Professional Documents
Culture Documents
of Physics and Electronic Engineering, Harbin Normal University, Harbin 150025, China
3 Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology,
The development of the computer network makes image files transportation via net-
work become more and more convenient. This paper is concerned with the image
encryption algorithm design based on the chaotic S-box mechanism. This paper pro-
poses an Image Encryption algorithm involving both chaotic dynamic S-boxes and
DNA sequence operation(IESDNA). The contribution of this paper is three folded:
Firstly, we design an external 256-bit key to get large key space; secondly, we design
a chaotic system to scramble image pixels; thirdly, DNA sequence operations are
adopted to diffuse the scrambled image pixels. Experimental results show that our pro-
posed image encryption algorithm can meet multiple cryptographic criteria and obtain
good image encryption effect. © 2017 Author(s). All article content, except where
otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license
(http://creativecommons.org/licenses/by/4.0/). [http://dx.doi.org/10.1063/1.4994860]
I. INTRODUCTION
The fast development of network and computer technologies has significantly changed people’s
communications. Transmission over the Internet for image information has become more and more
frequent. The security of the transmitted image information, however, has a serious threat because of
the sharing and openness of the networks.1 Therefore, people are increasingly paying more attention
to the confidentiality and security of image information transmission. Some chaotic systems have
been proven to be insecure.2 Therefore, it is necessary to research new technologies to further improve
the security of image encryption.
In the block cipher system, the construction of the substitution box (S-box) is a crucial point
because it is an essential part of the block cipher system.3 Furthermore, the performance of S-box
directly affects the passwords security system. Mathematically, a n × n size of S-box is a non-linear
mapping S: {0, 1}n → {0, 1}n , here {0, 1}n represents the vector spaces of n elements from GF(2)
where GF represents Galois field. Many algorithms have been proposed to construct S-boxes using
different chaotic systems in recent years.4–9 Moreover, the application of S-boxes is becoming more
and more widespread in image encryption, and its main function is pixel substitution in images.10–14
However, some S-box-only ciphers are vulnerable to chosen plaintext attacks15 and Zhang and Xiao
presented a successful cryptanalysis of this algorithm.16
Recently, lots of good characteristics of DNA computing, such as massive parallelism, huge
storage and ultra-low power consumption have been infiltrated into the field of cryptography.17,18
Therefore, DNA cryptography is a new cryptographic resolution. In these DNA-based cryptosystems,
DNA is used as the information carrier, and the DNA sequence operations and complementary rules
are used to encrypt images.19 Liu et al. proposed a novel confusion and diffusion image encryption
a
Corresponding author: hsdtianye@126.com
algorithm that combined DNA coding and chaotic maps. They used the DNA coding scheme to
encode each pixel of the original image to four nucleotides and applied the complementary rule to
transform each nucleotide into their base pair for different times generated by Chebyshev chaotic
maps.20 SaberiKamarposhti et al. proposed a hybrid method by using DNA sequences and logistic
map for image encryptions.21 Xue et al. proposed a chaotic image encryption algorithm based on
DNA encoding rules. They firstly used a DNA encoding rule to map the original image into DNA
sequence and then encrypted the DNA sequence by utilizing chaotic systems.22 In addition, Zhang et
al. presented a DNA computing addition operation based chaotic image encryption algorithm. By the
rules of DNA coding, the algorithm converted the image information and the chaotic cipher stream
into DNA sequence.23 Zhang et al. proposed a novel image encryption algorithm by combining 1D
and 2D Logistic maps and DNA addition operation.24 Recently, a new image encryption method
based on linear nonlinear coupled map lattices is proposed. The strategy of DNA computing and
one-time-pad policy was used to generate an image encryption algorithm.25 Chai et al. present a
novel image encryption scheme combining DNA sequence operations, chaotic system and SHA 256
hash.26 In this paper, we aim at acquiring high security using DNA sequence operation to modify the
results obtained in the permutation stage.
The contributions of this work are the following:
1) We design an external 256-bit secret key to get large key space, and this key and a plaintext
are applied to initialize the new chaotic map (NC map) and the cubic map and tent map (C-T
cascade map).
2) We employ a mechanism of image block encryption through which the high correlation among
image pixels is preliminarily reduced.
