Bulletin of Electrical Engineering and Informatics

Vol. 12, No. 3, June 2023, pp. 1607~1619

ISSN: 2302-9285, DOI: 10.11591/eei.v12i3.4728  1607

Colour image encryption based on hybrid bit-level scrambling,

ciphering, and public key cryptography

Ahmed Kamil Hasan Al-Ali, Jafaar Mohammed Daif Alkhasraji

Department of Electromechanical Engineering, University of Technology, Baghdad, Iraq

Article Info ABSTRACT

Article history: This paper proposes an image encryption technique using three stages
algorithms based on hyper-chaotic maps. In the first scenario, bit-level
Received Sep 8, 2022 scrambling (BLS) using a 2D coupled chaotic map (2D-CCM) is used to
Revised Nov 2, 2022 encrypt the bits of the basic colour image. In the second strategy, the
Accepted Dec 9, 2022 scrambled bit level is XORed with pseudo random bit generator (PRBG).
The PRBG is designed using a combination of chaotic maps, including,
logistic map (LM), sine map (SM), 5D chaotic map (5D-CM), enhanced
Keywords: quadratic map (EQM), and 2D henon SM (2D-HSM). The pubic key based
on the Chebyshev polynomial chaotic map is used as the final phase of the
Bit-level scrambling encryption algorithms. The performance analysis of the proposed image
Hyper-chaotic maps encryption technique is validated through various criteria such as
Image encryption fundamental space analysis, correlation coefficient, entropy, the number of
Pseudo random bit generator pixels changes rate (NPCR), and unified average-changing intensity (UACI).
Public key cryptography Also, the obtained results are compared with other recent studies. The
simulation results demonstrated that the proposed technique has robust
security and it provides the image with high protection against various
attacks.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Ahmed Kamil Hasan Al-Ali
Department of Electromechanical Engineering, University of Technology
Baghdad, Iraq
Email: ahmed.k.alali@uotechnology.edu.iq

1. INTRODUCTION
The development of multimedia technologies and computer networks has made transmitting and
sharing data an uncomplicated process through the network [1]. Among these data, is a digital image format.
Such a form is widely used because it is capable of carrying much data and it contains confidential besides
private personal information [2]. However, because of the transmission and sharing of digital images, a third
party could obtain the image or change the data of the original image. Thus, image encryption techniques
have an essential role in protecting digital image transmission from attackers. From this perspective, image
encryption is the process of transforming the original information image into another kind that is completely
different from the original [3]. Although various traditional encryption techniques have been proposed in the
literature, for instance, the data cryptography model, and the algorithm such as Rivest-Shamir-Adleman.
According to the previous studies, these techniques demonstrated a flaw in the application process of image
cryptography because digital images have some essential characteristics, such as a high degree of correlation
and redundancy between neighbouring pixels [4], [5]. Furthermore, the security of the image is based on
designing the public key cryptosystem. In this context, various public key cryptosystem techniques have been
proposed such as knapsack public key [6], public key based on algebraic coding [7], and lattice public key
[8]. In this paper, we focus on the Chebyshev polynomial. The Chebyshev polynomial is based on a chaotic

