Nonlinear Dyn (2024) 112:10583–10609

https://doi.org/10.1007/s11071-024-09620-w

RESEARCH

Novel image encryption algorithm using hybrid 3D-ICPCM

and hessenberg decomposition
Vivek Verma · Sanjeev Kumar · Narbda Rani

Received: 11 January 2024 / Accepted: 9 April 2024 / Published online: 30 April 2024
© The Author(s), under exclusive licence to Springer Nature B.V. 2024

Abstract This study introduces a novel image encryp- satisfies various security standards and surpasses the
tion algorithm by combining the capabilities of a three- efficacy of recently introduced schemes.
dimensional chaotic map (3D-ICPCM) and Hessenberg
decomposition. By harnessing the Ikeda chaotic map Keywords 3D chaotic map · Hessenberg decom-
and the Cosine polynomial chaotic map as seed maps, position · SHA-256 · Image scrambling · Spiral
a new 3D chaotic map is developed, which becomes the permutation
pivotal element of the algorithm’s structure. The pro-
posed method employs an approach to ensure robust-
ness and security, integrating advanced methods of con- 1 Introduction
fusion and diffusion. Encryption involves three pro-
cesses on image pixels: a bit-level confusion process With the fast-evolving field of multimedia applications
using bit reversal and spiral permutation, and a diffu- such as WhatsApp, Facebook, Twitter and advance-
sion process by applying the 3D-ICPCM sequence and ments in the IT industry, vast amount of information is
Hessenberg decomposition for effective pixel value dis- being transmitted all around the globe over the unpro-
tribution. A series of comprehensive experiments were tected channels. Consequently, ensuring the secu-
conducted to assess the proposed algorithm’s effective- rity of these digital communications is essential [1].
ness. The subsequent research and comparative evalua- The literature is filled with numerous research efforts
tions demonstrate that the proposed encryption system aimed to ensuring information security and prevent-
ing from network intrusions, tampering, eavesdrop-
ping and distributed-denial-of-service (DDoS) attacks.
V. Verma · S. Kumar
Department of Mathematics, Indian Institute of Technology In the current digital age, a significant portion of this
Roorkee, Roorkee, Uttarakhand 247667, India information is disseminated via visual media, such as
e-mail: vivek_v@ma.iitr.ac.in digital images and videos. The value of content con-
S. Kumar (B) veyed through this medium can range widely, which
Mehta Family School of Data Science and Artificial Intelligence, encompassing everything from digital signatures, con-
Indian Institute of Technology Roorkee, Roorkee, Uttarakhand tract agreements, and medical records of the patient to
247667, India
e-mail: sanjeev.kumar@ma.iitr.ac.in armed forced data as well as personal snapshots [2].
Preventing the unauthorized disclosure of such sensi-
N. Rani
Department of Mathematics, IK Gujral Punjab Technical Uni-
tive information has emerged as a substantial concern.
versity (Main Campus), Kapurthala, Punjab 144603, India Therefore, it is essential to address potential malicious
e-mail: narmadasharma1990@gmail.com threats with advanced and efficient algorithms [3,4].

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10584 V. Verma et al.

Since digital images inherently exhibit characteris- it scrambles the bytes of a plain image utilizing three
tics like considerable data volumes, high redundancy, distinct chaotic maps. Two of the three chaotic maps
strong pixel correlation, as well as low information have been assigned explicitly to the permutation task,
entropy therefore, the traditional text encryption meth- whereas the third map has been employed to produce
ods such as AES [5] and DES [6] fall short for image a stochastic value. The byte stream has been subse-
encryption due to their complexity and inefficiency. To quently modified externally by introducing this value
mitigate this issue and enhance image data security, an through an exclusive-or operation. Hua et al. [32] sug-
abundance of alternative encryption techniques have gested a novel 2D chaotic map by modifying the logis-
been explored. For example, one-time pad [7], cellular tic map with the sine chaotic map, named 2D-LASM,
automata [8–10], hybrid genetic algorithms [11–14], which has been subsequently utilized to establish a
S-box [15,16], neuron model [17], matrix decomposi- fresh encryption mechanism. Feng et al. [31] have
tion [18,19], quantum cryptography [20] and compres- been able to successfully cryptanalyze this proposed
sive sensing [21–26] each offering different techniques scheme. However, the existing chaotic maps encounter
for the encryption of image data. These methodologies specific issues, mainly when executed on devices with
convert the plain image into a cipher image resem- finite precision, such as computers, where their inher-
bling noise, typically requiring an external source of ently chaotic nature may be compromised. Further-
entropy for the encryption process. For instance, the more, the distribution of points within their orbits
various image encryption strategies involve modifying are frequently asymmetrical, whereas a uniform dis-
the pixel positions, using random sequences to deter- tribution is generally preferred for encryption purpose.
mine the scrambling order. Consequently, the security Although 1D and 2D chaotic systems can potentially
of these encryption systems is significantly dependent increase encryption performance, their reduced num-
on the random nature of these sequences, indicating ber of variables and parameters may lead to predictable
the importance of identifying a significant source of behavior. On the contrary, higher-dimensional chaotic
entropy to develop image encryption schemes. With maps offer an enhanced level of security by their more
the evolution of chaos theory, it has been identified intricate framework comprised of a more significant
that the chaotic maps exhibit characteristics ideal for number of variables and parameters.
the encryption procedure, including through informa- Ye et al. [21] have developed a new image
tion mixing, inherent randomness and high sensitivity compression-hiding algorithm using a 3D chaotic sys-
to initial values as well as parameters [27–29]. tem and 2D compressive sensing for asymmetric and
Matthews introduced the notion of “chaotic cryptog- visually coherent image encryption. The technique
raphy” in 1989, which subsequently sparked a notable generates a unique keystream from features of the plain
surge in research within this domain. Subsequently, image, processed through innovative models TransM
numerous encryption algorithms, founded upon chaos and GetM, and secured with RSA encryption. It also
theory, have been intensively studied and implemented utilized integer wavelet transformation for hiding the
throughout the scrambling and diffusion phases of the encrypted data within a carrier image, preserving visual
standard chaotic encryption framework [4,30–36]. Qi quality and enhancing security. Ye et al. [25] proposed
et al. [27] introduced a two-dimensional (2D) chaotic an enhanced 3D chaotic map characterized by broaden
model called tan-sin-coupling (TSCC) and applied it bifurcation range and introduced a reversible image
to the tasks of image steganography and encryption. hiding method based on singular value sampling and
The author of this article has integrated steganogra- compressive sensing. In this technique, two mathe-
phy and encryption techniques to safeguard informa- matical models has been used extraction of the plain
tion. The empirical findings indicate that the cipher text messages from the secret image, another one for the
has been characterized by a higher level of security, key stream generation. Further, the image has been
although at the cost of a lower embedding capacity of secured by using the compressive sensing and singu-
the scheme. In [37] a chaotic image encryption method lar value sampling techniques. which allowed the con-
utilizing content-sensitive dynamic function switching cealment of a large volume of data. Wu et al. [22]
has been proposed. Designed as a symmetric cipher, devised a method for reversible blind image hiding
this algorithm decrypts a sequence of one-dimensional technique aimed to enhancing the security of VMIE.
bytes. Effectively employing confusion and diffusion, In addition to the plain image, this approach utilizes

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10585

the Rivest-Shamir-Adleman (RSA) algorithm to gen- into one ciphertext image and cannot encrypt multi-
erate initial encryption keys. Following this, confiden- ple plaintext images into multiple ciphertext images
tial information has been appended to the host image in one encryption. Yu et al. [18] employed the singu-
using the Integer Wavelet Transform (IWT), thereby lar value decomposition (SVD) and a chaotic system
enabling the restoration of the plain image without to create a mechanism for compression, encryption,
requiring the original host image. Xian et al. [4] intro- and authentication. The SVD has been utilized to com-
duced a spatiotemporal chaotic system incorporating press the plaintext image after decomposing it into a
a DPFSV (double parameters fractal sorting vector) left singular value matrix, right singular value matrix,
model. This model response to the limitations of the and a singular value matrix. Then the compressed data
conventional fixed selection model that utilizes the cou- has been encrypted using a logistic tent sine map, and
pled map lattices (CML) to determine nodes. More- a proposed authentication scheme was implemented.
over, the study introduced a novel cryptographic sys- The aforementioned scheme exhibits exceptional com-
tem that combines permutation-diffusion synchronous pression efficacy and satisfactory security outcomes.
encryption method with a spatiotemporal chaotic sys- Nonetheless, the proposed scheme lacks a robust attack
tem. Qin et al. [38] introduced a approach for encrypt- analysis.
ing colour images through the utilization of a globally From the literature it has been concluded that there
designed coupled map lattice. Despite the algorithm are various limitations in the existing image cryp-
protection level is commendable, its encryption time is tosystems, including excessive storage needs, key gen-
suboptimal. Moreover, an optical single-channel color eration process, the inadequacy of diffusion process
image cryptosystem using chaotic fingerprint phase and vulnerability to various cyberattacks. Additionally,
mask and diffractive imaging has been suggested by the chaotic systems being utilized exhibits suboptimal
Wang et al. [39]. In this system, the chaotic finger- chaotic behavior, marked by transient cycle, lack of
print phase masks (CFPMs) produced by the finger- ergodicity, and orbits that can be easily predictable,
print, the chaotic Lozi map, and secure hash algorithm which compromises the effectiveness of scrambling
(SHA-256) has been exploited only as interim vari- and diffusion techniques. Therefore, it is essential
ables. The amplified amplitude part has been modulated to investigate and analyze the new chaotic maps
by CFPMs positioned at different diffraction distances that address these issues, aiming to generate chaotic
to obtain encrypted image into a noise-like diffraction sequences with significantly enhanced randomness.
pattern. Henceforth, to overcome the limitations inherent in
Su et al. [19] proposed a method that implements the chaotic maps a new three dimensional chaotic map
Hessenberg decomposition on pixel blocks. The pro- (3D-ICPCM) has been proposed. This map synergizes
cess of watermark embedding has been achieved by the dynamics of a 2D-Ikeda chaotic map with a 1D
evaluating the frequency domain coefficients at posi- cosine polynomial chaotic map, effectively enhanc-
tions (1, 2), (2, 1), and (2, 2) within the Hessenberg ing complexity, sensitivity to initial conditions, and
matrix and selecting the highest value among these expanding the parameters [42,43]. Moreover, to raise
for quantization. While matrix factorization methods the randomness within the image encryption process,
for watermarking produce reliable outcomes, but they a novel image encryption algorithm (3DICPCM-IEA),
have high time complexity, of O(n 3 ) or higher. Wei et which leverages the 3D-ICPCM as a non linear chaotic
al. [40] developed a 3D chaotic system utilizing closed- source in conjunction with Hessenberg decomposition
loop coupling, exhibiting favorable hyperchaotic char- has been introduced. This algorithm has been applied to
acteristics and randomness. Nevertheless, the intricate both color and grayscale images, with pixel modifica-
arrangement exhibits complexity, while the pace of iter- tion achieved through strategic use of the 3D-ICPCM’s
ation is sluggish. Wang et al. [41] introduced a cas- properties and sequences, combined with Hessenberg
cade modulation chaotic system (CMCS) capable of decomposition, thereby significantly improving the
producing diverse chaotic systems. The authors subse- proposed model’s reliability and practicality.
quently employed CMCS chaotic system to devise a The contributions and novelty of this study can be
multiple-image encryption algorithm featuring a block summarized as follows:
confusion-diffusion procedure. However, the algorithm
can only encrypt three gray images of different types

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10586 V. Verma et al.