3) We adopt a scheme of image permutation-diffusion encryption. The permutation process con-
cludes two parts: main permutation stage that calls the NC map and S-boxes to shuffle image
pixels; complementary permutation stage that makes IESDNA can resist differential attack
using an improved generalized two-dimensional Arnold transformation to scramble a sequence
generated by the C-T cascade map. In the diffusion process, a DNA sequence operation is
employed to change the permutated results instead of directly implementing image encryption
by overlaying a chaotic sequence produced by various chaotic maps and image pixel values.
4) We propose to use C-T cascade map rather than the random way to select DNA encoding rules
for coding the permutated results. The rest of the paper is organized as follows. In Section 2,
the system overview is introduced. Section 3 gives the related work. S-box construction and
evaluation criteria are depicted in Section 4, and the proposed image encryption scheme is
presented in Section 5. Section 6 presents the evaluation results, and Section 7 draws some
conclusions.
cipher image according to the corresponding positions of the plaintext image blocks. In summary,
Figure 1 depicts the encryption framework of the proposed IESDNA.
FIG. 2. The attractor, bifurcation diagram and Lyapunov exponent of a NC map. (a) Attractor; (b) Bifurcation diagram;
(c) Lyapunov exponents.
FIG. 3. Bifurcation diagram and Lyapunov exponent of a C-T cascade map. (a) Bifurcation diagram; (b) Lyapunov exponents.
085008-5 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
Since the cascade of a Cubic map and a Tent map expands the parameter ranges of that of the
Cubic and Tent maps, it achieves a larger key space (Initial values and system parameters as the
key); thus, the difficulty of decipher is enhanced, and the system safety is improved. Moreover, the
Lyapunov exponent of the C-T cascade system is greater than the Cubic and Tent map systems in the
chaotic region. The increase of the Lyapunov exponent will enhance the initial sensitivity of a system
so that the system dynamic property is essentially improved.
FIG. 5. Process of Line maps. (a) Left line map; (b) Right line map.
(k + 2)(k − 1)
L[2(i × N + j) − ] = A(i, M − i − j) (7)
2
For the folding algorithm, we can write,
B(i, j) = L[i × M + j] (8)
E. DNA rules
DNA is the main chemical ingredient of chromosomes. A DNA sequence has four kinds of
nucleic acid bases: A (Adenine), T (Thymine), G (Guanine), C (Cytosine), where, A and T are
complementary, G and C are complementary. If using the binary sequence 00, 0l, 10 and 11 to encode
four bases, we will obtain a total of 24 kinds of coding schemes; however, only eight kinds of encoding
schemes shown in Table III are in line with the Watson-Crick Complementary rules.36
1. DNA addition and subtraction rule
As there are eight kinds of encoding schemes, we can write eight kinds of DNA addition and
subtraction rules. Tables IV and V show the DNA addition and subtraction rules of the 1-st scheme.
Note: Other schemes can also be selected.
2. DNA XOR rule
XOR operation of the DNA sequence originates from the binary XOR operation. As there are
eight kinds of DNA encoding schemes, there are eight kinds of DNA XOR logic-operation rules.
Table VI gives the DNA XOR rule of the 1-st scheme.
Rule 1 2 3 4 5 6 7 8
00 A A T T G G C C
01 G C G C A T A T
10 C G C G T A T A
11 T T A A C C G G
085008-8 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
Addition A G C T
A A G C T
G G C T A
C C T A G
T T A G C
Subtraction A G C T
A A T C G
G G A T C
C C G A T
T T C G A
XOR A G C T
A A G C T
G G A T C
C C T A G
T T C G A
For example, if the gray value of the first pixel of an image is 76, the binary number of which is
[01001100], the corresponding DNA sequence is [CATA] using the 1-st DNA encoding rule. Likewise,
the gray value of the second pixel of the image is 168, the binary number of which is [10101000],
the DNA coding criterion 1 to obtain DNA sequence is [CCCA] using the 1-st DNA encoding rule.
Thus, a new sequence [TCGA] can be obtained by performing [CATA] XOR [CCCA] according to
Table VI.
A. S-box construction
Step 1: Define an integer array (denoted by S) of length 256.
Step 2: Substitute an initial value s0 into Eq. (2) to iterate 500 times, and begin to record from
the 501-th value, denote the obtained real value sequences by si ,i = 1,2,· · · .