Journal homepage: http://beei.org

1608  ISSN: 2302-9285

map [9]. To improve the security of the digital image, various techniques have been proposed, such as bit-
level permutation [10], deoxyribonucleic acid (DNA) [11], and chaotic systems [12]. It is worth to meeting
that most image encryption techniques are based on chaotic systems [13]–[15] because of their high
sensitivity to the initial value, ergodicity, and pseudo-randomness. Permutation and diffusion are the two
main processes in a chaotic system. These two processes are repeated many times until the perfect security
level is obtained. Wherein the permutation is used to remove the high correlation between adjacent pixels.
The purpose of diffusion is to modify the values of pixels and this process is obtained using chaotic maps.
The chaotic system can be divided into one-dimensional and high-dimensional chaotic systems. A
one-dimensional chaotic system, including a logistic map (LM) [16], 1D sine map (SM) [17], and iteration
LM [18], were proposed to improve the performance of image security. One advantage of a one-dimensional
chaotic system includes a simple structure and high-efficiency encryption while it suffers from a low level of
security. High-dimensional chaotic systems such as 5D chaotic maps (5D-CM) [19], 3D Lorenz mapping
[20], and 4D chaotic maps with DNA [21]. Although the high-dimensional chaotic maps improve image
security performance, the encryption time of the high-dimensional chaotic systems is long due to their
difficult implementation structure [2].
The contribution of this paper is to encrypt the colour images with high levels of security using
hyperchaotic systems in three scenarios. In the first stage, bit-level scrambling (BLS) is used, where the
index permutation is designed using 2D coupled chaotic map (2D-CCM). In the second phase, the stream
cipher system with the pseudo random bit generator (PRBG) is designed using multi random keys, where
each key is generated using chaotic maps, including LM, SM, 5D-CM, enhanced quadratic map (EQM), and
2D henon SM (2D-HSM). A public key based on the Chebyshev polynomial map is applied in the last step of
encryption. Different intelligibility measurements were used to test the quality of the encrypted colour
images, including correlation coefficient, entropy, number of pixels changes rate (NPCR), unified average
changing intensity (UACI), and histogram analysis.
This paper is organized as follows. In section 2, method is introduced. Pseudo-random bit generation
using hyper-chaotic maps is presented in section 3. The fourth section describes the hybrid image encryption
scheme structure. The hybrid image decryption system structure is illustrated in section 5. Experimental
results are presented in section 6. The key sensitivity analysis is presented in section 7, finally, the conclusion
is depicted in section 8.

2. METHOD
This section presents different types of chaotic maps . These chaotic maps are used to generate the
index permutation and PRBG. We will employ eight chaotic map formulas for investigation purpose. A
detailed description of chaotic maps is presented in the following subsections..

2.1. Logistic map

The LM can determine as [22]:

𝑧𝑖+1 = 4𝜌 𝑧𝑖 (1 − 𝑧𝑖 ) (1)

where 𝜌 is the control factor and it values between 0.89 to 1.

2.2. Sine map

The SM is a one-dimensional chaotic map and it can determine as [22]:

𝑧𝑖+1 = 𝜌 sin (𝜋𝑧𝑖 ) (2)

where 𝜌 is the control parameter and it values between 0.87 to 1.

2.3. 5D chaotic map

The 5D-CM is obtained by combining 3D Lorenz and LMs and it can be defined as [19]:

𝑍𝑖+1 = 𝑎 𝑍𝑖 (1 − 𝑍𝑖 )
𝑅𝑖+1 = 𝑏 𝑅𝑖 𝑆𝑖 − 𝑐 𝑊𝑖
𝑆𝑖+1 = 𝑍𝑖 + 𝑅𝑖 (3)
𝑈𝑖+1 = 𝑅𝑖 + 𝑑 𝑊𝑖
𝑊𝑖+1 = 𝑆𝑖 + 𝑍𝑖 𝑈𝑖

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1609

where 𝑍𝑖 , 𝑅𝑖 , 𝑆𝑖 , 𝑈𝑖 , and 𝑊𝑖 are state variables of the system, a, b, c, and d are system parameters. The
values of a, b, c, and d are 4, 0.5, 0.3, and 0.9, respectively. The initial values of Z(0)=0.9, R(0)=-0.28,
S(0)=0.183, U(0)=0.5, and W(0)=0.57).

2.4. 2D coupled chaotic map

The 2D-CCM can be defined as [23]:

𝑘
𝑍𝑖+1 = sin ( )
𝑠𝑖𝑛(𝑅𝑖 ) (4)
𝑅𝑖+1 = 𝛾 sin (𝜋(𝑍𝑖 + 𝑅𝑖 ))

where the control parameter 𝑘0, 𝛾 (0,1], and the initial value of 𝑅0 ≠ 0.

2.5. Enhanced quadratic map

The EQM can be represented mathematically as [15]:
2
𝑍𝑖+1 = (𝑟 − 25 (𝑟 − 𝑍𝑖 2 ) ) 𝑚𝑜𝑑 1 (5)

where r ∈ [0.05, 4].

2.6. 2D hyper-chaotic map

The 2D-HCM can be defined as [24]:

𝑍𝑖+1 = 𝑎1 + 𝑎2 𝑍𝑖 + 𝑎3 𝑅𝑖
(6)
𝑅𝑖+1 = 𝑏1 + 𝑏2 𝑍𝑖2

where 𝑎1 = 0.2, 𝑎2 = 0.3, 𝑎3 = 0.5, 𝑏1 = −1.7, 𝑎𝑛𝑑 𝑏2 = 3.7.