(1) This paper introduces a newly developed 3D where μ is a parameter and this map depicts chaotic
chaotic map to overcome the shortcomings observed nature when μ ∈ [0.5, 0.95].
in prior low-dimensional chaotic maps. These lim- (2) One dimensional cosine polynomial chaotic map
itations include insufficient key space, compro- (1D-CPCM)
mised security, and a simplistic structure. More- Talhaoui [42] introduced the 1D-CPCM and its
over, it mitigates the evident periodic window expression as follows in Eq. (2):
behavior observed in the Ikeda and cosine poly-
nomial chaotic maps. Z n+1 = cos(β(Z n 3 + Z n )) (2)
(2) The proposed image scrambling technique in our
work includes a dual scrambling procedure called where β is a chaos nature control parameter. The map
SBPS and BRMC, that ensures both pixel-level and depicts the chaotic nature for β > 1.05.
bit-level permutation which efficiently reduces the Remark:While chaotic maps have shown some suc-
correlation between the adjacent pixels of images. cess in encrypting images, there are still several limi-
(3) The second phase involves the application of dif- tations with this approach. For instance, some chaotic
fusion to a substantial portion of the Hessenberg maps may exhibit simple behavior and result in a rela-
decomposition and the chaotic sequences formed tively small key space, which can make them vulnera-
by a three-dimensional chaotic map to accomplish ble to attacks. To address these issues, we are exploring
the image data secure. the use of higher-dimensional chaotic maps, which can
Furthermore, the utilization of the SHA-256 hash func- be generated using two simpler chaotic maps as seed
tion output of the initial image is employed to calculate maps.
the encryption parameters in the key generation process Based on the Ikeda chaotic map and one dimensional
enhances the key space and the key sensitivity of the cosine polynomial chaotic map we proposed the fol-
proposed algorithm. lowing 3D chaotic map to enhance the dynamic nature
The rest of the article is summarized as follows: In and secret key space in Eq. (3):
Sect. 2, the propounded 3D-ICPCM chaotic map and its ⎧
⎪ 6
dynamical characteristic is described. In Sect. 3, details ⎪
⎪ tn = 0.4 −
⎪
⎪ (1 + X n 2 + Yn 2 )
the proposed image encryption algorithm. Section 4, is ⎨
dedicated to assessing the security and efficacy of the X n+1 = 1 + μ(X n cos tn − Yn sin tn + β Z n ) (3)
⎪
⎪
proposed algorithm via various cryptographic evalua- ⎪
⎪ Yn+1 = μ(X n sin tn + Yn cos tn + β Z n )
⎪
⎩
tions. Section 5, delineates the conclusion of the pro- Z n+1 = cos(β(X n 3 + Yn + Z n ))
posed work and outlines the direction for future work.
where μ is a chaos nature control parameter and β = 5
is a fixed parameter for getting the chaotic nature to the
2 Proposed 3D-ICPCM chaotic map
newly proposed chaotic map.
Chaotic systems are highly sensitive to initial condi-
tions, which means that small changes in the initial val- 2.1 Chaotic analysis of 3D-ICPCM map
ues can lead to significantly different outcomes. Due to
their unpredictability and sensitivity, chaotic maps are The dynamical nature of the proposed chaotic sys-
useful for creating secure encryption algorithms that tem is analyzed in this section using chaos trajectory
are resistant to different attacks [15]. (Phase diagram), bifurcation diagram, Lyapunov expo-
(1) Ikeda chaotic map nent, permutation entropy, 0-1 test and sensitivity anal-
This chaotic map was proposed by K. Ikeda [43] and ysis.
its expression mentioned below in Eq. (1):
⎧ 2.1.1 Phase (trajectory) and bifurcation diagram
⎪
⎪
6
⎨tn = 0.4 −
⎪ analysis
(1 + X n 2 + Yn 2 )
X n+1 = 1 + μ(X n cos tn − Yn sin tn ) (1)
⎪
⎪ Within a dynamic system, the trajectory that illustrates
⎪
⎩Y
n+1 = μ(X n sin tn + Yn cos tn ) chaos demonstrates the progression over time, start-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10587

Fig. 1 Phase diagram of the chaotic systems. a–d Trajectories map with initial conditions (0.53, 0.67). f Trajectory diagram of
of the newly designed 3D chaotic map with different planes and the 1D-CPCM with initial conditions (0.2)
initial values (0.3, 0.4, 0.5) e Trajectory diagram of the Ikeda

ing from a particular initial value. The distribution of comprehensive parameter range and superior random-
this trajectory across the phase plane reflects, to some like performance.
extent, the intrinsic unpredictability in the outputs pro-
duced by the chaotic system from a theoretical perspec- 2.1.2 Lyapunov exponent analysis
tive. The initial values for trajectory and bifurcation
analysis are set to (x0 , y0 , z 0 ) = (0.3, 0.4, 0.5) along The Lyapunov exponent is a metric employed in the
with the parameters are considered as μ = 0.98 and analysis of chaotic systems to quantify the degree of
β = 5. Figure 1 depicts the trajectory analysis for the sensitivity of such systems to minute variations in
various chaotic maps. their initial conditions. In order to ascertain the chaotic
The bifurcation diagram illustrates the long-term nature of a nonlinear system, it is necessary to verify
evolutionary phases of a time sequence produced by the positivity of its Lyapunov exponent. In the event
a nonlinear system given particular initial conditions. that a system exhibits chaotic behavior, its Lyapunov
These states include stable, unstable, periodic, and exponent will possess a positive value [20]. In a multidi-
chaotic phases [20]. Figure 2 displays the bifurcation mensional systems, there may exist multiple Lyapunov
diagram of proposed 3D-ICPCM, Ikeda chaotic map exponents that exceed zero, resulting in a more complex
and 1D-CPCM respectively. When the parameter μ behavior of the system. First, under the initial values
is within the range of [0, 1], the comparison findings (0.2, 0.2, 0.2) and β = 5, Fig. 3 presents the spectrum
demonstrate that the 3D-ICPCM outperforms both the of Lyapunov exponents for the 3D-ICPCM, highlight-
Ikeda chaotic map and 1D-CPCM regarding a more ing its relationship with the tunable parameter μ. It is

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10588 V. Verma et al.

Fig. 2 a Bifurcation diagram of new proposed 3D chaotic map adjustable parameter (μ ∈ [0, 1]) with initial value (0.53, 0.67);
with adjustable parameter( μ ∈ [0, 1], β = 5) with initial value c bifurcation plot of 1D-CPCM with adjusted parameter (β ∈
(0.3, 0.4, 0.5); b bifurcation plot of 2D-Ikeda chaotic map with [0, 2]) with initial value (0.2)

observed that when the parameter μ varies between The observed phenomenon indicates that slight vari-
0 and 1, the number of positive Lyapunov Exponent ations in the parameter lead to significant alterations
(LE) values for the proposed 3D-ICPCM map exceeds in the generated sequences, thereby demonstrating the
2. This observation indicates a significant presence of highly initial-value sensitivity of the proposed chaotic
hyper-chaotic dynamics within it. map.
Figure 5a depict the variations in sequences x and
y that arise due to alterations in the initial value of z.
2.1.3 Sensitivity analysis
Similarly, Fig. 5b, c illustrates the impact on sequences
y and z and x and z corresponding to the change in
The objective is to examine the system’s behavior
initial value of x and y respectively. The blue lines rep-
when subjected to minor perturbations in either the ini-
resent the output sequences of the chaotic map when
tial conditions or parameters [44]. In a system char-
the initial conditions are unchanged. The red lines rep-
acterized by chaos, tiny variations in the initial con-
resent the output sequences of the chaotic map when
ditions or parameters can result in significantly dif-
the initial conditions are slightly changed.
fering outcomes over an extended period. Figure 4
shows the sequences obtained using the initial condi-
tions (x0 , y0 , z 0 ), (x0 + 10−14 , y0 , z 0 ) and (x0 , y0 , z 0 ),
(x0 , y0 + 10−14 , z 0 ) and (x0 , y0 , z 0 ), (x0 , y0 , z 0 +
10−14 ) respectively, where number of iterations is 100.