Step 3: Substituted si into Eq. (15) to gain an integer Qi ranging of [0,255],
Step 4: If Qi has appeared in the array S, abandon it; otherwise, deposit it into S, when the array
is filled in, the S-box is generated.
Step 5: Arrange the integer sequence S in a 16 ×16 table to construct an initial prototype S-box.
Step 6: Call the above reversible two-dimensional map of the triangle (see Figure 5), perform p
times left map and q times right map on the prototype S-box to obtain p ×q lines, finally, fold these
results to generate p ×q S-boxes of scale 8 ×8.
085008-9 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
here, ai ∈{0,1}, (a1 , a2 , ..., an ) , (0, 0, ..., 0), wt() denotes the Hamming weight.
In fact, a reversible S-box is usually required, especially in a replacement network, the S-box
must be bijective.
2. Nonlinearity
Definition 1. Assume f (x): F2n → F2 is an n Boolean function, the nonlinearity of f (x) can take
the form,
Nf = min dH (f , l) (17)
l ∈Ln
where Ln is the set of all linear and affine functions, dH (f , l) is the Hamming distance between f
and l.
The nonlinearity denoted by the Walsh spectrum can take the form,
where, Γx and Γy are the mask values of the input and output, respectively, X is a set of all possible
input values of x, the elements of which is 2n .
The smaller the LP, the stronger the ability of the S-box for fighting against linear cryptanalysis
attacks, vise versa.
a0 = 7.999 (33)
here, t 0 , a0 are initial values of the chaotic map and the system parameter respectively.
085008-11 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
Step 2: Arrange pixels of P into a one-dimensional sequence I, with as size of M × N, from left
to right and top to bottom.
Step 3: Iterate equation (1) with t 0 and a0 . Discard previous 500 obtained values and record
the following M × N values, {t i , i = 1,2,· · · M × N}. Sort {t i } in ascending order to obtain {d i ,
i = 1,2,· · · M × N} and the corresponding indexes{ei , i = 1,2,· · · M × N}. Use these indexes to
scramble the sequence I. As a result, a scrambled sequence, I 0, is obtained.
Step 4: Pass s0 and a6 into Eqs. (2), (15) to construct an intermediate S-box. Use the left map
scheme of the Reversible two-dimensional map to scramble the intermediate S-box for m times, and
use the right map scheme to scramble it for n times. As a result, m × n S-boxes are obtained, {S k ,
k = 1,2,· · · , m × n}.
Step 5: Let k ← 1. For each sub-block Bk , k = 1,2,· · · , m × n, use a binary value
w =b7 b6 b5 b4 b3 b2 b1 b0 to represent a pixel of it. Denote i = b7 b6 b5 b4 by a binary representation
of a row index value, and convert it to a decimal value s. Denote j =b3 b2 b1 b0 by a binary represen-
tation of a column index value, and convert into a decimal value t. Replace the elements of the s +1th
row of the S-box with that of the t +1th column to obtain a sub-block Bk0 , k = 1,2,· · · , m × n.
Step 6: For each sub-block{Bk0 , k = 1,2,· · · , m × n}, use a 8-bit binary value to represent a
pixel of it. Convert the binary sub-block into a one-dimensional binary sequence, {Ak (l), k = 1,2,· · · ,
m × n,l =1,2,· · · ,8 × M/m × N/n},with the size of 8 × M/m × N/n.
Step 7: Iterate Eq. (2) with initial values x 0 , y0 , z0 , u0 , v0 , and control parameters
a1 ,a2 ,a3 ,a4 ,a5 .Discard previous 500 obtained values and record the following 4 × M/m × N/n values,
x j , yj , zj , uj , vj , j =1,2,· · · 4 × M/m × N/n. Convert these values into integers ranging from1,8 according
to Eqs. (51)–(55).
X
x0 = (ksum1 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 210 ) (34)
x ∈[1,m],y∈[1,n]
X
y0 = (ksum2 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 210 ) (35)
x ∈[1,m],y∈[1,n]
X
z0 = (ksum3 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 210 ) (36)
x ∈[1,m],y∈[1,n]
X
u0 = (ksum4 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 210 ) (37)
x ∈[1,m],y∈[1,n]
X
v0 = (ksum5 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 210 ) (38)
x ∈[1,m],y∈[1,n]
X
s0 = (ksum6 + kxor + psum + ksum + ( Bk (x, y)) mod 256)/(3 × 29 ) (39)
x ∈[1,m],y∈[1,n]
h=1 (40)
g=1 (41)
e=1 (42)
r =3 (43)
d =1 (44)
a1 = 1.9999 (45)
a2 = 1.9999 (46)
a3 = 1.9999 (47)
085008-12 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
a4 = 1.9999 (48)
a5 = 1.9999 (49)
a6 = 1.9999 (50)
B. Decryption process
The decryption algorithm is the inverse process of the encryption algorithm, and Figure 6 depicts
the decryption framework of the proposed IESDNA.
