Image Encryption Algorithm Based On Improved Itera
Image Encryption Algorithm Based On Improved Itera
1088/1402-4896/ad1ae1
PAPER
Abstract
The quick advancement of multimedia technology has led to the widespread utilization of digital
images across various industries. Meanwhile, image security has become an urgent issue in need of
resolution. To guarantee image security as well fulfill the requirements of real-time image
cryptosystems, a novel image encryption algorithm is proposed based on the designed improved
iterative chaotic map with infinite collapses (ICMIC), Gray code and semi-pixel level permutation and
diffusion. Firstly, two chaotic sequences are produced using the improved ICMIC and then they are
used to perform scrambling and diffusion operations. Secondly, the plain image is encoded using Gray
code and then we convert each Gray code into two semi-pixels. Thirdly, the designed semi-pixel level
permutation is used to scramble the semi-pixel vector. Finally, the scrambled semi-pixel vector is
diffused using the designed dynamic semi-pixel level diffusion. The performance analyses illustrate
that the proposed algorithm possesses robust security and high efficiency, making it ideally suited for
application in real-time image cryptosystems.
1. Introduction
With the continuous development of multimedia technology, digital images have been applied in different
fields, such as military, medicine, finance and transportation [1]. Meanwhile, the security of images has become
an increasingly important issue. During the process of images transmission and storage, they may face various
threats, such as data tampering, information disclosure and unauthorized access. Image encryption stands as a
prevalent method for enhancing image security. By converting a plain image into a cipher image with a noise-
like appearance, the content of the image can be protected from being known to unauthorized third parties [2].
Recently, researchers have proposed numerous advanced algorithms for image encryption, including the
algorithms using chaotic system [3], deep learning [4], DNA coding [5] and quantum coding [6].
Chaos-based image cryptosystems have become a current research hotspot, primarily because of its
sensitivity to initial conditions [7, 8]. Wen et al designed a two-dimensional (2D) hyper-chaotic map by crossing
Sine map and iterative chaotic map with infinite collapses (ICMIC), which exhibits more complex chaotic
behaviors compared with the Sine map and the ICMIC. Note that ICMIC is a chaotic map proposed by He et al
in 2001 [9] and infinite collapses can be observed in its phase portrait. However, it still has periodic states,
indicating that it cannot present chaotic behaviors throughout the entire control parameter range [10]. Liu et al
designed a novel hyper-chaotic map by using Sine map and ICMIC, which has a wider chaotic range compared
with other similar hyper-chaotic systems. However, its mathematical expression is relatively complicated, which
makes it unsuitable for image cryptosystems that require high real-time performance [11]. Mao et al designed a
novel one-dimensional (1D) chaotic map using ICMIC and Chebyshev map, which can present chaotic
behaviors throughout the entire control parameter range. However, its mathematical expression remains
complex, which makes it takes more time to generate chaotic sequences [12]. Wang et al designed a novel chaotic
map based on ICMIC and Logistic map, which can overcome the existing shortcomings of the two seed maps.
However, it is composed of two chaotic maps, making its mathematical expression not simple enough [13]. Gao
et al introduced a hyper-chaotic system based on the discrete memristor and ICMIC. However, its chaotic
performance is not good enough, i.e., the sequence distribution is nonuniform and the chaotic range is narrow
[14]. Wei et al presented a three-dimentional (3D) hyper-chaotic map based on ICMIC and Sine map. The use of
the closed-loop coupling method makes it has more complex chaotic behaviors compared with the two original
1D chaos. However, its mathematical expression is too complicated [15]. To overcome the shortcomings of the
above-mentioned chaotic maps, we design an improved ICMIC which can present chaotic behaviors
throughout the entire control parameter range and has a very simple mathematical structure.