2.7. 2D henon sine map

The 2D-HSM is defined as [25]:

𝑍𝑖+1 = (1 − 𝑎 𝑠𝑖𝑛 2 (𝑍𝑖 + 𝑅𝑖 ))𝑚𝑜𝑑 1

(7)
𝑅𝑖+1 = 𝑏 𝑍𝑖 𝑚𝑜𝑑 1

2.8. Chebyshev polynomial

The Chebyshev polynomial is defined as [26]:

𝑇𝑛 (𝑥) = (2𝑥 𝑇𝑛−1 (𝑥) − 𝑇𝑛−2 (𝑥))𝑚𝑜𝑑 𝑁 (8)

where 𝑇0 (𝑥) = 1, 𝑇1 (𝑥) = 𝑥, and 𝑁 is the total number of pixels in image. The value of Chebyshev
polynomial (𝑇𝑛 (𝑥)) is between -1 to 1. The semigroup property is one of the most essential properties of the
Chebyshev polynomial and it is used for the construction of the public key. The semigroup property is
defined as:

𝑇𝑠 (𝑇𝑟 (𝑥)𝑚𝑜𝑑 𝑁)𝑚𝑜𝑑 𝑁 = 𝑇𝑟 (𝑇𝑠 (𝑥)𝑚𝑜𝑑 𝑁)𝑚𝑜𝑑 𝑁 (9)

The commute under composition is a property of Chebyshev polynomial and it can be defined as:

𝑇𝑠 (𝑇𝑟 (𝑥)𝑚𝑜𝑑 𝑁)𝑚𝑜𝑑 𝑁 = 𝑇𝑠𝑟 𝑚𝑜𝑑 𝑁 (10)

3. PSEUDO RANDOM BIT GENERATOR (PRBG) BASED ON HYBRID CHAOTIC MAPS

Figure 1 shows the proposed PRBG based on a one, two, and five-dimensional chaotic map. Six
PRBGs are produced in the first using a hyperchaotic map and then the final key is generated by XORed
these keys according to:

𝑃𝑅𝐵𝐺𝑖 = 𝐾1 (𝑖) ⊕ 𝐾2 (𝑖) ⊕ 𝐾3 (𝑖) ⊕ 𝐾4 (𝑖) ⊕ 𝐾5 (𝑖) ⊕ 𝐾6 (𝑖) , 𝑖 = 1,2, . . ,8𝑀𝑁 (11)

Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)
1610  ISSN: 2302-9285

where 𝐾1 𝐾2 𝐾3 𝐾4 𝐾5 and 𝐾6 are the PRBG based on LM, SM, EGM, 2D-HCM, 2D-HSM, and 5D-CM,
respectively. The key space for each key is summarized as: LM (𝑥0 , 𝜌, 𝑦0 , 𝜌), SM (𝑥0 , 𝜌, 𝑦0 , 𝜌), EQM
(𝑥0 , 𝑟, 𝑦0 , 𝑟), 2D-HCM (𝑥0 , 𝑦0 , 𝑎1 , 𝑎2 , 𝑎3 , 𝑏1 , 𝑏2 ), 2D-HSM (𝑥0 , 𝑦0 , 𝑎, 𝑏 ), and 5D-CM
(𝑍0 , 𝑅0 , 𝑆0 , 𝑊0 , 𝑈0 , 𝑎, 𝑏, 𝑐, 𝑑 ). Therefore, there are 32 variables used as a key space to generate PRBG.

Figure 1. The proposed PRBG based on hyberchaotic maps

4. HYBRID IMAGE ENCRYPTION SYSTEM STRUCTURE

Figure 2 shows the proposed image encryption system using three sequential algorithms adopted in
this investigation. The first step of the encryption procedure is the implementation of a bit-level scrambling
algorithm to shuffle the bits of the channel image; the next step is the encryption process using the XOR
ciphering system. Finally, public key cryptography algorithms on the transmitter side were applied to the
output stream bits of the ciphering matrix. A more detailed description of each stage is presented in the
following subsections.

4.1. Bit-level scrambling algorithm (stage one)

The first stage of the encryption process is a BLS algorithm that is used to shuffle the bits of the
channel image (red (R) or green (G) or blue (B) channel) according to the permutation index generated using the
2D-CCM as described in subsection (2.4). Each channel matrix has the size. The exhaustive explication of the
bit-level-scrambling procedures is depicted in Algorithm 1. This algorithm is applied for each channel with its
initial conditions and produced the permutated binary matrix S (j, i), j={1 R, 2 G, and 3 B}, i=1 to 8 MN. The
BLS algorithm required four keys representing the initial conditions of 2D-CCM (𝑘, 𝑅0 , 𝑍0 ) for each channel.