Step 1: Consider the same initial values and parameters as the encryption process. Calculate
Eqs. (51)–(55) to obtain xj0, yj0, zj0, uj0, vj0, j = 1, 2,· · · , 4 × M/m × N/n, and calculate Eq. (56) to
obtain wj . Call the Step 10 in the above encryption process to obtain{Wk0(j), k = 1, 2,· · · , m × n,
j =1, 2, · · · , 4 × M/m × N/n}, and the Steps 11, 12 to achieve {Zk0 (j), k = 1, 2, · · · , m × n, j = 1, 2,
· · · 4 × M/m × N/n}.
Step 2: Pass s0 and a6 into Eqs. (2), (15) to construct an intermediate S-box. Use the left map
scheme of the ICT to scramble the intermediate S-box for m times, and use the right map scheme to
scramble it for n times. As a result, m × n S-boxes are obtained.
Step 3: Let k ← 1.Divide the ciphertext image C into sub-blocks {Dk , k = 1, 2, · · · , m × n}, with
the size of M/m × N/n, and convert each block into a binary sequence, {Dk0 , k = 1, 2, · · · , m × n},
with the length of 8M/m × N/n.
Step 4: Encode {Dk0 , k = 1, 2,· · · m × n}, with DNA map rules selected by vi to obtain a DNA
sequence Dk with the length of 4 × M/m × N/n.
Step 5: Operate according to DNA map rules selected by {uj0, j = 1, 2,· · · , 4 × M/m × N/n}. For
the first sub-block, use the following equation to decrypt
Ak0 (j) = Wk0(j) ⊕ Dk (j) − Zk0 (j), k = 1, j = 1, 2, · · · , 4 × M/m × N/n (59)
For k = 2, 3, . . . , m × n, use the following equation to decrypt
Ak0 (j) = (Dk (j − 1) + Wk0(j)) ⊕ Dk (j) − Zk0 (j), k = 2, · · · , m × n,
(60)
j = 1, 2, · · · , 4 × M/m × N/n
“-”, “+” and ⊕ are subtraction, addition and XOR operations of the above DNA sequences.
Step 6: For each sequence Ak0 , it is decoded according to DNA map rules selected by {xj0, j = 1,
2, · · · , 4 × M/m × N/n}, to obtain a binary sequence{Ak (l)k = 1, 2,· · · m × n, l = 1, 2, · · · , 8 × M/m
× N/n}, with the length of 8M/m × N/n.
Step 7: Convert {Ak (l), k = 1, 2,· · · m × n, l = 1, 2, · · · , 8 × M/m × N/n}, into a decimal sub-block
{Bk0 ,k = 1, 2,· · · m × n}with the size of M/m × N/n.
Step 8: In Encryption process Steps 4 to achieve S-boxes, {S k , k = 1, 2, · · · , m × n}. For each
pixel value in Bk , it is an element in the S-box. The row index value of the element is m, and m -1 = d 7
d 6 d 5 d 4 is a binary representation. The column index value is n, and n -1= d 3 d 1 d 0 is also a binary
representation. Thus, v = d 7 d 6 d 5 d 4 d 3 d 2 d 1 d 0 denotes a binary value of a pixel, and convert it
into a decimal value. {Bk , k = 1, 2,· · · , m × n} can be obtained by traversing all the pixels.
Step 9: Let k ← k + 1, repeat Steps 3-8 until k = m × n.
Step 10: Construct the plaintext image I 0 with each sub-block {Bk , k = 1, 2,· · · , m × n}.
Step 11: Iterate Eq. (1) with the initial value t 0 and the control parameter a0 , discard the previous
500 obtained values and record afterward M × N values, {t i , i = 1, 2,· · · M × N}. Use t i to scramble
inversely I 0 to obtain an one-dimensional sequence I with the length of M × N.