Fig. 3 a Lyapunov exponent of newly proposed 3D-ICPCM chaotic map with tunable parameter; b–c Lyapunov exponent of ikeda
chaotic map and 1D cosine polynomial chaotic map respectively with tunable parameter

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10589

Fig. 4 Sensitivity analysis based on the impact of minor initial- Fig. 5 a Sensitivity analysis for sequences x and y when ini-
condition modifications: a sequence x; b sequence y; c sequence tial value of z changed; b Sensitivity analysis for sequences y
z and z when initial value of x changed; c Sensitivity analysis for
sequences x and z when initial value of y changed

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10590 V. Verma et al.

chaotic map sequence φ(n)n=1,2,3,...N , value of 0-1 test

is mathematically calculated by Eq. (6) as follows:

⎧
⎪ Cov(λ, ω)
⎪
⎪ K = Corr(λ, ω) = √
⎪
⎪ var (λ) var (ω)
⎪
⎪
⎪
⎪ λ = (1, 2, ....., n)
⎪
⎪
⎪
⎪
⎪
⎪ ω = (S(1), S(2), ......., S(n)) wher e
⎪
⎪
⎪
⎪ 1  r
⎪
⎪ Cov(l, m) = ¯ (m(i) − m̄)
(l(i) − l)
⎪
⎪ r i=1
⎪
⎪
⎪
⎪ 1 r
⎪
⎪
⎪l¯ = l(i), Var(l) = Cov(l, l)
⎪
⎪ r i=1
⎨ 
1 N
⎪ S(n) = lim [ p(i + n) − p(i)]2
⎪
⎪ N →∞ N i=1

Fig. 6 Permutation entropy of different chaotic maps ⎪
⎪
⎪
⎪ 1−cos(nz)
⎪
⎪ +[q(i + n) − q(i)] 2 − (Eφ)2
⎪
⎪
1−cos(z)
⎪
⎪  n
2.1.4 Permutation entropy ⎪
⎪ p(n) = φ(i) cos(i z),
⎪
⎪and
⎪
⎪
⎪
⎪ 
i=1
Permutation entropy is a technique utilized for quanti- ⎪
⎪
n
⎪
⎪q(n) = φ(i) sin(i z);
fying the intricacy and uncertainty of time series data ⎪
⎪
⎪
⎪
i=1
that arise from chaotic systems. The method identifies ⎪
⎪ 1 N
⎪
⎩ E(φ) = lim φ(i)
dynamic variations in a given time series by comparing N →∞ N i=1
adjacent time series [45]. (6)
The mathematical formula for calculating the P E is
given by Eq. (4) with embedding dimension = 8 and In the context of a dynamic system, the system of equa-
delay = 1. Figure 6 shows that the Permutation entropy tion is considered to be chaotic if K ≈ 1. In this article,
of different chaotic maps. Given that the P E of an we choose N = 1000, z ∈ [π/5, 4π/5] to determine
ideal random sequence belongs to [0, 1], it is evident the value of K . The K value approaches unity when the
from the depicted figure that the P E metric of 3D- parameters μ are within the range of [0, 1]. Figure 7
ICPCM exceeds that of other chaotic maps and higher depicts the 0-1 test results of chaotic maps. Experi-
degree of stability compared to others. Consequently, mental outcomes demonstrates that the proposed 3D-
the resulting chaotic sequence generated by this map ICPCM map exhibits good chaotic behavior in com-
possesses a higher level of randomness parison to 1D-CPCM and 2D Ikeda map.


K
PE = − Pl ln(Pl ) (4)
2.1.6 NIST-SP 800 test
l=1

Normilzation process: To measure the randomness of the chaotic sequences

generated by the proposed 3D-ICPCM, employed the
PE NIST SP800-22 test suite, which is extensively utilized
PE = (5)
ln(n!) in the field of cryptography [20]. This suite comprises
fifteen distinct tests, each yielding a p-value ranges
from [0, 1]. A test is considered successful if its p-
2.1.5 0-1 test analysis value exceeds the confidence threshold p. For the pro-
posed chaotic map’s randomness evaluation, set p at the
The 0-1 test [24] is a procedure to determine the chaotic recommended threshold of 0.01. The p-values for the
characteristic of the proposed 3D chaotic map. When suite’s individual tests are presented in Table 1. All the
we set the initial conditions of newly proposed chaotic p-values obtained from the NIST statistical tests sur-
map are (x0 , y0 , z 0 ) = (0.42, 0.63, 0.52). For the pass the 0.01 threshold, indicating that the sequences

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10591

Fig. 7 a–c represent Boolean motion states, while the 3D-ICPCM exhibits a chaotic state at this particular period. d–e Comparison of
K-value analysis between proposed 3D-ICPCM with 1D-CPCM and 2D Ikeda chaotic map respectively

generated by the 3D-ICPCM exhibit random distribu- cific steps of the proposed scheme are outlined below,
tion. and the corresponding encryption flowchart is illus-
trated in Fig. 9.

3 Proposed image encryption algorithms

In this section, we introduce a novel image encryption 3.1 Key generation process
technique that merges the intricacies of a 3D chaotic
map with Hessenberg decryption. This method unfolds This process generates a 256-bit hash value, denoted
through a trio of phases: initially, it employs SHA-256 as hexadecimal number K . Every 16-bit hash value
and the 3D chaotic map for key generation. Subse- in K is utilized accordingly to Eq. (7) to determine
quently, the pixel confusion stage utilizes bit reversal (h 1 , h 2 , h 3 , h 4 ).
and spiral permutation techniques to scramble the pixel ⎧
locations. Finally, the diffusion phase applies the 3D ⎪ hex2dec(K (49 : 64))
⎪
⎪ h1 = ,
⎪
⎪ 214
chaotic map in conjunction with Hessenberg decom- ⎪
⎪ hex2dec(K (33 : 48))
position, effectively reducing pixel correlation and uni- ⎪
⎨h 2 = ,
214 (7)
formly distributing pixel values to defense against sta- ⎪ hex2dec(K (17 : 32))
⎪
⎪ h3 = ,
tistical attacks. The capability of this scheme is val- ⎪
⎪ 214
⎪
⎪ (1 :
idated on images of diverse dimensions and types, ⎪
⎩h 4 =
hex2dec(K 16))
including both grayscale and colored images. The spe- 214

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10592 V. Verma et al.

Table 1 NIST test values for proposed 3D chaotic map (P- Pass; F-Fail)
Subtests X-seq Result Y-seq Result Z-seq Result

Frequency test 0.0156 P 0.1543 P 0.7540 P

Block Frequency test 0.6646 P 0.1150 P 0.2868 P
Runs Test 0.7147 P 0.8388 P 0.1632 P
Longest-Run-of-Ones Test 0.2820 P 0.9961 P 0.6233 P
Binary Matrix Rank Test 0.9972 P 0.1490 P 0.7712 P
Discrete Fourier Transform Test 0.8803 P 0.3127 P 0.9141 P
Non-Overlapping Matching Test 0.5068 P 0.5032 P 0.4993 P
Overlapping Matching Test 0.9661 P 0.9969 P 0.9937 P
Universal Statistical Test 0.9989 P 0.9908 P 0.9981 P
Linear Complexity Test 0.1290 P 0.8012 P 0.3916 P
Serial Test 0.7535 P 0.4571 P 0.8626 P
Approximate Entropy Test 1.0 P 1.0 P 1.0 P
Cumulative Sums Test 0.0135 P 0.2289 P 0.9124 P
Random Excursion Test
State -4 0.2916 P 0.6888 P 0.7649 P
State -3 0.7371 P 0.7940 P 0.1343 P
State -2 0.6318 P 0.2591 P 0.2405 P
State -1 0.0285 P 0.6143 P 0.1139 P
State 1 0.4535 P 0.1414 P 0.5537 P
State 2 0.2070 P 0.0890 P 0.4072 P
State 3 0.2787 P 0.7525 P 0.2914 P
State 4 0.3950 P 0.6158 P 0.6565 P
Random Excursion Variant Test
State -4 0.6757 P 0.5929 P 0.5537 P
State -3 0.2293 P 0.4478 P 0.7852 P
State -2 0.1206 P 0.4624 P 0.5464 P
State -1 0.2683 P 0.6713 P 0.6634 P
State 1 0.8743 P 0.8875 P 0.3840 P
State 2 0.7150 P 0.4624 P 0.8016 P
State 3 0.5716 P 0.3427 P 0.2932 P
State 4 0.6757 P 0.5212 P 0.3744 P

Secondly, the initial values of the 3D chaotic map is ⎧

⎨x0 = mod((h 1 + s) + 1),
⎪
generated in the following manners: determine the aver-
y0 = mod((h 2 + s) + 1), (9)
age value of each pixel in the two-dimensional image ⎪
⎩
and normalize the outcome to obtain the variable s. The z 0 = mod((h 3 + h 4 + s) + 1)
initial values x, y and z of the chaotic map are deter-
mined through the utilization of h 1 , h 2 , h 3 , h 4 , and s.
3.2 Bit reversal matrix conversion


n 
n
I (i, j) The bit-reversal technique can be used to introduce
i=1 j=1
s= , (8) non-linearity into a cryptographic algorithm, making
255 × n × n it more difficult for an attacker to analyze or break.
By applying bit-reversal to the data before performing

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10593

encryption or decryption, the input data becomes less

predictable and less susceptible to statistical attacks. In Algorithm 2 SPBS Algorithm
this technique, the algorithm scrambles the binary val- Input: Rectangular image P1 ; Size of image blocks
ues of an input image by swapping the position of each m 0 ; Spiral pattern sequence for each block s1 ; Block
pair of bits in the binary value of each pixel in the image, scrambling sequence s2
which makes the pixel values more complex and harder Output: Scrambled image P1
to predict. The technique being proposed is a modified Begin:
version of the methodology initially defined by Sura et 1. Calculate number of image blocks: M1 =  mn0 ×
al. [46]. Let us take an example of n × n matrix and  mn0
Algorithm 1 illustrates the proposed bit reversal proce- 2. Generate spiral patterns for each block size m 0 :
dure in detail. spiral_pattern = spiral(m 0 )
Algorithm 1 BRMC Algorithm inverse_spiral_pattern = fliplr(spiral_pattern)
Input: I is a matrix of size n × n 3. Apply scrambling to each block:
Output: P1 and P2 (array of size n × n) for i = 1 to M1
Begin for j = 1 to M1
Calculate block indices and locations by the
1. I (i, j) = (I (i, j))2 , where ( )2 convert chaotic sequence s1
to the binary values k = s1 ((i − 1) × M1 + j)
2. Initialize an array T of length n × n × 16 (k − 1) mod M1 → m k
3. For each element i in I , do (k − 1)
if I {i, j}(k) = 0 → nk
M1
set T (2i − 1) = 0 and T (2i) = 1 (i − 1) mod M1 → m i
else (i − 1)
→ ni
set T (2i − 1) = 1 and T (2i) = 0 M1
4. T1 = T [1 to 8] Extract block B from P1 at location i based
T2 = T [9 to 16] on m 0
5. P1 = (T1 )10 , where ( )10 con- if s2 ((i − 1) × M1 + j) mod2 = 0
vert to the decimal values B = apply inverse_spiral_pattern(m 0 )
6. P2 = (T2 )10 else
B = apply spiral_pattern(m 0 )
Place rearranged B into P  at location deter-
mined by s1 (i)
3.3 Spiral permutation block scrambling
end
end
The conventional process for pixel scrambling entails
4. The scrambled image P1 is obtained.
the rearrangement of an image’s pixels in a spi-
ral pattern, commencing from the innermost ele-
ments and progressing towards the outermost elements. Similarly, we can apply the Algorithm 2 on matrix
Although spiral scrambling techniques offer a consid- P2 also, to obtain the matrix P2 . For illustration, con-
erable degree of security and resilience, they are subject sider a 6 × 6 matrix is divided into 4 sub-blocks of size
to certain constraints. During the scrambling procedure 3 × 3 and then the spiral permutation block scrambling
that relies on scanning techniques, it proves challenging procedure is shown in Fig. 8.
to prevent the occurrence of unaltered pixel locations.
To address this issues, this article exhibits a novel spiral
permutation block scrambling procedure, which aims 3.4 Hessenberg decomposition (HD)
to address the issues and improve the scrambling effi-
cacy. This approach not only assures the algorithm’s The Hessenberg decomposition is a mathematical pro-
security but also alters the arrangement of every pixel cedure that factorizes a square matrix into the product
in a single encryption cycle. The proposed procedure of an upper Hessenberg and an orthogonal matrix. An
scrambling is described in Algorithm 2. upper Hessenberg matrix is a type of square matrix

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10594 V. Verma et al.

