Professional Documents
Culture Documents
ABSTRACT
The recent researches of image encryption algorithms have been more and more interested to combine
the encryption methods and the complex behavior of chaotic signals, in order to reach higher security
performance. A digital image can be represented by two dimensional arrays which have numeric value
called pixel, and can be encrypted by various combination of those pixels. There are various
encryption methods to secure image from unauthorized parties. This paper is an implementation of a
color image encryption algorithm based on a novel three dimensional chaotic system using a random
permutation technique.
1. INTRODUCTION
With the ever-increasing growth development engineers have designed image encryption
of image transmission through computer algorithms based on chaotic random permutation
networks, especially the Internet, the security of [6, 9]. Due to desirable properties of nonlinear
digital images has become more important. So it dynamical systems such as high sensitive
is necessary to encrypt image data before dependence on initial conditions and control
transmission over the network to preserve its parameter, ergodicity, unpredictability, mixing,
security and prevent unauthorized access. Image etc., which are analogous to the confusion and
encryption techniques try to convert original diffusion properties of Shannon [1], the chaos-
image to another image that is hard to understand; based encryption has suggested a new and
to keep the image confidential between users, in efficient way to deal with the intractable problem
other word, it is essential that nobody could get to of fast and highly secure image encryption [2].
know the content without a key for decryption. Our proposed work focused on the image
In recent years, plenty of color image encryption using random permutation technique.
encryption approaches have been proposed, such The current pixel is moved to calculated new
as DES, AES, RSA, etc. However, due to bulk position using chaos theory.
volume of data, high correlation among adjacent The rest of paper is organized as follows.
pixels, high redundancy and real time Section 2 presents the novel chaotic system.
requirement [4], these ciphers may not be the Section 3 details the chaotic permutation
most desired candidates for image encryption, techniques for encryption image. The simulation
particularly for fast and real-time communication results are presented in section 4 while section 5
applications [5]. is devoted to the conclusion.
In order to overcome image encryption
problems, in recent years, many scientists and
1
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
2. PROPOSED SYSTEM
𝑥̇1 = 𝑥2
{𝑥̇ 2 = −𝑥1 + 𝑥2 𝑥3 (1) Fig. 1. Novel chaotic 3D attractor
𝑥̇ 3 = 1 − 𝑥22
2.3.1. Lyapunov exponent
The Nosé-Hoover system (1) has the Lyapunov
exponentsL1 = 0.014, L2 = 0 and L3 = −0.014. By linearizing the Jacobian matrix JE round the
The system (1) is chaotic as it has a positive equilibrium point E and solving the following
Lyapunov exponent and is conservative as the equation:
sum of the Lyapunov exponents is zero. Thus, the
system (1) is volume-conserving. |λi − JE | = 0 (3)
2.2. Tang & al system Therefore, the novel chaotic system (2) has
three eigenvalues shown in figure 1.
The following 3D chaotic system which was
introduced by Tang et al. [9] in 2012 is 𝜆1 = 40.051, 𝜆2 = -5.039, 𝜆3 =0.828
considered:
2
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
(c)
Fig.3 . Sensitivity to two initial conditions
[5,- 2, 1] and [6, -1,3]
(a): 𝑥1 (b): 𝑥2 (c): 𝑥3
3
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
Step 6: Repeat step 5 until reaches M×N, the 4. SIMULATION RESULTS AND
length of the whole 1-D signal; SECURITY ANALYSIS
Step 7: Reshape the 1-D signal into the 2-D The proposed encryption algorithm is
block; implemented in MATLAB for computer
simulations. We take a color “baby” image of
Step 8: Recombine the three-color bloc and get 384x384 in size for experimental purposes.
encrypted image I’;
The Figure 5 shows the encrypted image using
the proposed method.
Original
image I
Chaotic
RGB three-color generator
separation block random
sequence
Reshape
each block
to 1-D
signal
Position pixel
permutation
Sort lx then,
I2=I1(lx)
generate the
Reshape 1-D permutation
signal to 2-D key
block
RGB three-color
combination
block
Encrypted
image
4
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
5
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
𝑁 ∑𝑁 𝑁 𝑁
𝑗=1(𝑋𝑗 × 𝑌𝑗 ) − ∑𝑗=1 𝑋𝑗 × ∑𝑗=1 𝑌𝑗
𝐶𝑟 =
2
√(𝑁 ∑𝑁 2 𝑁 𝑁 2 𝑁
𝑗=1 𝑋𝑗 − (∑𝑗=1 𝑋𝑗 ) ) × (𝑁 ∑𝑗=1 𝑌𝑗 − (∑𝑗=1 𝑌𝑗 ) )
2
Table I
Correlation coefficient corresponding to original and
decrypted image
Fig. 8. Horizontal, vertical and diagonal correlation Encrypted -0.0540 -0.0795 -0.0252
of original image
6
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
4.3. PSNR
2552
𝑃𝑆𝑁𝑅[𝑑𝐵] = 10 log10 ( ) (4)
𝐸𝑄𝑀(𝐼𝑜 ,𝐼𝑅 )
Fig. 11. Error decrypted image
where EQM is the mean square error between the
two images (𝐼𝑜 original , 𝐼𝑅 recovered). PSNR high means: Mean square error between
the original image and reconstructed image is
𝑚−1 𝑛−1 very low. It implies that the image been properly
1 restored. In the other way, the restored image
𝐸𝑄𝑀(𝐼𝑜 , 𝐼𝑅 ) = ∑ ∑(𝐼𝑜 (𝑥, 𝑦) − 𝐼𝑅 (𝑥, 𝑦))2
𝑚𝑛 quality is better; in our case, the value of PSNR is
𝑥=0 𝑦=0
as follow:
4.4. Decrypted image
PSNR (Original/Decrypted) = Inf
To recover the "Baby" image, we apply the Contrariwise, a low PSNR means: Mean
inverse of the proposed algorithm in Figure 4. square error between the original image and
The result is shown in figure 10. encrypted image is very high. It implies that the
image been correctly encrypted. In our case the
value of PSNR is as follow:
PSNR(Original/Encrypted) = 16.9466
7
International Journal of Computer Networks and Communications Security
VOL. 6, NO. 1, January 2018,
Available online at: www.ijcncs.org
E-ISSN 2308-9830 (Online) / ISSN 2410-0595 (Print)
proposed system has good security and the RC5 block cipher algorithm for digital
complexity. Then we proposed image encryption images”. Optical Engineering, 2006. 45( 10).
base on permutation. The different combination [11]. Ibrahim S I Abuhaiba and M.A.S.
of the permutations for image encryption has Hassan, Image Encryption Using Differential
been performed. It is observed that image Evolution Approach in Frequency Domain.
encryption using this technique given good Signal & Image Processing : An
results. International Journal(SIPIJ), 2011. 2, No.1:
p. 51-69.
6. REFERENCES