The permutation refers to the change of pixel position, which is a crucial step in designing encryption
algorithm [16, 17]. Wang et al proposed a cross-plane scrambling method by using chaotic sequence and cyclic
shift operation, which can simultaneously scramble the R, G and B components of a color image. To improve the
scrambling effect, the column and row of each component are further scrambled using Inside-Out shuffling
[18]. Chen et al first split an image into different sub-blocks and then those blocks are scrambled using the
designed intra and inter scrambling methods. In addition, Fisher-Yates shuffling is utilized to further improve
the scrambling effect [19]. Wang et al used three different scrambling methods to scramble a color image. Firstly,
the 3D pixel matrix is scrambled using 3D spiral transformation. Secondly, the 2D pixel matrix is scrambled
using index sequence of a chaotic sequence. Thirdly, the 1D pixel vector is scrambled using 1D Arnold
transformation [20]. Wang et al designed a bidirectional spiral transformation, which can overcome the
shortcomings of the traditional one. After the pixel-level permutation, a DNA-level permutation is used to
further increase the scrambling effect [21]. Chen et al first split an image into different parts and then the
designed V-shape traversal is used to respectively scramble each part. After the V-shape traversal, index
permutation and row column index permutation are used to further scramble the image [22]. However, most of
the above scrambling methods require the combination with other scrambling methods to achieve a satisfactory
scrambling effect. In our work, we design a semi-pixel level permutation, which can simultaneously alter the
position and value of pixels and has a desirable scrambling effect without combining other scrambling methods.
The diffusion refers to the change of pixel value, which is another crucial step in designing encryption
algorithm [23, 24]. Rani et al designed a bit-level differential coding scheme to diffuse the scrambled image. After
that, circular shift operation is utilized to further diffuse the encoded binary sequence [25]. Liang et al utilized
random deoxyribonucleic acid (DNA) coding (DNA/RNA coding refers to the conversion between pixels and
DNA/RNA sequences through replacement rules between binary numbers and bases) and operations to
perform the first round of diffusion. And then, XOR operation and a mask image are utilized to perform the
second round of diffusion [26]. Gao et al first used DNA XOR operation and chaotic encoding sequence to
diffuse the scrambled image. Subsequently, the DNA mutation operation is used to further diffuse the image
after DNA operations [27]. Hu et al designed an adaptive diffusion scheme by using additive modules and XOR
operations. However, the designed diffusion scheme is somewhat complicated due to the multiple uses of
addition and modulo operations [28]. Zhang et al designed an ribonucleic acid (RNA) level permutation and
substitution, which can simultaneously alter the value and position of pixels. However, the diffusion scheme
requires multiple rounds to achieve a satisfactory diffusion effect [29]. Wang et al introduced a diffusion method
using central dogma of molecular biology, which has an excellent effect. However, the diffusion method is
composed of many processes, including DNA coding, operations, transcription and translation, making it run
inefficiently [30]. To summarize, most of the above diffusion methods require multiple rounds of diffusion,
which makes the diffusion efficiency relatively low. To ensure the diffusion effect while improving the efficiency
of diffusion, we design a dynamic semi-pixel level diffusion method, which can achieve a desirable diffusion
effect with one round of diffusion and has very high efficiency.
The following are the contributions made by our work. (1) We design an improved ICMIC, which has a
wider chaotic range and more complex chaotic behaviors than the traditional ICMIC. (2) We design a semi-pixel
level scrambling method, which can simultaneously change the position and value of pixels. (3) A dynamic semi-
pixel level diffusion method is designed, which has a desirable diffusion effect and a high execution efficiency. (4)
An image encryption algorithm is proposed based on the improved ICMIC and the designed semi-pixel level
permutation and diffusion, which has robust security and high efficiency.
The rest of this paper is organized as follows. Section 2 presents the improved ICMIC and its dynamics
performance analyses. Section 3 shows the proposed encryption and decryption algorithms. Sections 4 and 5
show the experimental results and the performance analyses of our algorithm. Section 6 shows the conclusions.
2
Phys. Scr. 99 (2024) 025232 Q Wang et al
2.1. Definition
We can describe the ICMIC [31] as
l
x n + 1 = sin ⎛ ⎞ ,
⎜ ⎟ (1)
x
⎝ n⎠
where the initial conditions λ ä (0, + ∞ ) and x ä (− 1, 1).