Fig. 8 6 × 6 matrix
transferred by SPBS via a
3 × 3 sub-block

characterized by the presence of zero elements below where v is the non zero vector, Im is identity
the first sub-diagonal [47]. Hessenberg decomposition matrix.
involves the factorization of a matrix M through an
orthogonal transformation. The process of decomposi-
tion can be illustrated in the following manner:
⎧ 3.5 Proposed image encryption/decryption algorithm
⎪
⎪ H = (S1 S2 ......S N −3 S N −2 )
T
⎪
⎪
⎨ M(S1 S2 ......S N −3 S N −2 ) This section provides a description of the specific steps
(10)
⎪
⎪ H = QT M Q involved in implementing the proposed encryption and
⎪
⎪
⎩M = Q H QT decryption algorithm.

where Q is the orthogonal matrix derived from the

QR decomposition; H is upper Hessenberg matrix,
3.5.1 Encryption algorithm
H = (h ik ) ∈ Rm×m satisfy h ik = 0 (i > k + 1);
S is Householder matrix.
⎡ ⎤ Take the initial image In×n , where n represents the size
h 11 h 12 ..... h 1m of input image.
⎢h 21 h 22 ..... h 2m ⎥
⎢ ⎥ Steps 1. Apply the SHA-256 hash function on the
⎢ . . . . ⎥
H =⎢ ⎢ ⎥ (11) image to generate the initial values of chaotic map as
⎥
⎢ . . . . ⎥ mentioned in key generation procedure.
⎣ . . . h (m−1)m ⎦ Step 2. Iterate the 3D chaotic map n 2 +1000 times to
0 h m(m−1) h mm generate chaotic sequences X, Y and Z , according the
initial conditions and discard the 1000 values to neglect
(Im − 2vv T ) the transient error. According to Eq. (13), sequences are
S= (12) discretized to generate the sequences L 1 , L 2 and L 3 .
vT v

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10595

Fig. 9 Block diagram of

proposed encryption scheme

⎧
⎪ ⎧
⎨ L 1 = |Z |,
⎪
⎨ X 0 = (cos(X ) × 10 ) mod 256
14
 
L 2 = (|Y |) × 105 mod 256 (13)
⎪
⎩   Y0 = (cos(Y ) × 10 ) mod 256
14 (14)
L 3 = (X (1 : m 0 × m 0 ) ) × 105 mod 256 ⎪
⎩
Z 0 = (cos(Z ) × 1014 ) mod 256
Step 3(a). Pixel scrambling process: apply Algorithm
1 to obtain the two scrambled matrices P1 and P2 . Step 6. Generate a new matrix Mn×n by using the
3(b). Similarly, apply Algorithm 2 to obtained the chaotic sequence obtained in Step-5.
two new matrices P1 and P2 . Step 7. To reduce the correlation between the adja-
Step 4. After that for image concatenation process, cent pixels, apply the Arnold map scrambling on matrix
apply the XOR operation between these two new matri- Mn×n to obtain the another matrix M1 .
ces P1 and P2 to obtained the matrix Sn×n . Step 8. Initiate the diffusion process with the imple-
Step 5. The iterative process of the chaotic map uti- mentation of Hessenberg decomposition on the M1 and
lizes the initial values to generate the sequences by S separately to obtain upper Hessenberg and orthogo-
Eq. (14). nal matrices P1 , H1 and P, H respectively.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10596 V. Verma et al.

Step 9. The cipher image C is obtained by applying original counterparts, demonstrating the algorithm’s
inverse Hessenberg decomposition on I mg1 and P1 , effectiveness in both encryption and decryption.
where I mg1 = H1 + H .

3.5.2 Decryption algorithm 4.1 Key space analysis

The decryption process entails the sequential execution Key-space analysis assesses the strength of a crypto-
of the processes involved in the encryption process, but graphic system by assessing the total number of pos-
in reverse order, while utilizing the same set of keys sible keys that can be used in the algorithm [24].
that were generated earlier. The specific details are as Typically, a larger key space indicates a more secure
follows: algorithm, making key-space analysis an indispens-
Step 1. Load the cipher image Cn×n . able instrument for evaluating the security of cryp-
Step 2. Extract the correct decryption keys K . tographic systems. Based on the analysis, a key that
Step 3. Generate the pseudo random sequence using is longer than 2128 is considered to provide a high
3D-ICPCM, according to the Eqs. (7)–(9). level of encryption security [49]. According to this, we
Step 4. Generate the matrix M1 by using the Eq. (14). assume eight floating numbers with 10−15 computer
Step 5. Implement the inverse 2D Arnold scrambling precision. In this proposed image encryption method,
on M1 to obtain the matrix M2 . the key K = (K 1 , K 2 , K 3 ) is used for deciphering at
Step 6. Apply the inverse Hessenberg decomposition the receiver end. Here, the subkey K 1 contains the hash
on cipher image C and M2 obtained in Step-1 and Step- value obtained from SHA-256 which is used in the gen-
5 to further decrypt the encrypted image. eration of chaotic map’s initial values. So, the size of
Step 7. Apply inverse bit scrambling techniques subkey K 1 generated by SHA-256 is 2256 bits. Further,
SPBS and BRMC respectively on the image obtained the size of subkey K 2 generated by control parameter μ
in Step 6, to restore the original values of the image of the proposed 3D-ICPCM map is 1015 . Also, if each
pixels. entry of square image of size N × N is 2r1 bits then the
2
Step 8. In case of color images, concatenate all the size of subkey K 3 is 2r1 N . Hence, the total size of the
components after performing component wise decryp- key space is:
tion process to obtain the original color images.
2256 × 1015 × 2(r1 N
2)
≈ 2256 × 249 × 2(r1 N
2)

which is higher than 2128 . This infers that the model

4 Simulation and security analysis under consideration is resilient enough to withstand
statistical attacks.
The proposed image encryption and decryption algo-
rithm based on 3D-ICPCM and Hessenberg decompo-
sition have been tested against various types of robust 4.2 Key sensitivity analysis
attacks. The proposed algorithms have been imple-
mented on both gray and colour images (Fig. 9). The degree to which the accompanying cipher image
For experimental analysis, the control parameters of or reconstructed image changes when the encryption
proposed 3D chaotic map is set as μ = 0.98 and the ini- or decryption key is significantly modified is consid-
tial values of the chaotic map is obtained from the origi- ered one of the most fundamental parameters when
nal images by a hashing algorithms. We select these ini- assessing image encryption techniques. The greater the
tial values as a secret key for encryption and decryption degree of change, the greater the sensitivity of the key
methodology. The reliability of the proposed algorithm [50]. With a plain image, SHA-256 generates the ini-
was investigated by drawing examples from the USC- tial key for this algorithm. After executing a number of
SIPI image database collection [48]. Figure 10 presents operations on the key, the initial values of 3D-ICPCM
a display of original, encrypted, and decrypted images. chaotic map are obtained. The next step in assessing
This comparison confirms that the encrypted images key sensitivity involves decrypting the cipher image
effectively obscure any discernible details of the origi- encrypted with the slightly modified keys. Figure 11
nals, while the decrypted images closely resemble their represents the decrypted images corresponding to the

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10597

Fig. 10 Visual analysis:

a–c original image of
Boat.512, 7.1.05, 4.2.03;
d–f cipher image of
Boat.512, 7.1.05, 4.2.03;
g–i decipher image of
Boat.512, 7.1.05, 4.2.03

cipher images with the original key and altered keys. 08505930ea2b35681bd7b54d53d29ec12490931bb2f
Thus, making a slight modification to the correct key 355b83d1bcddd97fccad3
used to decrypt the cipher image does not provide any Wrong key with change at first position in key K 1 :
information about the original image. Let us consider 78505930ea20a5681bd7b54d53d29ec12490931bb2f
“Boat.512” as an illustration: 355b83d1bcddd97fccaeb
Correct key K 1 : Wrong key with change at first position in key K 1 :
08505930ea20a5681bd7b54d53d29ec12490931bb2f3 98505930ea20a5681bd7b54d53d29ec12490931bb2f
55b83d1bcddd97fccaeb 355b83d1bcddd97fccaeb
Wrong key with change at last position in key K 1 : Wrong key with change at first position in key K 1 :
08505930ea20a5681bd7b54d53d29ec12490931bb2f 18505930ea20a5681bd7b54d53d29ec12490931bb2f
355b83d1bcddd97fccae3 355b83d1bcddd97fccaeb
Wrong key with change at second last position in Wrong key with change μ = 0.98 + 10−15 in the
key K 1 : key K 2 , where μ is the
08505930ea20a5681bd7b54d53d29ec12490931bb2f parameter of the 3D-ICPCM map.
355b83d1bcddd97fcca1b
Wrong key with changes at last and second last
position in key K 1 :

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10598 V. Verma et al.