Step 12: Convert I to achieve an image P from right to left and from bottom to top.
085008-14 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
FIG. 7. Original image and its encrypted image. (a) Original image; (b) Encrypted image.
FIG. 8. Performance distribution of S-boxes. (a) Average nonlinearities distribution; (b) Average values distribution of depen-
dent matrixes; (c) Average value distribution of BIC-nonlinearity; (d) Average value distribution of BIC-SAC; (e) Maximum
value distribution of DP; (f) Maximum value distribution of LP.
085008-15 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
B. Security analysis
1. Analysis of a key space
To be able to resist brute-force attack, a sufficiently large key space is required. In this paper, a set
of 256 bits binary numbers is selected as a key since it is convenient and practical. Theoretically, the
key space with such a large key space 2256 makes the password system resist the brute-force attack.
FIG. 9. Results of the key sensitivity. (a) Correct decrypted image; (b) Error decrypted image.
085008-16 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
Images Scale DR
1 C1 (i, j) , C2 (i, j)
D(i, j) = 0 C (i, j) = C (i, j) (61)
1 2
X
D(i, j)
i,j
DR == × 100% (62)
M ×N
here, M and N respectively denotes the width and height of an image, C 1 (i, j) and C 2 (i, j) respectively
represent the pixels at the position (i, j) of two ciphertext images which originate from the plaintext
and the one pixel changed plaintext. Table VII depicts the experimental results.
3. Histogram analysis
The histogram represents the distribution of gray level of all pixels in an image. The example
of the lena image with 256×256, of which the histogram analysis is shown in Figure 10, is used.
Figure 10(a) shows the histogram of the plaintext image, Figure 10(b) corresponds to the histogram
of the encrypted image. Contrast with the plaintext, the ciphertext image information distribution is
uniform; thus, it is difficult for an attacker to attack ciphertext by some statistical methods and the
security of the ciphertext is improved.
FIG. 10. The histograms of the plaintext image and ciphertext image. (a) The histogram of a plaintext image; (b) The histogram
of a ciphertext image.
085008-17 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
FIG. 11. Correlation of plaintext image and the ciphertext image in multiple directions. (a) Plaintext horizontal correlation; (b)
Plaintext vertical correlation; (c) Plaintext diagonal correlation; (d) Ciphertext horizontal correlation; (e) Ciphertext vertical
correlation; (f) Ciphertext diagonal correlation.
N N
here, E(x) = N1 xi , D(x) = N1 (xi − E(x))2 , x and y are two adjacent pixels in the image, γ xy is the
P P
i=1 i=1
correlation coefficient.
Figure 11 depicts the correlation of 3000 pixels randomly selected from the plaintext and
ciphertext images in three different directions including horizontal, vertical and diagonal directions.
Table VIII gives the correlation coefficient of the Lena image and its encrypted image in three direc-
tions and lists the corresponding results obtained by recent chaotic system based image encryption
algorithms.
Table VIII and Figure 11 show that IESDNA has a smaller correlation coefficient, and can better
achieve the purpose of destructing the correlation between adjacent pixels, and as a result, IESDNA
makes the ciphertext have a better random distribution characteristic.
M N
1 X X |C1 (i, j) − C2 (i, j)|
UACT = × 100% (66)
M × N i=1 j=1 255
where M and N respectively denotes the width and height of an image, C 1 (i, j) and C 2 (i, j) repre-
sent the pixels values at the position (i, j) of two ciphertext images originating from the plaintext
and the one pixel changed plaintext, respectively. if C 1 (i, j) , C 2 (i, j), then D(i, j) = 1, otherwise
D(i, j) = 0.
Table IX depicts the NPCR and UACI of the ciphertexts of different images. As can be seen, all
the NPCR values are more than 99%, and all the UACI values are above 33.33%, Table X depicts the
NPCR and UACI of the lena ciphertext obtained by different methods.10,11,14,44–46 The results show
that IESDNA is sensitive to the plaintext, and can resist the differential attack to some extent.
6. Information entropy analysis
Entropy refers to the uncertainty of information, the larger the information entropy, the greater
the uncertainty of information, vise versa. It solves the problem of quantifying information. In this
paper, the information entropy to reflect the randomness of the occurrence of the pixel gray values
in the encryption result is used.