The ICMIC has many shortcomings, such as discrete chaotic range, weak sensitivity to the initial
conditionsand uneven sequence distribution. To overcome the shortcomings of the ICMIC, we design an
improved ICMIC as shown below.
p 6m
x n + 1 = sin ⎛⎜ 2 ⎞⎟ mod 1, (2)
⎝ xn ⎠
where mod (·) represents the modulo operation and the initial conditions μ ä (0, + ∞ ) and x ä (0, 1).
3
Phys. Scr. 99 (2024) 025232 Q Wang et al
Figure 2. Sensitivity analyses: (a) ICMIC with λ1 = 2 and λ2 = 2 + 10−14; (b) ICMIC with x1(0) = 0.6 and x2(0) = 0.6 + 10−14; (c)
Improved ICMIC with μ1 = 2 and μ2 = 2 + 10−14; d) Improved ICMIC with x1(0) = 0.6 and x2(0) = 0.6 + 10−14.
4
Phys. Scr. 99 (2024) 025232 Q Wang et al
increase of μ. Stated another way, the improved ICMIC has a continuous chaotic range and a strong sensitivity to
its initial conditions.
where ⌊·⌋ represents the round-down operation, x ¢ and x respectively refer to the binary sequence and the
chaotic sequence. If we let μ = 3 and x(0) = 0.5, table 1 shows the 15 subtest results of the chaotic sequence
5
Phys. Scr. 99 (2024) 025232 Q Wang et al
Frequency 0.449649 ✓
Block Frequency 0.134478 ✓
Rank 0.048977 ✓
FFT 0.468160 ✓
Runs 0.618081 ✓
Longest Run 0.782957 ✓
Universal 0.651447 ✓
Non-overlapping Template 0.680676 ✓
Linear Complexity 0.883160 ✓
Overlapping Template 0.789538 ✓
Approximate Entropy 0.208240 ✓
Serial test 1 0.570542 ✓
Serial test 2 0.524826 ✓
Excursions (X=1) 0.326169 ✓
Excursions Variant (X=1) 0.595838 ✓
Cumulative Sums-forward 0.988909 ✓
Cumulative Sums-reverse 1.000000 ✓
produced by the improved ICMIC. Obviously, all p-values > 0.01 illustrates that the improved ICMIC can pass
the NIST test and has high randomness.
6
Phys. Scr. 99 (2024) 025232 Q Wang et al
difference, so that there are fewer bit errors caused by one bit error during transmission. For a decimal number,
its encoding processes are as follows. Firstly, the decimal number is converted into its corresponding binary
form. After that, starting from the least significant bit of the binary number, the XOR operation is performed
with each bit and its left bit. Finally, keeping the most significant bit of the binary number unchanged, we can
obtain the corresponding Gray code of the decimal number. The above processes can be described using
equations (6) and (7).
G (n) = B (n) , (6)
G (i ) = B (i ) Å B (i + 1) , i Î [1, n - 1] , (7)
where B and G represent n-bit binary number and the corresponding Gray code, respectively. Figure 7 presents
two specific examples for explaining the process of Gray encoding.
For a gray image, its pixel values range from 0 to 255. Therefore, the Gray code of each pixel value and the
decimal form of the Gray code are shown in table 2. In our work, we first store the table 2 in Table_Gray
beforehand and then we can quickly obtain the Gray code of each pixel value and the decimal form of the Gray
code by looking up the table. Through this method, the time consumed in the Gray encoding process can be
greatly reduced. Below are the specific steps for encoding a gray image.
Step 1: Reshaping pixel matrix into pixel vector
P is reshaped into a pixel vector P1 of length mn.
Step 2: Gray encoding
Using the look-up table method, we can obtain the decimal form of the Gray code corresponding to each
pixel value. By this method, P1 is encoded into a Gray code vector P2 of length mn, where P2 is composed of Gray
code in decimal form.