Fig. 11 a Encrypted image;

b decrypted image with the
correct key K 1 ; c–e
decrypted Boat.512 images
with changes in the correct
key K 1 at last position;
second last position; last as
well as second last positions
respectively; f–h decrypted
Boat.512 images with
change in the correct key
K 1 at the first position; i
decrypted Boat.512 image
with change in the correct
key K 2

4.3 Histogram analysis quantitatively analyze the evenness in histogram of the

cipher images and quantify the ability of the encryption
A histogram demonstrates how the intensity values of algorithm to resist statistical analysis attacks. However
each pixel are distributed out statistically. The origi- the following formula has been employed to assure his-
nal image’s histogram reveals a non-uniform distribu- togram uniformity [37]:
tion of intensity values, while the encrypted image’s
histogram reflects the hidden effects of the encryption 
n
(xk − yk )2
scheme [51]. Despite also showing a non-uniform dis- χ2 = (15)
yk
tribution, the encrypted image’s histogram exhibits a k=1
distinct pattern resulting from the encryption process.
The way curves of the histograms were uniformly and where yk = M N
256 and x k symbolize to the expected
smoothly distributed out made sure that the encryp- and obtained frequencies respectively of every possi-
tion algorithm could withstand statistical attacks [52]. ble pixel value in the image, for k = 1, 2, 3, 4, 5.....255
Figure 12 shows the histogram of original grayscale maximum pixel intensity. When the ciphertext image’s
and color images and corresponding encrypted images chi-square value is less than χ 2 (α=0.01) (255) =
respectively. 310.4574, χ 2 (α=0.1) (255) = 293.2478, χ 2 (α=0.05)
In order to further analyze the distribution of the (255) = 284.3359 [24]. It signifies that the chi-square
pixels’ tonal values, numerically chi-square test is to test is passed. Table 2 displays the χ 2 results for
the encrypted test images acquired by using proposed

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10599

Fig. 12 Histogram of grayscale and color test images

Table 2 Chi-square values for test images at significance level Table 3 Calculated entropy values of test images
(α = 0.05)
Images Gray/color Plain image Cipher Image
Obtained chi-square values of propounded algorithm
Test images Plaintext image Cipher image Result 4.2.01 Color 6.6530 7.9993
4.2.03 Color 7.6244 7.9992
Boats.512 435888.3417 256.26 Pass
4.2.07 Color 7.2978 7.9993
7.1.04 105290.26484 243.2988 Pass
7.1.02 Gray 4.0045 7.9993
7.1.05 557365.5410 244.6712 Pass
7.1.04 Gray 6.1074 7.9993
4.2.01 951969.3020 280.0117 Pass
7.1.05 Gray 6.5632 7.9993
4.2.03 1018634.615 273.1426 Pass
Boat.512 Gray 7.0333 7.9993
5.3.01 Gray 7.5237 7.9998
5.1.10 Gray 7.3118 7.9974
scheme. This table illustrates that the chi-square values
for all encrypted images are below the optimal thresh-
old, indicating that the resulting images possess a uni-
form histogram and successfully pass this test. calculated as follows:


K −1
4.4 Information entropy H (Y ) = − p(yi ) log2 p(yi ) (16)
i=0
Information entropy measures the degree of random-
ness and uncertainty in the structure of a dataset related where K represents the total number of possible sym-
to a random variable [53]. Mathematically, the infor- bols within the sample space, and p(yi ) indicates the
mation entropy H (Y ) of a random variable Y can be probability of the symbol yi occurring [26]. For a

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10600 V. Verma et al.

Table 4 Calculated LSE values of test images selected as in Eq. (17).

  
Images Components Cipher image k
H PBi
Hk,TB (P) = (17)
Boat.512 – 7.9029 k
i=1
7.1.02 – 7.9034  
7.1.04 – 7.9026
where H PBi represents the information entropy of
7.1.05 – 7.9032
non overlapping blocks Pi , k denotes the total number
of blocks, and TB depicts the total number of blocks.
R 7.9024
For an encrypted image to satisfy the LSE require-
4.2.03 G 7.9030
ments, its LSE values must fall within the specified
B 7.9019
range of (7.901515698, 7.903422936). The required
range is determined based on a significance level of α =
0.001 along with the parameters (k, TB ) = (30, 1936).
The results of calculated LSE values at the sig-
cryptosystem to be considered ideal, the entropy of
nificance level α = 0.001 are displayed in Table 4.
encrypted data should be near to log2 K bits. This will
Analysis of Table 4, demonstrates that the proposed
result in an uniform histogram, ensuring the security of
method assures a substantial degree of randomization
the information content.
and uniform distribution across localized segments of
Table 3 displays the entropy values derived for both
the cipher image.
initial and cipher images using the proposed algorithm.
Table 5 framed the comparative analysis between
In theory, the maximum entropy value achievable for an
the erst-while algorithms and propounded algorithm
encrypted grayscale image is 8, serving as a benchmark
for grayscale and color images. As demonstrated by
for evaluating the diffusion effectiveness of an image
the values in bold, it has been observed that the pro-
encryption system. The closer the entropy value is to
pounded model gives better results in comparison to
this theoretical maximum, the more efficient the dif-
other existing image encryption algorithms.
fusion process is considered. In the context of cipher
images generated through our proposed method, the
information entropy values span from 7.9971 to 7.9998, 4.5 Adjacent pixel correlation analysis
indicating a high level of diffusion performance.
The Local Shannon Entropy (LSE) is a statistical In the original image, pixels often exhibit a strong pos-
measure utilized to assess the degree of randomness itive correlation, nearing a value near 1, which gives
or uncertainty in a digital image. The randomness of the plain image its coherent and meaningful aspect.
pixel values throughout the image is evaluated through An effective encryption algorithm should aim to sig-
the computation of entropy for each pixel, which is nificantly reduce this correlation among pixels, ideally
determined by the probability distribution of gray lev- bringing it close to zero. The decorrelation process adds
els in the pixel’s immediate surroundings [54]. Entropy a noise-like or textured characteristic to the image, indi-
is more significant for pixels distinguished by a high cating that the encryption has succeeded [34,60]. To
degree of randomness; conversely, pixels that display calculate the correlation between pixels, we employ
more predictable patterns have a lower entropy value. the following procedure: randomly pick the 4000 pairs
For an encrypted image, it is crucial that the pixel dis- of adjacent pixels in the vertically, horizontally, and
tribution remains uniform, not just across the whole diagonally directions, then the correlation is calculated
image, but also within its distinct, non-overlapping seg- in each direction by using Eq. (18)
ments. This uniformity ensures that any part of the
Cov( p, q)
image, when randomly picked up, should reflect the r pq = √ √ (18)
similar level of uniform distribution as observed in the D( p) D(q)
entire image. The LSE is a valuable tool for verifying When r pq approaches to 1, it indicates a strong cor-
the encryption integrity of a digital image by examin- relation between adjacent pixels, whereas as its value
ing its distribution and randomness patterns. The LSE nearing to 0 signifies a weak correlation.
is subsequently computed by averaging the informa- Figure 13 shows the scatter plots of test images. The
tion entropy values of the blocks that were randomly linearity of relationship between pixels in the original

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10601

Table 5 Comparative analysis of Information entropy for grayscale and color images
Images Color Components Proposed Algo. [55] [56] [57] [52] [38] [58] [59]

7.1.01 Gray – 7.9993 7.9992 7.9990 7.9993 7.9992 – – –

7.1.02 Gray – 7.9993 7.9993 7.9991 7.9991 7.9992 – – –
7.1.03 Gray – 7.9993 7.9992 7.9990 7.9992 7.9993 – – –
7.1.04 Gray – 7.9993 7.9992 7.9992 7.9992 7.9993 – – –
7.1.05 Gray – 7.9993 7.9993 7.9992 7.9991 7.9992 – – –
7.1.08 Gray – 7.9993 7.9991 7.9990 7.9990 7.9991 – – –
Boat.512 Gray – 7.9993 7.9991 7.9992 7.9991 7.9992 – – –
5.3.01 Gray – 7.9998 7.9998 7.9998 7.9998 7.9998 – – –
5.1.10 Gray – 7.9974 7.9974 7.9974 7.9969 7.9975 – – –
R 7.9992 – – – – 7.9923 7.9991 7.9914
4.2.03 Color G 7.9992 – – – – 7.9802 7.9991 7.9914
B 7.9993 – – – – 7.9986 7.9993 7.9915
R 7.9993 – – – – 7.9962 7.9989 7.9911
4.2.07 Color G 7.9993 – – – – 7.9928 7.9991 7.9912
B 7.9993 – – – – 7.9724 7.9989 7.9915

image confirms the strong correlation between adja- 

0, i f S1 (i, j) = S2 (i, j)
cent pixels. However, symmetrically dispersed scat- E(i, j) =
tered plots have infiltrated the low correlation in the pix- 1, i f S1 (i, j) = S2 (i, j)
els of the encrypted image. This suggests that the pro- (19)
posed algorithm is more effective at resisting against 
M 
N
the cyberpunks’ statistical attacks. E(i, j)
i=1 i=1
The mean correlation coefficient score for images N PC R(S1 , S2 ) = ξ = × 100%
is indicated in Table 6. By reducing the correlation M×N
(20)
between adjacent pixel values, its proximity to 0 and
negative values ensure the algorithm’s resilience to sta- 1
U AC I (S1 , S2 ) = δ =
tistical attacks by diminishing the correlation between M×N

M 
N
adjacent pixel values. |S1 (i, j) − S2 (i, j)|
Table 7 demonstrates the comparative examina- 255
i=1 i=1
tion of correlation coefficients with existing models.
Clearly, the proposed model exhibits more resilience × 100% (21)
against statistical attacks than previous state-of-the-art
models. where M × N is size the of image, E(i, j) is the differ-
ence between two encrypted images S1 and S2 . Any
4.6 Resistance differential attack analysis scheme pass this test only if obtained ξ > ξα∗ and
δ ∈ (δα∗− , δα∗+ ) at α level of significance.
In this analysis, the sensitivity of an encryption algo- Theoretical values of N PC R(ξ ∗ ) and U AC I (δ ∗ ) are
rithm is investigated by comparing the two encrypted given as follows [51]:
images with or without one bit change [51]. The num- For 256 × 256 pixels image ξ0.05 ∗ = 0.9957 and δα∗− =
ber of pixels change rate (N PC R) and unified average ∗+
0.3328, δα = 0.33647
change intensity (U AC I ) tests are employed to assess For 512 × 512 pixels image ξ0.05 ∗ = 0.9959 and δα∗− =
the ability of an image encryption scheme to resist dif- ∗+
0.3337, δα = 0.3355
ferential attack. The definition of NPCR and UACI can For 1024 × 1024 pixels image ξ0.05 ∗ = 0.9960 and
be described as follows [63]: ∗− ∗+
δα = 0.3342, δα = 0.3351

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10602 V. Verma et al.