Use m to represent the information source, and use equation (67) to calculate the information
entropy,47
2Xn −1
1
H(m) = p(mi )log2 (67)
i=0
p(m i)
here, p(mi ) denotes the probability of the occurrence of pixel gray value mi . According to
equation (67), if the information entropy H(m) = 8, the information is then completely random,
and the disorder of the information is the strongest, that is, the information entropy of the ciphertext
should be closer to 8.
Table XI lists the information entropy of ciphertexts of different images obtained by IESDNA.
From the table, we can see that information entropy of IESDNA is closer to 8; thus, the encryption
effect of it may be good.
7. Analysis of crop resistance
Images in the transmission process often encounter crop-operations from some anonymous
attackers. As a result data information may be lost and the decryption performance of recipients may
be affected. For a good image encryption algorithm, a decrypted image can restore main information
Images Entropy
Lena 7.9974
Boat 7.9973
Cameraman 7.9975
Peppers 7.9973
of the plaintext image even though the decrypted image is subjected to crop-attacks. In the security
analysis, generally, the following steps are performed to verify the anti-crop capability of a proposed
image encryption method: crop a ciphertext image such that part information of the image is lost,
and use a key to decrypt the cropped ciphertext image.26
The crop-operation on an encrypted image is used to check to the crop-resistance ability of the
proposed algorithm. As shown in Figure 12, the black rectangle area in Lena image is the crop part, all
pixels-values of which are zero. Figure 12(a) shows an encrypted image that is crop by one-sixteen.
Figure 12(b) depicts the decrypted image of (a). Figure 12(c) shows an encrypted image that is crop
by one-eighth. Figure 12(d) depicts the decrypted image of (c). Figure 12(e) shows an encrypted
image that is crop by one-fourth. Figure 12(f) depicts the decrypted image of (e). Figure 12(g) shows
an encrypted image that is crop by one-half. Figure 12(h) depicts the decrypted image of (g). The
following observations can be seen from the results: 1) when an encrypted image is crop by one-
sixteen, the corresponding decrypted image is clear with only a small amount of noise-like dots, 2)
As the crop area increases, the decrypted image becomes blurred, and 3) even though the image is
crop by one-half, the information of the plaintext image can be obtained from the decrypted image.
Table XII shows the peak signal-to-noise ratio (PSNR) of the decrypted image under different
crop areas. It can be seen that the proposed algorithm can effectively resist the crop attack.
FIG. 12. Test of cropping. (a) An one-sixteen-crop encrypted image; (b) The decrypted image of (a); (c) An one-eighth-crop
encrypted image; (d) The decrypted image of (c); (e) An one-fourth-crop encrypted image; (f) The decrypted image of (e);
(g) An one-half-crop encrypted image; (h) The decrypted image of (g).
085008-20 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
a proposed image encryption method: add salt-pepper noise and Gaussian noise of different intensities
to a ciphertext image, and use a key to decrypt the modified ciphertext image.26
The anti-noise ability of the proposed algorithm is checked by adding salt-pepper-noise and
Gaussian-noise with difference intensities to the encrypted image.
a. Salt-pepper-noise attack. In order to verify the anti-noise ability of the proposed algorithm,
various density of salt-pepper-noise is added to the Lena image. The decryption result is shown in
Figure 13. The added noise intensities are 0.001, 0.01, 0.1 and 0.2, respectively. Table XIII shows
the peak signal-to-noise ratio (PSNR) of the corresponding decrypted images in the order of 9.1841,
9.0250, 8.4001 and 7.8534, respectively. From the figures, we can see that the decrypted images
become more and more blurred with the increase of noise intensity, but the outline of the original
image can still be seen. It shows that the proposed algorithm can resist salt-pepper-noise attack to
some extent.
b. Gaussian-noise attack. In order to verify the anti-noise ability of the proposed algorithm,
the Lena image is added with several types of Gaussian white noise of various variances and zero-
mean-value. The decryption results are shown in Figure 14. The noise intensities added in the image
are 0.2, 0.3, 0.4 and 0.5 respectively. Table XIV shows the peak signal to noise ratio (PSNR) of the
corresponding decrypted images that are 9.1175, 8.7851, 8.5786 and 8.4955, respectively. From the
FIG. 13. Test of salt-pepper noise addition. (a) Noise intensity is 0.001; (b) Noise intensity is 0.01; (c) Noise intensity is 0.05;
(d) Noise intensity is 0.1.