7
Phys. Scr. 99 (2024) 025232 Q Wang et al
Pixel value Gray code Decimal ... Pixel value Gray code Decimal
scrambled pixel vector. Figure 8 is a specific example and the detailed scrambling processes are described as
below.
Step 1: Converting Gray code into semi-pixels
We convert each Gray code in P2 into two semi-pixels and the conversion method is as follows.
8
Phys. Scr. 99 (2024) 025232 Q Wang et al
Input: Plain image P; Chaotic sequences X and Y ; Gray code table Table _Gray .
Output: Cipher image P7.
// Gray encoding
1 P1 ¬ reshape (P , 1, mn);
2 for (i ¬ 1; i mn ;++ i ) do
3 P2 (i ) ¬ Table _Gray (P1 (i ) + 1, 3)
4 end
// Converting Gray code into semi-pixels
5 ℓ ¬ 1;
6 for (i ¬ 1; i mn ;++ i ) do
7 P3 (ℓ ) ¬ ⌊P2 (i ) 16⌋;
8 P3 (ℓ + 1) ¬ P2 (i ) mod 16 ;
9 ℓ ¬ ℓ + 2;
10 end
// Semi-pixel level permutation
11 [~,X idx ] ¬ sort (X ) ;
12 for (i ¬ 1; i 2mn ;++ i ) do
13 P4 (i ) ¬ P3 (X idx (i ));
14 end
9
Phys. Scr. 99 (2024) 025232 Q Wang et al
(Continued.)
// Dynamic semi-pixel level diffusion
15 R oper ¬ ⌊(X + 100) ´ 1010⌋ mod 4 ;
16 CV ¬ ⌊(Y + 100) ´ 1010⌋ mod 16;
17 for (i ¬ 1; i 2mn ;++ i ) do
18 Switch Roper(i) do
19 Case 0 do
20 P5 (i ) ¬ (P4 (i ) + CV (i )) mod 16;
21 Case 1 do
22 P5 (i ) ¬ (P4 (i ) - CV (i )) mod 16;
23 Case 2 do
24 P5 (i ) ¬ P4 (i ) Å CV (i );
25 Case 3 do
26 P5 (i ) ¬ P4 (i ) CV (i );
27 end
28 end
29 end
// Converting semi-pixels into pixels
30 ℓ ¬ 1;
31 for (i ¬ 1; i 2mn - 1; i ¬ i + 2) do
32 P6 (ℓ ) ¬ P5 (i ) ´ 16 + P5 (i + 1);
33 ℓ ¬ ℓ + 1;
34 end
35 P7 ¬ reshape (P6, m, n);
36 return P7;
where P3¢ is the diffused semi-pixel vector and its length is 2mn.
Step 4: Semi-pixel level reverse permutation
X ¢ is sorted in ascending order and then we can get its index sequence X ¢idx . After that, P3¢ is scrambled by
where P4¢ is the scrambled semi-pixel vector and its length is 2mn.
Step 5: Converting semi-pixels into Gray code
10
Phys. Scr. 99 (2024) 025232 Q Wang et al
We convert every two semi-pixels in P4¢ into its decimal form by using equation (20).
P5¢ (ℓ) = P ¢4 (i ) ´ 16 + P ¢4 (i + 1) , i = 1, 3,¼,2mn - 1, (20)
where P ¢5 is a Gray code vector which is composed of the Gray code in decimal form and its length is mn.
Step 6: Gray decoding
We can quickly find the pixel value corresponding to the Gray code in P ¢5 using the look-up table method.
Therefore, P ¢5 can be decoded into a pixel vector P6¢ of length mn. And then, we reshape P6¢ into a pixel matrix P7¢
with the size of m × n and P7¢ is the decrypted image.
Input: Cipher image P ¢ ; Chaotic sequences X ¢ and Y ¢ ; Gray code table Table _Gray .
Output: Decrypted image P7¢ .