Fig. 13 Correlation scatter plots: first row original coefficients in horizontal, vertical, diagonal directions for image 4.2.03 and Boat.512
respectively; second row corresponding encrypted correlation coefficients in each direction respectively

Table 6 Correlation values of grayscale and color images

Image Color Plain Image Cipher image
Horizontal Vertical Diagonal Horizontal Vertical Diagonal

4.2.01 Color 0.9541 0.9941 0.9497 −0.0183 −0.0150 −0.0266

4.2.03 Color 0.8942 0.8391 0.8116 −0.0263 −0.0368 −0.0281
7.1.02 Gray 0.9334 0.9493 0.8716 −0.0014 −0.0061 −0.0160
7.1.03 Gray 0.9486 0.9199 0.8934 −0.0229 −0.0041 −0.0039
7.1.04 Gray 0.9794 0.9681 0.9572 −0.0058 −0.2934 −0.0205
7.1.05 Gray 0.9525 0.9233 0.9108 −0.0343 −0.0014 −0.0118
7.1.08 Gray 0.9562 0.9384 0.9312 −0.0300 −0.0038 −0.0299
Boat.512 Gray 0.9781 0.9839 0.9736 −0.0156 −0.0051 −0.0088

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10603

Table 7 Comparative analysis of correlation analysis for test images

Images Directions Proposed algo. [52] [61] [55] [62]

H −0.0156 0.0034 0.00248 −0.0045 0.00340

Boat.512 V −0.0051 0.0237 −0.0653 −0.0011 0.00191
D −0.0088 0.0055 −0.0036 −0.0113 0.00176
H −0.0058 0.0101 −0.003117 −0.0005 –
7.1.04 V −0.0293 −0.0139 0.007290 0.0026 –
D −0.0205 −0.0040 0.000242 0.0030 –
H −0.0343 −0.0195 −0.00404 −0.0126 –
7.1.05 V −0.0014 0.0016 −0.00038 0.0044 –
D −0.0118 −0.0154 −0.00019 0.0004 –
H −0.0263 – – – 0.00113
4.2.03 V −0.0368 – – – 0.00098
D −0.0281 – – – 0.00213

Table 8 Obtained NPCR and UACI values of test images

Test images Size NPCR UACI Result rate (in %)

4.2.01 512 × 512 0.9961 0.3350 100

4.2.03 512 × 512 0.9961 0.3346 100
7.1.04 512 × 512 0.9961 0.3354 100
7.1.05 512 × 512 0.9960 0.3348 100
7.1.07 512 × 512 0.9960 0.3354 100
7.1.08 512 × 512 0.9960 0.3371 100
Boat.512 512 × 512 0.9960 0.3349 100
5.3.01 1024 × 1024 0.9960 0.3349 100

Table 9 Comparative analysis of NPCR values for test images

Images Components Proposed Algo. [55] [56] [57] [52] [38] [58] [59]

7.1.01 – 99.602 99.609 99.59 99.586 99.622 – – –

7.1.04 – 99.612 99.597 99.62 99.587 99.612 – – –
7.1.05 – 99.603 99.612 99.61 99.599 99.613 – – –
Boat.512 – 99.605 99.615 99.61 99.603 99.609 – – –
5.3.01 – 99.605 99.612 99.62 99.605 99.605 – – –
R 99.609 – – – – 99.624 99.789 99.605
4.2.03 G 99.606 – – – – 99.591 99.785 99.616
B 99.608 – – – – 99.597 99.790 99.625
R 99.621 – – – – 99.631 99.847 99.609
4.2.07 G 99.612 – – – – 99.601 99.838 99.621
B 99.611 – – – – 99.638 99.847 99.594

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10604 V. Verma et al.

Table 10 Comparative analysis of UACI values for test images

Images Components Proposed Algo. [55] [56] [57] [52] [38] [58] [59]

7.1.01 – 33.519 33.419 33.25 33.448 33.502 – – –

7.1.04 – 33.472 33.424 33.21 33.484 33.464 – – –
7.1.05 – 33.488 33.423 33.21 33.485 33.453 – – –
Boat.512 – 33.451 33.435 33.42 33.499 33.463 – – –
5.3.01 – 33.490 33.445 33.31 33.471 33.492 – – –
R 33.514 – – – – 33.684 33.344 33.447
4.2.03 G 33.441 – – – – 31.955 33.343 33.449
B 33.432 – – – – 33.414 33.348 33.440
R 33.405 – – – – 33.557 33.329 33.501
4.2.07 G 33.477 – – – – 33.721 33.328 33.441
B 33.510 – – – – 33.570 33.337 33.453

Table 11 Calculated SSIM score of test images of correlation, contrast, and distortion of brightness
Test images Gray/color Original vs cipher Original vs decrypted between two variables which in the case are the ini-
tial and the decrypted image [64]. For an algorithm to
4.2.01 Color 0.0075 1 attain a desirable SSIM score, it must not only effi-
4.2.03 Color 0.0098 1 ciently eliminate noise but also preserve the integrity
7.1.04 Gray 0.0106 1 of the object edges present in the image. The standard
7.1.05 Gray 0.0105 1 SSIM values ranges form to 0 to 1. Thereby, SSIM
Boat.512 Gray 0.0098 1 value close to 0 indicates that the two images are quite
5.3.01 Gray 0.0084 1 distinct from one another. An SSIM value close to 1
quantify that the two images are identically same [65].
The calculative formula for SSIM is as mentioned:
The N PC R and U AC I calculated values of the test
images are illustrated in Table 8. From the calculated
(2μ p μq + C1 )(2σ pq + C2 )
values of N PC R(ξ ) and U AC I (δ) of each test image SS I M( p, q) =
show that the proposed scheme performs satisfactorily. (μ2p + μq2 + C1 )(σ p2 + σq2 + C2 )
Thus, the proposed model is resistant to differential (22)
attacks and highly susceptible to even minor alterations
in the plain image. where C1 and C2 are constants used to stabilize images,
Tables 9 and 10 summarise a comparison of the μ p , μq , σ p , σq , σ pq indicates the average pixel values,
proposed model of N PC R and U AC I values, respec- variance and covariance of the plain image and cipher
tively, for test images with the existing models. There- image, respectively. Getting a value of 1 in Table 11
fore, the N PC R and U AC I values calculated from the signifies that the characteristic similarity between both
proposed model fall within the critical range, indicating the encrypted and plain images in terms of all of their
that the suggested model is more resilient to differential defining features. Also, in addition the structural dis-
attacks compared to the benchmark models. similarity between cipher images and plain images has
been validated for the propounded scheme by SSIM
values close to 0. Thus, it depicts difference in between
4.7 Structural similarity index metric (SSIM) the pixel value of cipher images and plain images. Con-
sequently, it becomes difficult for the invaders to access
The Structural Similarity Index (SSIM) is designed to the initial plane image. Hence, the propounded model
evaluate the quality of reconstructed images, includ- for image encryption is feasible and efficient against
ing important factors such as similarity of edges, loss statistical attacks.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10605

Fig. 14 Clipping attack

analysis: The first row from
a–c is the encrypted 4.2.03
image with different
cropping; the second row
from d–e is corresponding
decipher images

4.8 Robust attack analysis image. Effective image encryption techniques must be
resilient against noise attacks [22,66]. To assess this
4.8.1 Cropping attack analysis resilience speckle noise, salt and pepper noise, Gaus-
sian noise are added to encrypted image. Decryption is
The transmission of digital images via the internet then attempted after applying noise at a level of 0.01,
is susceptible to data loss, which may occur due with outcomes presented in Fig. 15. The findings show
to network issues or intentional interference by an that the suggested approach effectively mitigates salt
adversary aiming to compromise the image’s integrity and pepper, speckle, and Gaussian noise. This demon-
[10]. An encryption algorithm must possess suffi- strates that the 3D-ICPCM image encryption algorithm
cient robustness to guarantee that, regardless of any is resilient against noise attacks.
potential interference during transmission, the recip-
ient can recover visually coherent information upon
decrypting the image. Figure 14 illustrates a occlu-
4.9 Analysis of algorithmic complexity
sion attack on encrypted images. Even after the loss
of one-third, one-fourth, and one-half of the data in
This approach evaluates an algorithm’s effectiveness
the cipher image, it has been discovered that the pro-
based on the amount of time and memory it consumes
posed model’s decryption procedure can still retrieve
to address a problem. The proposed algorithm is simu-
the plain image, although with some data loss [21].
lated in an environment equipped with an i5 − 7200U
Hence, when the encrypted image is subjected to a crop-
CPU @2.5G H z and 8G B of RAM. In the proposed
ping attack during transmission, the suggested model
algorithm, Process 1- is the time complexity of origi-
provides enhanced security and may effectively with-
nal image representing coefficient which is O(N × N ).
stand the cropping attack.
Process 2- SHA-256 hash function for generating the
initial values of chaotic map and its time complexity
4.8.2 Noise attack analysis is O(N ) because we generate separate keys for dif-
ferent images. Process 3- Iterates the 3D chaotic map
Noise can disrupt image transmission, making it diffi- to generate the chaotic sequences whose time com-
cult to recover the original image from the noisy cipher plexity is O(3 × N × N ). Process 4- BRMC algo-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10606 V. Verma et al.