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1611

Figure 2. Proposed image encryption system

Algorithm 1. BLS algorithm

Input: Plain Image A ∈ ℝ3×M×N , k j , γj , R 0,j , Z0,j , j=1,2,3.
Output: Permutated Sequence, S∈ {0,1} of size 3×8MN
Step 1: Reshape A3×M×N to A3×MN .
Step 2: Convert A to binary stream vector, bj = {bj,1 , bj,2 , … . , bj,8MN }.
Step 3: For j=1 to 3 Compute
For i=1 to 8*M*N-1, Compute
𝐤𝐣
𝐙𝐣.𝐢+𝟏 = 𝐬𝐢𝐧 ( )
𝐬𝐢𝐧 (𝐑 𝐣,𝐢 )
𝐑 𝐣,𝐢+𝟏 = 𝛄 𝐬𝐢𝐧(𝛑(𝐙𝐣,𝐢 + 𝐑 𝐣,𝐢 ))
End For
End For
Step 4: [~ PIj]=sort(Zj).
Step 5: For j=1 to 3 Compute
For i = 1 to 8*M*N, compute
S (j, i) = b (1, PIj(i))
End For
End For

4.2. XOR ciphering system (stage two)

The second step of encryption is XOR Ciphering system in which the j-th channel and i-th bit of S (j,
i) (j=1, 2, 3 and i=1,.., 8 MN) is XORed with PRBG key, PRBG (j, i) (KR (i)=PRBG (1, i), KG (i)=PRBG(2,i),
KB (i)=PRBG (3, i), i=1,.., 8 MN). Correspond for j-th channel and produced the j-th ciphered message C (j, i)
according to:

𝐶(𝑗, 𝑖) = 𝑆(𝑗, 𝑖)⨁PRBG(𝑗, 𝑖) (12)

where ⨁ is XOR bitwise operation. The j-th stream binary keys, PRBG (j, i), j=1, 2, 3, i=1,.., 8 MN are
generated using PRBG which is described in section (3), where each channel has self-initial conditions.

4.3. Public key cryptography encryption algorithm (stage three)

This section presents public key cryptography encryption algorithm. Algorithm 2 shows the third
stage public key cryptography encryption algorithm at the transmitter side. The algorithm of public key
cryptography encryption is applied to the output stream bits of the ciphering matrix C (j, i). The objective is
to produce the encrypted color image E (j, n, m), j=1, 2, 3, n=1,.., N, m=1,…, M..
Algorithm 2. Public key cryptography encryption algorithm
Input: C(j,i) ∈ {0,1}, j=1,2,3, i=1,2,..,8NM, large number Q, 𝛼 <Q, secret
keys 𝑠𝑗 , j=1,2,3
Output: E(j,n,m) ∈ ℝ3×𝑁×𝑀 , j=1,2,3, n=1,2,..N, m=1,2..,M.
Step 1: Compute q=log2(Q)
3×𝛽×𝑞
Step 2: Reshape 𝐶 3×8𝑁𝑀 to 𝐶𝑓 , 𝛽 = 8𝑁𝑀/𝑞.

Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)
1612  ISSN: 2302-9285
3×β
Step 3: Convert q bits in Cf from binary to decimal Cd
Step 4: Compute Γj = Tsj (α) mod Q, j=1,2,3
Step 5: For j=1 to 3 Compute
For t=1 to β Compute
e(j,t)=Cd(j,t) Γj mod Q
End For
End For
Step 6: Convert e3×β to binary number B 3×β×q
Step 7: Reshape B 3×β×q to B 3×N×M×8
Step 8: Convert to decimal number E 3×N×M

5. HYBRID IMAGE ENCRYPTION SYSTEM STRUCTURE

Figure 3 shows the proposed image decryption system. The proposed system consists of three stages
of the algorithms. The first stage is the public key cryptography decryption algorithms, the second
deciphering and the third stage is bit-level descrambling (BDS) algorithms. A detailed description of each
stage is presented in the following subsections.