FIG. 14. Test of Gaussian-noise addition. (a) Variance value is 0.2; (b) Variance value is 0.3; (c) Variance value is 0.4;
(d) Variance value is 0.5.
085008-21 Y. Tian and Z. Lu AIP Advances 7, 085008 (2017)
results, we can see that the decrypted images become blurred with the increase of the noise intensity,
but the outline of the original image can still be obtained. It shows that the proposed algorithm can
resist Gaussian-noise attack to some extent.
VII. CONCLUSIONS
A block encryption algorithm based on dynamic S-boxes and DNA sequence operation is pro-
posed in this paper. It first uses a new chaotic, NC, map to scramble an input plaintext image and
then leverages an S-box generated by a Cubic-Tent, C-T, cascade map and an improved generalized
two-dimensional Arnold map to substitute the scrambled image. Next, A DNA sequence operation
is further applied to diffuse the scrambled image. Theoretical analysis and experimental results show
that the proposed image encryption algorithm can effectively resist violent attack, correlation analysis,
information entropy analysis and differential attack, anti-crop attack and anti-noise attack.
ACKNOWLEDGMENTS
The authors would like to appreciate for the financial supports by the National Natural Science
Foundation of China (Nos: 60603092, 60975042, 51472066).
1 X. Y. Wang, Y. Q. Zhang, and X. M. Bao, “A colour image encryption scheme using permutation-substitution based on
chaos,” Entropy 17, 3877–3897 (2015).
2 D. D. Wheeler, “Problems with chaotic cryptosystems,” Cryptologia 13, 243–250 (1989).
3 K. Ning, “A pseudo DNA cryptography method,” arXiv 0903, 2693 (2009).
4 G. Jakimoski and L. Kocarev “Chaos and cryptography: block encryption ciphers based on chaotic maps,” IEEE Transactions
on Circuits and Systems I: Fundamental Theory and Applications 48, 163–169 (2001).
5 M. Khan, T. Shah, H. Mahmood, M. A. Gondal, and I. Hussain, “A novel technique for the construction of strong S-boxes
20 H. Liu and X. Wang, “Image encryption using DNA complementary rule and chaotic maps,” Applied Soft Computing 12,
1457–1466 (2012).
21 M. SaberiKamarposhti, I. AlBedawi, and D. Mohamad, “A new hybrid method for image encryption using DNA sequence
and chaotic logistic map,” Australian Journal of Basic and Applied Sciences 6, 371–380 (2012).
22 X. Xue, Q. Zhang, X. Wei, L. Guo, and Q. Wang, “An image fusion encryption algorithm based on DNA sequence and
trol synchronization and chaotic masking communication application,” Optik-International Journal for Light and Electron
Optics 127, 4024–4030 (2016).
29 Y. Wang, Z. Liu, J. Ma, and H. He, “A pseudorandom number generator based on piecewise logistic map,” Nonlinear
operation and hyper-chaotic system,” Journal of Systems and Software 85, 290–299 (2012).
37 C. Adams and S. Tavares, “The structured design of cryptographically good S-boxes,” Journal of Cryptology 3, 27–41
(1990).
38 A. F. Webster and S. E. Tavares, “On the design of S-boxes,” in Proceedings of Advances in CryptologyCRYPTO 85.Lecture
CryptologyCRYPTO 90.Lecture Notes in Computer Science (Springer-Verlag, 1990), Vol. 537, pp. 2–21.
41 M. Matsui, “Linear cryptanalysis method for DES cipher,” in Proceedings of Advances in Cryptology EURO-
CRYPT93.Lecture Notes in Computer Science (Springer-Verlag, 1994), Vol. 765, pp. 386–397.
42 T. Gao and Z. Chen, “A new image encryption algorithm based on hyper-chaos,” Physics Letters A 372, 394–400 (2008).
43 C. Zhu, “A novel image encryption scheme based on improved hyperchaotic sequences,” Optics Communications 285,
29–37 (2012).
44 A. Belazi, A. A. A. El-Latif, A. V. Diaconu, R. Rhouma, and S. Belghith, “Chaos-based partial image encryption scheme
based on linear fractional and lifting wavelet transforms,” Optics and Lasers in Engineering 88, 37–50 (2017).
45 Y. Liu, J. Wang, J. Fan, and L. Gong, “Image encryption algorithm based on chaotic system and dynamic S-boxes composed