Fig. 15 “Boat.512”; the

first row from a–c cipher
images with a noise
intensity of 0.01 using
speckle noise, salt and
pepper noise, and Gaussian
noise respectively; the
second row from d–f is
corresponding decipher
images respectively

rithm applies on the initial image and its time com- with Hessenberg decomposition. The integration of the
plexity is O(N × N × 16). Process 5- SPBS technique 3D-ICPCM enhances the randomness of the proposed
applies on the plain image and its time complexity algorithm, impacting both position and bit levels signif-
is O(N × N × dm 2 ). Process 6- Arnold scrambling icantly. In the diffusion process, the use of Hessenberg
applies on the matrix generated by chaotic sequences decomposition adds an extra level of security to the pro-
and the time complexity of scrambling process is posed encryption algorithm. Also, the Hash function
O(n 0 × N × N ) where n 0 is the number of iterations, SHA-256 has been utilized for secret key generation in
here n 0 is set at 36 so complexity is O(36 × N × N ). the encryption algorithm, which enhances the security
Process 7- The Hessenberg decomposition is used to of the image. The simulation results validate the ability
decompose the image and its time complexity depends of the proposed model to withstand differential and var-
upon the size of image(matrix) being decomposed so ious statistical attacks. Moreover, the comparative anal-
the time complexity of Hessenberg decomposition is ysis demonstrates that the proposed model surpasses
O(N × N ). Thus, the proposed encryption scheme for existing ones regarding security and reliability. It has
plaintext images has a time complexity of O(N 2 ). been observed that the proposed encryption algorithm
is susceptible to data loss when encrypted image is sub-
jected to higher noise intensities. In future, we aim to
enhance the resilience of the proposed encryption algo-
5 Conclusions and future work
rithm against noise, by minimizing the data loss occur
during the decryption procedure, Furthermore, we will
In this study, firstly, a novel 3D-ICPCM chaotic map
focus on using quantum circuits to optimize encryption
has been introduced to overcome the constraints of ran-
efficiency.
domness and key space observed in the Ikeda chaotic
map and the 1D-CPCM chaotic map. It has been ana- Acknowledgements The first author Vivek Verma, is thankful
lyzed that 3D-ICPCM shows an improved hyperchaotic to Science and Engineering Board (SERB), New Delhi, Govern-
nature, which has been assured by positive values of ment of India, for the financial support for carrying out this work
LE, bifurcation diagrams, trajectory analysis, and NIST under SERB project core grant number CRG/2020/002040.
tests. Secondly, a novel image encryption algorithm
has been introduced based on 3D-ICPCM combined

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10607

Funding The work is supported by the Science and Engineering 11. Liang, Q., Zhu, C.: A new one-dimensional chaotic map
Board (SERB), New Delhi, Government of India, under Grant for image encryption scheme based on random dna coding.
number CRG/2020/002040. Optics & Laser Technology 160, 109033 (2023). https://doi.
org/10.1016/j.optlastec.2022.109033
Data availability statement The study’s supporting data can be 12. Kalpana, J., Murali, P.: An improved color image encryption
found in USC-SIPI (https://sipi.usc.edu/database). based on multiple dna sequence operations with dna syn-
thetic image and chaos. Optik 126(24), 5703–5709 (2015).
Declarations https://doi.org/10.1016/j.ijleo.2015.09.091
13. Abdelfatah, R.I., Saqr, H.M., Nasr, M.E.: An efficient med-
Competing interests The authors declare no competing inter- ical image encryption scheme for (wban) based on adaptive
ests. dna and modern multi chaotic map. Multimedia Tools and
Applications 82(14), 22213–22227 (2023). https://doi.org/
10.1007/s11042-022-13343-8
14. SaberiKamarposhti, M., Sahlabadi, M., Lin, C.-C.,
References Muniyand, R.C.: Using 2d hénon map, cycling chaos and
dna sequence for new secure color image encryption algo-
1. Zhang, F., Kodituwakku, H.A.D.E., Hines, J.W., Coble, J.: rithm. Arabian Journal for Science and Engineering, 1–13
Multilayer data-driven cyber-attack detection system for (2023) https://doi.org/10.1007/s13369-023-08298-3
industrial control systems based on network, system, and 15. Abduljabbar, Z.A., Abduljaleel, I.Q., Ma, J., Al Sibahee,
process data. IEEE Transactions on Industrial Informatics M.A., Nyangaresi, V.O., Honi, D.G., Abdulsada, A.I., Jiao,
15(7), 4362–4369 (2019). https://doi.org/10.1109/TII.2019. X.: Provably secure and fast color image encryption algo-
2891261 rithm based on s-boxes and hyperchaotic map. IEEE Access
2. Chen, I.-T., Tsai, J.-M., Chen, Y.-T., Lee, C.-H.: 10, 26257–26270 (2022). https://doi.org/10.1109/ACCESS.
Lightweight mutual authentication for healthcare iot. Sus- 2022.3151174
tainability 14(20), 13411 (2022). https://doi.org/10.3390/ 16. Huang, L., Li, W., Xiong, X., Yu, R., Wang, Q., Cai, S.:
su142013411 Designing a double-way spread permutation framework uti-
3. Xian, Y., Wang, X., Teng, L.: Double parameters fractal sort- lizing chaos and s-box for symmetric image encryption.
ing matrix and its application in image encryption. IEEE Optics Communications 517, 128365 (2022). https://doi.
Transactions on Circuits and Systems for Video Technology org/10.1016/j.optcom.2022.128365
32(6), 4028–4037 (2021). https://doi.org/10.1109/TCSVT. 17. Li, K., Bao, H., Li, H., Ma, J., Hua, Z., Bao, B.: Memris-
2021.3108767 tive rulkov neuron model with magnetic induction effects.
4. Xian, Y., Wang, X., Teng, L., Yan, X., Li, Q., Wang, X.: IEEE Transactions on Industrial Informatics 18(3), 1726–
Cryptographic system based on double parameters fractal 1736 (2021). https://doi.org/10.1109/TII.2021.3086819
sorting vector and new spatiotemporal chaotic system. Infor- 18. Yu, C., Li, H., Wang, X.: Svd-based image compression,
mation Sciences 596, 304–320 (2022). https://doi.org/10. encryption, and identity authentication algorithm on cloud.
1016/j.ins.2022.03.025 IET Image Processing 13(12), 2224–2232 (2019). https://
5. Heron, S.: Advanced encryption standard (aes). Network doi.org/10.1049/iet-ipr.2018.5912
Security 2009(12), 8–12 (2009). https://doi.org/10.1016/ 19. Su, Q., Chen, B.: A novel blind color image watermark-
S1353-4858(10)70006-4 ing using upper hessenberg matrix. AEU-International Jour-
6. Standard, D.E., et al.: Data encryption standard. Federal nal of Electronics and Communications 78, 64–71 (2017).
Information Processing Standards Publication 112 (1999) https://doi.org/10.1016/j.aeue.2017.05.025
7. Takaoglu, M., Ozyavas, A., Ajlouni, N., Takaoglu, F.: Highly 20. Hu, M., Li, J., Di, X.: Quantum image encryption scheme
secured hybrid image steganography with an improved based on 2d s ine 2-l ogistic chaotic map. Nonlinear
key generation and exchange for one-time-pad encryption Dynamics 111(3), 2815–2839 (2023). https://doi.org/10.
method. Afyon Kocatepe Universitesi Fen Ve Muhendislik 1007/s11071-022-07942-1
Bilimleri Dergisi 23(1), 101–114 (2023). https://doi.org/10. 21. Ye, G., Du, S., Huang, X.: Image compression-hiding algo-
35414/akufemubid.1128075 rithm based on compressive sensing and integer wavelet
8. Mohamed, N.A.E.-S., El-Sayed, H., Youssif, A.: Mixed transformation. Applied Mathematical Modelling 124, 576–
multi-chaos quantum image encryption scheme based 596 (2023). https://doi.org/10.1016/j.apm.2023.08.015
on quantum cellular automata (qca). Fractal and 22. Wu, H., Ye, G., Yap, W.-S., Goi, B.-M.: Reversible blind
Fractional 7(10), 734 (2023). https://doi.org/10.3390/ image hiding algorithm based on compressive sensing and
fractalfract7100734 fusion mechanism. Optics & Laser Technology 167, 109755
9. Dehghani, R., Kheiri, H.: Chaotic-based color image encryp- (2023). https://doi.org/10.1016/j.optlastec.2023.109755
tion using a hybrid method of reversible cellular automata 23. Huang, X., Dong, Y., Ye, G., Shi, Y.: Meaningful
and dna sequences. Multimedia Tools and Applications, 1– image encryption algorithm based on compressive sens-
22 (2023) https://doi.org/10.1007/s11042-023-16118-x ing and integer wavelet transform. Frontiers of Com-
10. Musanna, F., Dangwal, D., Kumar, S.: Novel image puter Science 17(3), 173804 (2023). https://doi.org/10.
encryption algorithm using fractional chaos and cellu- 1007/s11704-022-1419-8
lar neural network. Journal of Ambient Intelligence and 24. Su, Y., Wang, X., Unar, S., Zhao, X., Liu, P.: Secure image
Humanized Computing, 1–22 (2022) https://doi.org/10. storage system based on compressed sensing and 2d-sllim
1007/s12652-021-02982-8