Figure 3. Proposed image decryption system

5.1. Public key cryptography encryption algorithm (stage three)

This section presents public key cryptography encryption algorithm. Algorithm 3 shows the third
stage public key cryptography decryption algorithm at the reception side. The public key cryptography
decryption algorithm is applied to the encrypted colour matrix E 3×N×M to produce the sequence Dp (j, i), j=1,
2, 3, i=1,.., 8 NM..

Algorithm 3. Public key cryptography decryption algorithm

Input: E (j, n, m) ∈ {0,1}, j=1,2,3, n=1,2,..,N, m=1,…,M, large number Q, 𝛼
<Q, secret keys 𝑟𝑗 , j=1,2,3
Output: Dp(j,i) ∈ ℝ3×8𝑁𝑀 , j=1,2,3, i=1,2,..8NM.
Step 1: Compute q=log2(Q)
3×𝛽×𝑞
Step 2: Reshape 𝐸 3×𝑁×𝑀 to 𝐷𝑓 , 𝛽 = 8𝑁𝑀/𝑞.
3×𝛽
Step 3: Convert q bits in Df from binary to decimal 𝐷𝑑
Step 4: Compute Γ𝑗 = 𝑇𝑟𝑗 (𝛼) mod Q, j=1, 2, 3
Step 5: For j=1 to 3 Compute
For t=1 to β Compute
d (j, t)=Ed(j,t)/ Γ𝑗 mod Q
End For
End For
Step 6: Convert 𝑑 3×𝛽 to binary number 𝐵 3×𝛽×𝑞

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1613

Step 7: Reshape 𝐵 3×𝛽×𝑞 to 𝑫3×8𝑁𝑀

𝒑

5.2. XOR deciphering system (stage two)

The second stage of decryption is XOR Deciphering system in which the j -th channel and i-th bit of
Dp (j, i) (j=1, 2, 3 and i=1,.., 8 MN) is XORed with PRBG key, PRBG (j, i) K R (i)= PRBG (1, i), KG
(i)=PRBG (2, i), KB (i)=PRBG (3, i), i=1,.., 8 MN). Correspond for each j -th channel and produced the j-th
deciphered message 𝐷𝑐 (𝑗, 𝑖) according to:

𝐷𝑐 (𝑗, 𝑖) = 𝐷𝑝 (𝑗, 𝑖)⨁𝑃𝑅𝐵𝐺(𝑗, 𝑖) (13)

where ⨁ is XOR bitwise operation. The j-th stream binary keys, PRBG (j, i), j=1, 2, 3, i=1,.., 8 MN are
generated using PRBG which is described in section 3, where each channel has self-initial conditions. We
assume that the synchronization between the keys at the transmitter and receiver is established.

5.3. Bit-level descrambling algorithm (stage one)

The last stage of the decryption process is the BDS algorithm that is used to disorganize the
information of the channel image pursuant to the permutation index yielded using the 2D-CCM as described
in subsection 2.4. The exhaustive explication of the BDS procedures is depicted in Algorithm 4. This
algorithm is applied for each channel with its initial conditions and produces the original decrypted colour
matrix. As for the transmitter side, the BDS algorithm required four keys representing the initial conditions of
2D-CCM (𝑘, 𝑅0 , 𝑍0 ) for each channel.

Algorithm 4. BLD algorithm

Input: The matrix 𝑫𝒄 ∈ 𝐵𝑖𝑛𝑎𝑟𝑦 3×8𝑀𝑁 , 𝑘𝑗 , 𝛾𝑗 , 𝑅0,𝑗 , 𝑍0,𝑗 , j=1,2,3.
Output: Decrypted colour image, D∈ ℝ3×𝑵×𝑴 .
Step 1: For j=1 to 3 Compute
For i=1 to 8*M*N-1, Compute
𝑘𝑗
𝑍𝑗.𝑖+1 = sin ( )
sin (𝑅𝑗,𝑖 )
𝑅𝑗,𝑖+1 = 𝛾 sin(𝜋(𝑍𝑗,𝑖 + 𝑅𝑗,𝑖 ))
End For
End For
Step 4: [~ PIj]=sort(Zj).
Step 5: For j=1 to 3 Compute
For i = 1 to 8*M*N, compute
𝑫𝒍 (j, PIj(i)) = 𝑫𝒄 (1,i)
End For
End For
Step 6: Reshape 𝑫𝒍 𝟑×𝟖𝑴𝑵 to 𝑫𝒍 𝟑×𝑵×𝑴×𝟖
Step 7: Convert to 8 bits binary to Decimal 𝑫𝟑×𝑵×𝑴