// Converting pixels into semi-pixels
1 P1¢ ¬ reshape (P ¢ , 1, mn);
2 ℓ ¬ 1;
3 for (i ¬ 1; i mn ;++ i ) do
4 P ¢2 (ℓ ) ¬ ⌊P ¢1(i ) 16⌋;
5 P ¢2 (ℓ + 1) ¬ P ¢1(i ) mod 16 ;
6 ℓ ¬ ℓ + 2;
7 end
// Dynamic semi-pixel level reverse diffusion
8 R ¢oper ¬ ⌊(X ¢ + 100) ´ 1010⌋ mod 4 ;
9
CV ¢ ¬ ⌊(Y ¢ + 100) ´ 1010⌋ mod 16 ;
10 for (i ¬ 1; i 2mn ;++ i ) do
11 switch R ¢oper (i ) do
12 case 0 do
13 P ¢3 (i ) ¬ (P ¢2 (i ) - CV ¢ (i )) mod 16 ;
14 case 1 do
15 P ¢3 (i ) ¬ (P ¢2 (i ) + CV ¢ (i )) mod 16 ;
11
Phys. Scr. 99 (2024) 025232 Q Wang et al
(Continued.)
16 case 2 do
17 P ¢3 (i ) ¬ P ¢2 (i ) Å CV ¢ (i );
18 case 3 do
19 P ¢3 (i ) ¬ P ¢2 (i ) CV ¢ (i );
20 end
21 end
22 end
// Semi-pixel level reverse permutation
23 [~,X ¢idx ] ¬ sort (X ¢);
24 for (i ¬ 1; i 2mn ;++ i ) do
25 P ¢ 4 (X ¢idx (i )) ¬ P ¢3 (i );
26 end
// Converting semi-pixels into Gray code
27 ℓ ¬ 1;
28 for (i ¬ 1; i 2mn - 1; i ¬ i + 2) do
29 P5¢ (ℓ ) ¬ P4¢ (i ) ´ 16 + P4¢ (i + 1);
30 ℓ ¬ ℓ + 1;
31 end
// Gray decoding
32 for (i ¬ 1; i mn ;++ i ) do
33 idx ¬ find (Table _Gray (:,3) == P5¢ (i )) ;
34 P6¢ (i ) ¬ Table _Gray (idx , 1) ;
35 end
36 P7¢ ¬ reshape (P6¢, m, n);
37return P7¢ ;
4. Experimental results
We set the external keys as ξ1 = 6, ξ2 = 0.6, ξ3 = 5 and ξ4 = 0.2. The gray images with the size of 256 × 256 are
from the USC-SIPI image database [35]. From the experiment results in figure 10, we can learn that the cipher
images are noise-like and we cannot visually observe any useful information from them. In other words, our
algorithm can well hide the private information of the plain images. In addition, the cipher images can be
correctly decrypted using our designed decryption algorithm.
5. Performance analyses
where sign(·) stands for the sign function, P1 and P2 refer to two different cipher images. We can learn from the
NPCR values in table 3 that a tiny change in the key can produce two almost completely different cipher images.
We respectively change μ2 and x2(0) into μ2 + 1014 and x2(0) + 10−14, and then the new key sets are
denoted as SK3 and SK4, respectively. We respectively use SK0, SK3 and SK4 to decrypt the Peppers cipher
images and the decrypted images are presented in figures 11(d)-(f). Obviously, the cipher image can be correctly
12
Phys. Scr. 99 (2024) 025232 Q Wang et al
13
Phys. Scr. 99 (2024) 025232 Q Wang et al
decrypted using SK0, while it cannot be correctly decrypted using SK3 and SK4. In other words, a tiny change in
the key will not yield the correct decrypted image.
5.3. Histogram
The histograms of different images are presented in figure 12. For the plain images, the histograms have
relatively large fluctuations, making it easy for attackers to infer the information in the plain image. However, for
the cipher images, the histograms are very flat. In other words, the cipher images can well hide the statistical
characteristics of the plain images.
