “A Novel Image
presented. The proposed encryption system
Encryption Scheme Using Pixel Shuffler and
included two major parts, chaotic pixels shuffling
A5/1,” Proceedings of The 2010 International
and W7. A chaotic map is used to generate a
Conference on Artificial Intelligence and
permutation matrix in two orientations to build
Computational Intelligence (AICI10), Sanya,
shuffler. Henon map is a good candidate for
China, 2010.
permutation matrix generation. Permutation matrix
[4] L. Xiangdong, Z. Junxing, Z. Jinhai, and H.
has a desirable property that its inverse is known.
Xiqin, “Image Scrambling Algorithm Based on
Thus, much memory is saved in decryption process.
Chaos Theory and Sorting Transformation,”
All parts of the proposed chaotic encryption system
IJCSNS International Journal of Computer
were simulated using MATLAB. Shuffling expands
Science and Network Security, vol. 8, no. 1,
diffusion property and decreases vertical, horizontal
2008, pp. 64–68.
and diagonal correlation of two adjacent pixels. The
[5] Z.H. Guan, F. Huang, and W. Guan, “Chaos-
number of occurrence for each grey level in the
Based Image Encryption Algorithm,” Physics
image is not changed after pixel shuffling. So
Letters A, vol. 346, 2005, pp. 153– 157.
shuffled image histogram is the same as plain-
[6] M. Suneel, “Cryptographic Pseudo-random
image histogram. The proposed schemes key space
Sequences from the Chaotic Henon Map,”
is large enough to resist all kinds of brute-force
Sadhana, vol. 34, no. 5, 2009, pp. 689–701.
attacks. Theoretical and experimental results
[7] V. Patidar, N.K. Pareek, G. Purohit, and K.K.
indicate that the cipher-image histogram of the
Sud, “Modified Substitution–Diffusion Image
proposed scheme is so even that the entropy
Cipher Using Chaotic Standard and Logistic
measured is almost equal to the ideal value. The
Maps,” Commun Nonlinear SciNumer Simulat,
uniformity of cipher-image histogram was justified
vol. 15, 2010, pp. 2755–2765.
by the chi-square test. Research results show that
[8] M. Henon, “A Two-Dimensional Mapping with
the proposed scheme has resistance to statistical
a Strange Attractor,” Communication in
attacks. The differential analysis results illustrate
Mathematical physics, vol. 50, 1976, pp. 69–
that a small change in the original image will result
77.
in a negligible change in the ciphered image. The
[9] R. Forre, “The Henon Attractor as Key Stream
MAE experiment result showed that the MAE value
Generator,” Abstracts of Eurocrypt 91, 1991,
increases through each level of algorithm.
pp. 76–80.
Sensitivity analysis shows that the proposed
[10] R.A. Rueppel, “Analysis and Design of Stream
encryption scheme is sensitive to the key; a small
Ciphers,” Springer, 1986.
change of the key will generate a complete different
[11] S. Thomas, D. Anthony, T. Berson, and G.
decryption result and cannot get the correct plain-
Gong, “The W7 Stream Cipher Algorithm,”
image. Randomness test results showed that the
Internet Draft, April 2002.
new scheme has no defect and passes all the
[12] Alex Biryukov, Adi Shamir, David Wagner,
statistical tests with high P-values. According to
“Real time cryptanalysis of A5/1 on a PC,”
latter discussions, it seems that the proposed
FSE'2000, LNCS 1978, Springer, 2001, pp. 1–
encryption scheme can be a potential candidate for
18.
image encryption.
[13] A.S. Alghamdi, H. Ullah, M. Mahmud, and
M.K. Khan, “Bio-Chaotic Stream Cipher-
Based Iris Image Encryption,” Proceedings of

122
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

the International Conference on Computational AUTHOR PROFILES:

Science and Engineering, 2009, pp. 739–744.
Alireza Jolfaei received the Bachelor's degree in
[14] X.Y. Yu, J. Zhang, H.E. Ren, G.S. Xu1, and
Biomedical Engineering in the
X.Y. Luo, “Chaotic Image Scrambling
field of Bio-electric with the
Algorithm Based on S-DES,” Journal of
honor degree from Islamic Azad
Physics: Conference Series, vol. 48, 2006, pp.
University, Science and
349–353.
Research branch, Tehran, Iran in
[15] D.R. Stinson, Cryptography theory and
2007 and Master's degree in
practice. Chapman and Hall/CRC, pp. 45–70.
Telecommunication in the field
[16] J.K. Jan and Y.M. Tseng, “On the Security of
of Cryptography with the honor
Image Encryption Method,” Information
degree from IHU, Tehran, Iran in 2010. He was a
Processing Letters, vol. 60, no. 5, 1996, pp.
chosen student in the first meeting of honor
261–265.
students of Islamic Azad University, Science and
[17] S. Li, C. Li, G. Chen, N.G. Bourbakis, and K.T.
Research Branch in 2005. Currently, he is a TA at
Lo, “A General Quantitative Cryptanalysis of
the faculty and research center of communication
Permutation-only Multimedia Ciphers against
and information technology, IHU, Tehran, Iran. His
Plaintext Attacks,” Signal Processing: Image
research interest includes: Cryptography,
Communication, vol. 23, no. 3, 2008, pp. 212–
Information Systems Security, Network Security,
223.
Image Processing and Electrophysiology.
[18] P. L'ecuyer and R. Simard, “TestU01: A C
Library for Empirical Testing of Random
Number Generators,” ACM Transactions on Abdolrasoul Mirghadri received the B.Sc., M.Sc.
Mathematical Software, vol. 33, no. 4, Article and PHD degrees in
22, 2007. Mathematical Statistics, from
[19] A. Jolfaei and A. Mirghadri, “Survey: Image the faculty of Science, Shiraz
Encryption Using A5/1 and W7,” Journal Of University in 1986, 1989 and
Computing, vol. 2, no. 8, 2010, ISSN 2151– 2001, respectively. He is an
9617. assistant professor at the faculty
[20] C.E. Shannon, “Communication Theory of and research center of
Secrecy Systems,” Bell Syst Tech J, vol. 28, communication and information
1949, pp. 656–715. technology, IHU, Tehran, Iran since 1989. His
[21] H.H. Ahmed, H.M. Kalash, and O.S. Farag research interest includes: Cryptography, Statistics
Allah, “Encryption Quality Analysis of RC5 and Stochastic Processes. He is a member of ISC,
Block Cipher Algorithm for Digital Images,” ISS and IMS.
Journal of Optical Engineering, vol. 45, 2006.
[22] A.N. Pisarchik and M. Zanin, “Image
Encryption with Chaotically Coupled Chaotic
Maps,” Physica D, vol. 237, no. 20, 2008, pp.
2638–2648.
[23] G. Chen, Y. Mao, and C. Chui, “A Symmetric
Image Encryption Scheme Based on 3d Chaotic
Cat Maps,” Chaos, Solitons & Fractals, vol.
12, 2004, pp. 749–761.
[24] A. Rukhin et al., “A Statistical Test Suite for
Random and Pseudorandom Number
Generators for Cryptographic Applications,”
NIST Special Publication 800-22, Revision 1a,
2010.

123
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

(e) (f)
Figure 1. W7 key stream generator. Figure 4. Shuffling using Henon map: (a) plain-
image, (b) plain-image histogram, (c) horizontal
permutation, (d) vertical and horizontal
permutation, (e) horizontal permutated image
histogram, (d) histogram of permuted image in
vertical and horizontal directions.

Figure 2. Diagram of C2 Block in W7 stream

cipher. (a) (b)

(c) (d)
Figure 5. Visual test result: (a) plain-image, (b)
encrypted image, (c) plain-image histogram, (d)
Figure 3. Block diagram of the proposed scheme. cipher-image histogram.

(a) (b)

(a) (b)

(c) (d)

(c) (d)

124
 Journal of Theoretical and Applied Information Technology

© 2005 - 2010 JATIT & LLS. All rights reserved.

www.jatit.org

Table 1. Correlation coefficients of two

adjacent pixels in the proposed method.
Correlation Coefficient Analysis
Adjacent pixels orientation
Image
Vertical Horizontal
Plain-image 0.9924 0.9976
Cipher-image 0.0038 0.0096
(e) (f)
Figure 6. Correlation of two adjacent pixels: (a) Table 2. A comparison of MAE of methods
distribution of two vertically adjacent pixels in used in the proposed scheme.
the plain-image, (b) distribution of two vertically Method MAE
adjacent pixels in the cipher-image, (c) Row permuted image 56.0851
distribution of two horizontally adjacent pixels Column-row permuted image 61.4098
in the plain-image, (d) distribution of two Encrypted image after shuffling 74.7693
horizontally adjacent pixels in the cipher-image,
(e) distribution of two diagonally adjacent pixels Table 3. NPCR and UACI of proposed method.
in the plain-image, (d) distribution of two NPCR UACI
diagonally adjacent pixels in the cipher-image. 0.0015 % 0.0005%

(a) (b)

(c) (d)
Figure 7. Key sensitivity test: (a) plain-image, (b)
plain-image histogram, (c) Decrypted image

with wrong key, (d) Decrypted image with

wrong key histogram.
Figure 8. Results of NIST tests for the image
sequence encrypted by the proposed scheme.

125