6. EXPERIENTIAL RESULTS
Three standard colour images, including (Lena, Peppers, and Barbara) are used as input images to
ensure the efficiency of the proposed image encryption technique. Figure 4 shows the simulation findings of
the new image cryptography method. Figures 4(a) and (b) show the original images and histogram of the
origininal images, respectively. Figure 4(c) and (d) show the cipher images and histogram of cipher images,
respectively. The decrypted images are shown in Figure 4(e). It can be noted that the cipher images are no
valid information and the decrypted images are similar to the plain images. Also, it can be seen from
Figures 4(b) and (d) that the histogram of the cipher images is unlike that of the plain images and the
histogram of the cipher images is close to the uniform distribution. The simulation results demonstrate that
the proposed technique is secure.

6.1. Key space

The size of the key space is determined by multiplying the key for each parameter and initial
condition. The security level of image encryption improves when the number of key spaces is increased.
Large key size is more robust against brute-force attacks that must exceed 2100 [14], [19], [22]. In this paper,
the following parameters and initial condition: 2D-CCM (Z (0), R (0), k, 𝛾 ), LM (𝑥0 , 𝜌, 𝑦0 , 𝜌), SM
(𝑥0 , 𝜌, 𝑦0 , 𝜌), EQM (𝑥0 , 𝑟, 𝑦0 , 𝑟), 2D-HCM (𝑥0 , 𝑦0 , 𝑎1 , 𝑎2 , 𝑎3 , 𝑏1 , 𝑏2 ), 2D-HSM (𝑥0 , 𝑦0 , 𝑎, 𝑏 ), and 5D-CM
(𝑍0 , 𝑅0 , 𝑆0 , 𝑊0 , 𝑈0 , 𝑎, 𝑏, 𝑐, 𝑑 ), and public key cryptography (s1, s2, s3). Therefore, there are 39 keys with a
precision 10-15 and the total key space is (1015)39≈21911. Thus, the suggested image cryptography approach is
robust against brute-force attacks.

Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)
1614  ISSN: 2302-9285

(a) (b) (c) (d) (e)

Figure 4. The results of the proposed cryptography technique based on simulation: (a) plain images,
(b) histogram of plain images, (c) cipher images, (d) histogram of cipher image, and (e) decrypted images

6.2. Correlation coefficients

The correlation coefficient is a statistical measurement of the relationship between adjacent pixels.
In the plain image, the value of the correlation coefficient should be high in the horizontal, vertical, and
diagonal directions. Conversely, the correlation coefficient value of the cipher image should be closer to zero.
The (14)-(17) of the correlation coefficient between adjacent pixels [27].
𝑐𝑜𝑣𝑎𝑟 (𝑥,𝑦)
𝜌𝑥𝑦 = (14)
√𝑣𝑎𝑟(𝑥)𝑣𝑎𝑟(𝑦)

1
𝐸(𝑥) = ∑𝑇𝑖=1 𝑥𝑖 (15)
𝑇

1
𝑣𝑎𝑟(𝑥) = ∑𝑇𝑖=1(𝑥𝑖 − 𝐸(𝑥))2 (16)
𝑇

1
𝑐𝑜𝑣𝑎𝑟(𝑥, 𝑦) = ∑𝑇𝑖=1(𝑥𝑖 − 𝐸(𝑥))(𝑦𝑖 − 𝐸(𝑦)) (17)
𝑇

where 𝜌𝑥𝑦 denotes the correlation coefficient between adjacent pixels (x, and y). T stand for the total number
of pixels selected from the image. 𝐸(𝑥), and 𝑣𝑎𝑟(𝑥) represent the mean and variance of x, respectively.
Table 1 shows the correlation coefficients of the proposed method using three colour images. It can
be seen that the correlation coefficients of cipher images are close to zero, which demonstrates that the
suggested image cryptography is an effective approach for reducing the interconnection between
neighbouring pixels. Table 2 shows a comparison of the correlation coefficient with other image encryption
techniques using Lena image. The results show that the proposed image encryption achieves almost the best
results compared with other recent works.

6.3. Entropy
Entropy is widely used to measure the uncertainty of a random variable. The entropy can be defined
as [1]:

𝐻(𝑥𝑖 ) = − ∑256
𝑖=1 𝑃(𝑥𝑖 ) log 2 (𝑃(𝑥𝑖 )) (18)

where 𝑃(𝑥𝑖 ) is the probability of 𝑥𝑖 .