14
Phys. Scr. 99 (2024) 025232 Q Wang et al
5.5. χ2 test
The χ2 value [3] is calculated by:
2
255
( pi - q)
c2 = å q
, (22)
i=0
where pi refers to the frequency of all pixels and q = mn/256. For a gray image, its χ2 value is 284.3359 when we
set the significance level as 0.1. As seen in table 4, the χ2 values of the cipher images are below 284.3359,
indicating that our algorithm can pass the χ2 test.
15
Phys. Scr. 99 (2024) 025232 Q Wang et al
Table 4. χ2 test.
2L - 1
IE = - å p (s i ) log 2( p (s i )) , (23)
i=0
where p(si) stands for the probability of si occurring and L = 8 for a gray image. Table 5 shows that our algorithm
can bring the entropy of the Peppers cipher image closer to its ideal value 8 compared with the algorithms
mentioned in [12, 37–40]. Put differently, our algorithm can enhance the randomness and unpredictability of
the pixels.
5.7. Correlation
The correlation coefficient [38] is calculated by
rxy =
(
å i = 1 x i - å i = 1x i T
T T
)( y - å y T )
i
T
i=1 i
, (24)
2 2
( ( T
åi=1 x i - åTi = 1x i T ) ) (å ( y - å y T ) )
T
i=1 i
T
i=1 i
where xi and yi stand for two adjacent pixels and T refers to the total number of pixels chosen from the image. For
different plain and cipher images, table 6 presents their correlation coefficients in the horizontal (H), vertical (V)
and diagonal (D) directions. Obviously, our algorithm has the capability to bring the correlation coefficients of
the Peppers cipher image closer to 0 in comparison to the algorithms reported in [12, 37–40]. Besides, figure 14
shows the correlation distribution diagrams of the plain and cipher images. As can be seen, our algorithm can
significantly reduce the correlation between adjacent pixels.
16
Phys. Scr. 99 (2024) 025232 Q Wang et al
Plain Cipher
Algorithm Image
H V D H V D
17
Phys. Scr. 99 (2024) 025232 Q Wang et al
Figure 15. Experimental results and histograms of pure black and white images.
Correlation coefficient
Image χ2 value Information entropy
H V D
1 m n
∣ C1 (i , j ) - C 2 (i , j )∣
UACI = åå
m ´ n i=1 j=1 255
´ 100%, (25)
where C1 and C2 stand for two cipher images derived from the plain images that differ by just a single pixel. We
can learn from the NPCR and UACI values in table 8 that our algorithm can make the two indicators closer to
their optimal values (i.e., 99.6094% and 33.4635%) in comparison to the algorithms reported in [12, 37–40].
That is to say, our algorithm can render the differential attacks ineffective.
18
Phys. Scr. 99 (2024) 025232 Q Wang et al
5.10. Robustness
The first row in figure 16 shows the cropped Peppers cipher images, where the cropping sizes respectively are 1/
32, 1/16, 1/8 and 1/4. The first row in figure 17 presents the disturbed Peppers cipher images with salt and
pepper noise, where the noise intensities respectively are 0.01, 0.05, 0.1 and 0.15. After that, we respectively
decrypt the cropped/disturbed cipher images and the second rows in figures 16- 17 present the corresponding
decrypted images. It is obvious that the decrypted images remain recognizable, which means our algorithm is
robust to the above two types of attacks.
19
Phys. Scr. 99 (2024) 025232 Q Wang et al
Time (in
Step seconds) Percentage
table 9. As we can see from table 10 that our algorithm takes about 0.1944 second to encrypt a 256 × 256 gray
image, which surpasses the algorithms reported in [41–44].
6. Conclusions
In our work, we propose an image encryption algorithm using the improved ICMIC, Gray code, semi-pixel level
permutation and diffusion. The improved ICMIC can overcome the shortcomings of the traditional ICMIC and
has wide chaotic range and complex chaotic behaviors. The designed semi-pixel level scrambling method can
simultaneously alter the position and value of pixels and has a desirable scrambling effect by using it only once.
The designed dynamic semi-pixel level diffusion method has an excellent diffusion effect and high efficiency.