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

10608 V. Verma et al.

in cloud environment. Nonlinear Dynamics 111(3), 2779– 39. Wang, Y., Zhao, Q., Zhang, H., Li, T., Xu, W., Liu, S.,
2814 (2023). https://doi.org/10.1007/s11071-022-07930-5 Su, Y.: Optical single-channel color image encryption based
25. Ye, G., Wu, H., Liu, M., Huang, X.: Reversible image-hiding on chaotic fingerprint phase mask and diffractive imaging.
algorithm based on singular value sampling and compressive Applied Optics 62(4), 1009–1018 (2023). https://doi.org/
sensing. Chaos, Solitons & Fractals 171, 113469 (2023). 10.1364/AO.479983
https://doi.org/10.1016/j.chaos.2023.113469 40. Wei, C., Li, G.: A selective image encryption scheme using
26. Musanna, F., Dangwal, D., Kumar, S.: A novel chaos-based licc hyperchaotic system. IET Image Processing 16(12),
approach in conjunction with mr-svd and pairing function 3342–3358 (2022). https://doi.org/10.1049/ipr2.12569
for generating visually meaningful cipher images. Multime- 41. Wang, T., Ge, B., Xia, C., Dai, G.: Multi-image encryp-
dia Tools and Applications 79(33), 25115–25142 (2020). tion algorithm based on cascaded modulation chaotic sys-
https://doi.org/10.1007/s11042-020-09034-x tem and block-scrambling-diffusion. Entropy 24(8), 1053
27. Li, Q., Wang, X., Wang, H., Ye, X., Zhou, S., Gao, S., Shi, (2022). https://doi.org/10.3390/e24081053
Y.: A secure image protection algorithm by steganography 42. Talhaoui, M.Z., Wang, X., Midoun, M.A.: A new one-
and encryption using the 2d-tscc. Chinese Physics B 30(11), dimensional cosine polynomial chaotic map and its use
110501 (2021). https://doi.org/10.1088/1674-1056/abfa01 in image encryption. The Visual Computer 37, 541–551
28. Alghamdi, Y., Munir, A., Ahmad, J.: A lightweight image (2021). https://doi.org/10.1007/s00371-020-01822-8
encryption algorithm based on chaotic map and random sub- 43. Ikeda, K.: Multiple-valued stationary state and its instabil-
stitution. Entropy 24(10), 1344 (2022). https://doi.org/10. ity of the transmitted light by a ring cavity system. Optics
3390/e24101344 communications 30(2), 257–261 (1979). https://doi.org/10.
29. Akraam, M., Rashid, T., Zafar, S., et al.: A chaos-based 1016/0030-4018(79)90090-7
image encryption scheme is proposed using multiple chaotic 44. Murillo-Escobar, D., Murillo-Escobar, M.A., Cruz-
maps. Mathematical Problems in Engineering 2023 (2023) Hernandez, C., Arellano-Delgado, A., Lopez-Gutierrez,
https://doi.org/10.1155/2023/2003724 R.M.: Pseudorandom number generator based on novel
30. Erkan, U., Toktas, A., Enginoglu, S., Akbacak, E., Thanh, 2d henon-sine hyperchaotic map with microcontroller
D.N.: An image encryption scheme based on chaotic loga- implementation. Nonlinear Dynamics, 1–17 (2022) https://
rithmic map and key generation using deep cnn. Multimedia doi.org/10.1007/s11071-022-08101-2
Tools and Applications 81(5), 7365–7391 (2022). https:// 45. Bandt, C., Pompe, B.: Permutation entropy: a natu-
doi.org/10.1007/s11042-021-11803-1 ral complexity measure for time series. Physical review
31. Feng, W., He, Y., Li, H., Li, C.: Cryptanalysis and improve- letters 88(17), 174102 (2002). https://doi.org/10.1103/
ment of the image encryption scheme based on 2d logistic- PhysRevLett.88.174102
adjusted-sine map. Ieee Access 7, 12584–12597 (2019). 46. Yousif, S.F., Abboud, A.J., Alhumaima, R.S.: A new
https://doi.org/10.1109/ACCESS.2019.2893760 image encryption based on bit replacing, chaos and
32. Hua, Z., Zhou, Y.: Image encryption using 2d logistic- dna coding techniques. Multimedia Tools and Applica-
adjusted-sine map. Information Sciences 339, 237–253 tions 81(19), 27453–27493 (2022). https://doi.org/10.1007/
(2016). https://doi.org/10.1016/j.ins.2016.01.017 s11042-022-12762-x
33. Patel, S., Bharath, K., Kumar, R.: Symmetric keys image 47. Huang, T., Xu, J., Yang, Y., Han, B.: Robust zero-
encryption and decryption using 3d chaotic maps with watermarking algorithm for medical images using double-
dna encoding technique. Multimedia Tools and Appli- tree complex wavelet transform and hessenberg decompo-
cations 79, 31739–31757 (2020). https://doi.org/10.1007/ sition. Mathematics 10(7), 1154 (2022). https://doi.org/10.
s11042-020-09551-9 3390/math10071154
34. Rashid, A.A., Hussein, K.A.: Image encryption algorithm 48. Weber, A.G.: The usc-sipi image database: Version 5. http://
based on the density and 6d logistic map. International Jour- sipi.usc.edu/database/ (2006)
nal of Electrical and Computer Engineering 13(2), 1903 49. Musanna, F., Kumar, S.: A novel fractional order chaos-
(2023). https://doi.org/10.11591/ijece.v13i2.pp1903-1913 based image encryption using fisher yates algorithm and
35. Rahul, B., Kuppusamy, K., Senthilrajan, A.: Dynamic dna 3-d cat map. Multimedia Tools and Applications 78, 14867–
cryptography-based image encryption scheme using mul- 14895 (2019). https://doi.org/10.1007/s11042-018-6827-2
tiple chaotic maps and sha-256 hash function. Optik 289, 50. Wang, X., Chen, S.: Chaotic image encryption algo-
171253 (2023). https://doi.org/10.1016/j.ijleo.2023.171253 rithm based on dynamic spiral scrambling transform and
36. Mansouri, A., Wang, X.: A novel block-based image encryp- deoxyribonucleic acid encoding operation. IEEE Access 8,
tion scheme using a new sine powered chaotic map gener- 160897–160914 (2020). https://doi.org/10.1109/ACCESS.
ator. Multimedia Tools and Applications 80, 21955–21978 2020.3020835
(2021). https://doi.org/10.1007/s11042-021-10757-8 51. Rani, N., Mishra, V., Sharma, S.R.: Image encryption model
37. Yavuz, E.: A novel chaotic image encryption algorithm based on novel magic square with differential encoding
based on content-sensitive dynamic function switching and chaotic map. Nonlinear Dynamics 111(3), 2869–2893
scheme. Optics & Laser Technology 114, 224–239 (2019). (2023). https://doi.org/10.1007/s11071-022-07958-7
https://doi.org/10.1016/j.optlastec.2019.01.043 52. Wang, X., Chen, X.: Image encryption algorithm based
38. Wang, X., Qin, X., Liu, C.: Color image encryption algo- on cross-scrambling and rapid-mode diffusion. The
rithm based on customized globally coupled map lattices. Visual Computer, 1–28 (2022) https://doi.org/10.1007/
Multimedia Tools and Applications 78, 6191–6209 (2019). s00371-022-02645-5
https://doi.org/10.1007/s11042-018-6326-5

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Novel image encryption algorithm 10609

53. Li, T., Zhang, D.: Hyperchaotic image encryption based on 62. Zhang, X., Liu, Z., Yang, X.: Fast image encryption algo-
multiple bit permutation and diffusion. Entropy 23(5), 510 rithm based on 2d-fcsm and pseudo-wavelet transform. Non-
(2021). https://doi.org/10.3390/e23050510 linear Dynamics 111(7), 6839–6853 (2023). https://doi.org/
54. Alawida, M.: A novel chaos-based permutation for image 10.1007/s11071-022-08185-w
encryption. Journal of King Saud University-Computer and 63. Alawida, M., Teh, J.S., Samsudin, A., et al.: An image
Information Sciences 35(6), 101595 (2023). https://doi.org/ encryption scheme based on hybridizing digital chaos and
10.1016/j.jksuci.2023.101595 finite state machine. Signal Processing 164, 249–266 (2019).
55. Wang, X., Chen, X., Zhao, M.: A new two-dimensional sine- https://doi.org/10.1016/j.sigpro.2019.06.013
coupled-logistic map and its application in image encryp- 64. Rani, N., Sharma, S.R., Mishra, V.: Grayscale and colored
tion. Multimedia Tools and Applications, 1–37 (2023) image encryption model using a novel fused magic cube.
https://doi.org/10.1007/s11042-023-14674-w Nonlinear Dynamics 108(2), 1773–1796 (2022). https://doi.
56. Zhou, Y., Bao, L., Chen, C.P.: Image encryption using a org/10.1007/s11071-022-07276-y
new parametric switching chaotic system. Signal processing 65. Hore, A., Ziou, D.: Image quality metrics: Psnr vs. ssim. In:
93(11), 3039–3052 (2013). https://doi.org/10.1016/j.sigpro. 2010 20th International Conference on Pattern Recognition,
2013.04.021 pp. 2366–2369 (2010). https://doi.org/10.1109/ICPR.2010.
57. Wu, Y., Yang, G., Jin, H., Noonan, J.P.: Image encryption 579
using the two-dimensional logistic chaotic map. Journal of 66. Musanna, F., Kumar, S.: Generating visually coherent
Electronic Imaging 21(1), 013014–013014 (2012). https:// encrypted images with reversible data hiding in wavelet
doi.org/10.1117/1.JEI.21.1.013014 domain by fusing chaos and pairing function. Computer
58. Valandar, M.Y., Barani, M.J., Ayubi, P.: A fast color image Communications 162, 12–30 (2020). https://doi.org/10.
encryption technique based on three dimensional chaotic 1016/j.comcom.2020.08.008
map. Optik 193, 162921 (2019). https://doi.org/10.1016/j.
ijleo.2019.06.021
59. Zhang, Y.-Q., He, Y., Li, P., Wang, X.-Y.: A new color image Publisher’s Note Springer Nature remains neutral with regard
encryption scheme based on 2dnlcml system and genetic to jurisdictional claims in published maps and institutional affil-
operations. Optics and Lasers in Engineering 128, 106040 iations.
(2020). https://doi.org/10.1016/j.optlaseng.2020.106040
60. Ding, L., Ding, Q.: A novel image encryption scheme Springer Nature or its licensor (e.g. a society or other partner)
based on 2d fractional chaotic map, dwt and 4d hyper- holds exclusive rights to this article under a publishing agreement
chaos. Electronics 9(8), 1280 (2020). https://doi.org/10. with the author(s) or other rightsholder(s); author self-archiving
3390/electronics9081280 of the accepted manuscript version of this article is solely gov-
61. Wang, X., Guan, N., Yang, J.: Image encryption algorithm erned by the terms of such publishing agreement and applicable
with random scrambling based on one-dimensional logis- law.
tic self-embedding chaotic map. Chaos, Solitons & Fractals
150, 111117 (2021) mnjkhm25

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”),
for small-scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are
maintained. By accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use
(“Terms”). For these purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or
a personal subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or
a personal subscription (to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the
Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data
internally within ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking,
analysis and reporting. We will not otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of
companies unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that
Users may not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to
circumvent access control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil
liability, or is otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by
Springer Nature in writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer
Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates
revenue, royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain.
Springer Nature journal content cannot be used for inter-library loans and librarians may not upload Springer Nature journal
content on a large scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any
information or content on this website and may remove it or features or functionality at our sole discretion, at any time with or
without notice. Springer Nature may revoke this licence to you at any time and remove access to any copies of the Springer Nature
journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express
or implied with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or
warranties imposed by law, including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be
licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other
manner not expressly permitted by these Terms, please contact Springer Nature at

onlineservice@springernature.com