The security analyses demonstrate that our algorithm is capable of withstanding various types of attacks, such as
brute-force attacks, statistical attacks, differential attacks, cropping attacks, and noise attacks. The execution
time analysis indicates that our algorithm takes about 0.1944 second to encrypt a 256 × 256 gray image, making
it exceptionally well-suited for real-time image cryptosystems.
Acknowledgments
This research is supported by the Fundamental Research Funds for the Central Universities (No.:
2020ZDPY0223).
The data that support the findings of this study are available upon reasonable request from the authors.
20
Phys. Scr. 99 (2024) 025232 Q Wang et al
The authors declare that they have no known competing financial interests or personal relationships that could
have appeared to influence the work reported in this paper.
ORCID iDs
References
[1] Kaur M and Kumar V 2020 A comprehensive review on image encryption techniques Arch. Comput. Meth. Eng. 27 15–43
[2] Yang Y, Cheng M, Ding Y et al 2023 A visually meaningful image encryption scheme based on lossless compression spiht coding IEEE
Trans. Serv. Comput. 16 1–15
[3] Wang Q, Zhang X and Zhao X 2022 Image encryption algorithm based on improved zigzag transformation and quaternary dna
codingJournal of Information Security and Applications Journal of Information Security and Applications 70 103340
[4] Zhou Q et al 2023 Optical image encryption based on two-channel detection and deep learning Opt. Lasers Eng. 162 107415
[5] Wang Q, Zhang X and Zhao X 2023 Color image encryption algorithm based on novel 2d hyper-chaotic system and dna crossover and
mutation Nonlinear Dyn. 111 22679–705
[6] Ma Y and Zhou N 2023 Quantum color image compression and encryption algorithm based on fibonacci transform Quantum Inf.
Process. 22 1–33
[7] Zhou S, Wang X and Zhang Y 2023 Novel image encryption scheme based on chaotic signals with finite-precision error Inf. Sci. 621
782–98
[8] Erkan U, Toktas A and Lai Q 2023 2d hyperchaotic system based on schaffer function for image encryption Expert Syst. Appl. 213
119076
[9] He D, He C, Jiang L et al 2001 Chaotic characteristics of a one-dimensional iterative map with infinite collapses IEEE Transactions on
Circuits and Systems I: Fundamental Theory and Applications 48 900–6
[10] Wen J, Xu X, Sun K et al 2023 Triple-image bit-level encryption algorithm based on double cross 2d hyperchaotic map Nonlinear Dyn.
111 6813–38
[11] Liu X, Sun K and Wang H 2022 A novel image encryption scheme based on 2d silm and improved permutation-confusion-diffusion
operations Multimedia Tools Appl. 81 1–27
[12] Mao N et al 2023 Real-time image encryption algorithm based on combined chaotic map and optimized lifting wavelet transform
Journal of Real-Time Image Processing 20 35
[13] Wang X et al 2022 A novel meaningful image encryption algorithm based on newly-designed coupled map lattice and adaptive
embedding Optik 270 170073
[14] Gao X et al 2022 An effective multiple-image encryption algorithm based on 3d cube and hyperchaotic map Journal of King Saud
University-Computer and Information Sciences 34 1535–51
[15] Wei C and Li G 2022 A selective image encryption scheme using licc hyperchaotic system IET Image Proc. 16 3342–58
[16] Singh M and Singh A K 2022 A comprehensive survey on encryption techniques for digital images Multimedia Tools Appl. 81 1–33
[17] Ghadirli H M, Nodehi A and Enayatifar R 2019 An overview of encryption algorithms in color images Signal Process. 164 163–85
[18] Wang X et al 2023 A color image encryption and hiding algorithm based on hyperchaotic system and discrete cosine transform
Nonlinear Dyn. 111 1–24
[19] Chen R et al 2023 Medical image characteristic region recognition encryption algorithm based on intra and inter blocks scrambling and
lscc chaotic map Multimedia Tools Appl. 82 1–29
[20] Wang X, Liu L and Song M 2023 Remote sensing image and multi-type image joint encryption based on nccs Nonlinear Dyn. 111 1–27
[21] Wang Q, Zhang X and Zhao X 2023 Color image encryption algorithm based on bidirectional spiral transformation and dna coding
Phys. Scr. 98 025211
[22] Chen R et al 2023 An image encryption algorithm based on the lscmm chaotic map and bidirectional dynamic diffusion Multimedia
Tools Appl. 82 1–26
[23] Ahmed F et al 2023 A dna based colour image encryption scheme using a convolutional autoencoder ACM Transactions on Multimedia
Computing, Communications and Applications 19 1–21
[24] Elsaid S A, Alotaibi E R and Alsaleh S 2023 A robust hybrid cryptosystem based on dna and hyperchaotic for images encryption
Multimedia Tools Appl. 82 1995–2019
[25] Rani N, Mishra V and Sharma S R 2023 Image encryption model based on novel magic square with differential encoding and chaotic
map Nonlinear Dyn. 111 2869–93
[26] Liang Q and Zhu C 2023 A new one-dimensional chaotic map for image encryption scheme based on random dna coding Opt. Laser
Technol. 160 109033
[27] Gao X et al 2023 A color image encryption algorithm based on hyperchaotic map and dna mutation Chin. Phys. B 32 030501
[28] Hu Y and Nan L 2023 Image encryption algorithm based on 1d-sfacf with cross-cyclic shift and adaptive diffusion Phys. Scr. 98 055209
[29] Zhang D et al 2023 An image encryption algorithm based on joint rna-level permutation and substitution Multimedia Tools Appl. 82
23401–26
[30] Wang Q, Zhang X and Zhao X 2023 Image encryption algorithm based on 2d hyper-chaotic system and central dogma of molecular
biology Phys. Scr. 98 085244
[31] Wang X, Liu C and Jiang D 2022 An efficient double-image encryption and hiding algorithm using a newly designed chaotic system and
parallel compressive sensing Inf. Sci. 610 300–25
[32] Wang J et al 2022 A novel content-selected image encryption algorithm based on the ls chaotic model Journal of King Saud University-
Computer and Information Sciences 34 8245–59
21
Phys. Scr. 99 (2024) 025232 Q Wang et al
[33] Zhou Z et al 2023 Novel multiple-image encryption algorithm based on a two-dimensional hyperchaotic modular model Chaos,
Solitons Fractals 173 113630
[34] Abd-El-Atty B, El-Affendi M and El-Latif A A A 2023 A novel image cryptosystem using gray code, quantum walks, and henon map for
cloud applications Complex & Intelligent Systems 9 609–24
[35] The usc-sipi image database: https://sipi.usc.edu/database/, (1977).
[36] Çelik H and Doğan N 2023 A hybrid color image encryption method based on extended logistic map Multimedia Tools Appl. 82 1–24
[37] Wu X et al 2023 A novel image compression and encryption scheme based on conservative chaotic system and dna method Chaos,
Solitons Fractals 172 113492
[38] Wang X and Liu L 2021 Application of chaotic josephus scrambling and rna computing in image encryption Multimedia Tools Appl. 80
23337–58
[39] Luo Y et al 2020 Image encryption scheme by combining the hyper-chaotic system with quantum coding Opt. Lasers Eng. 124 105836
[40] Mfungo D E et al 2023 A novel image encryption scheme using chaotic maps and fuzzy numbers for secure transmission of information
Applied Sciences 13 7113
[41] Masood F et al 2022 A lightweight chaos-based medical image encryption scheme using random shuffling and xor operations Wirel.
Pers. Commun. 127 1405–32
[42] Mondal B and Singh J P 2022 A lightweight image encryption scheme based on chaos and diffusion circuit Multimedia Tools Appl. 81
34547–71
[43] Mansoor S and Parah S A 2023 Haie: a hybrid adaptive image encryption algorithm using chaos and dna computing Multimedia Tools
Appl. 82 28769–96
[44] Ferdush J, Begum M and Uddin M S 2021 Chaotic lightweight cryptosystem for image encryption Advances in Multimedia 2021 1–16
22