For cipher image, the value of entropy should be ideally close to 8 [28]. Table 3 shows the
information entropy of the proposed algorithm using three cipher colour images. The outcomes demonstrate
that the values of entropy for three cipher image are close to 8, which indicate that cipher image pixels are

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1615

uniformly distributed [28]. Table 4 shows a comparison of the information entropy of the proposed algorithm
with other image encryption techniques using the Lena image. It can be seen that the values of entropy of the
proposed image encryption technique tend to 8, which proves that the suggested approach has strong
protection and that a third party can scarcely find useful information from a cipher image.

Table 1. Correlation coefficient of the proposed method using three colour images
Images Channel Horizontal Vertical Diagonal
Plain Cipher Plain Cipher Plain Cipher
Lena Red 0.9621 0.0782 0.9923 -0.0097 0.8981 0.0800
Green 0.9687 -0.0660 0.9884 0.0210 0.8913 -0.1045
Blue 0.9295 -0.0177 0.9647 -0.0349 0.7729 -0.0080
Peppers Red 0.8804 -0.0025 0.9831 -0.0951 0.9816 -0.0183
Green 0.8867 0.0991 0.9870 -0.0608 0.9784 -0.0843
Blue 0.8486 0.0609 0.9900 -0.1298 0.9065 -0.0673
Barbara Red 0.9342 -0.0410 0.8977 0.0126 0.9308 -0.0878
Green 0.8771 0.0592 0.9254 -0.0436 0.9206 0.0291
Blue 0.8818 0.0639 0.9557 -0.1453 0.9240 -0.0951

Table 2. Comparison of the correlation coefficient with other image encryption techniques using Lena image
Methods Channel Horizontal Vertical Diagonal
Plain Cipher Plain Cipher Plain Cipher
Ours Red 0.9621 0.0782 0.9923 -0.0097 0.8981 0.0800
Green 0.9687 -0.0660 0.9884 0.0210 0.8913 -0.1045
Blue 0.9295 -0.0177 0.9647 -0.0349 0.7729 -0.0080
[29] Red 0.9813 0.0092 0.9803 0.0203 0.9668 -0.0073
Green 0.9691 0.0002 0.9594 -0.0025 0.9433 -0.0131
Blue 0.9455 0.0076 0.9294 0.0006 0.9099 0.0111
[30] Red 0.9777 0.0090 0.9508 -0.0027 0.9259 -0.0013
Green 0.9670 -0.0013 0.9370 -0.0051 0.9111 -0.0155
Blue 0.9496 -0.0025 0.9171 -0.0103 0.8867 -0.0078
[31] Red 0.9775 -0.0021 0.9880 0.0027 0.9737 -0.00032
Green 0.9662 0.0017 0.9817 0.0023 0.9605 0.0010
Blue 0.9304 0.0012 0.9568 -0.0011 0.9219 0.00069

Table 3. The information theory of the proposed algorithm using three cipher colour images
Images Channels Proposed
Lena Red 7.9981
Greeen 7.9980
Blue 7.9979
Peppers Red 7.9971
Greeen 7.9979
Blue 7.9978
Barbara Red 7.9980
Greeen 7.9976
Blue 7.9979

Table 4. Comparability of data entropy of the proposed algorithm with other image encryption techniques
using Lena image
Methods Channels Proposed
Ours Red 7.9981
Greeen 7.9980
Blue 7.9979
[23] Red 7.9912
Greeen 7.9913
Blue 7.9914
[29] Red 7.9980
Greeen 7.9979
Blue 7.9978
[32] Red 7.9895
Greeen 7.9894
Blue 7.9894
[33] Red 7.9966
Greeen 7.9972
Blue 7.9967

Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)
1616  ISSN: 2302-9285

6.4. Peak signal to noise ratio

The (19) of peak signal to noise ratio (PSNR) can be represented as [1]:
2552
𝑃𝑆𝑁𝑅 = 10𝑙𝑜𝑔10 1 (19)
∑𝑀 ∑𝑁 [𝐼(𝑖,𝑗)−𝐶(𝑖,𝑗)]2
𝑀𝑁 𝑖=1 𝑗=1

where I is the original plain image and c is the encrypted or decrypted image. Table 5 shows the PSNR tests
for different images in case of encryption and decryption. In this table, it can be noticed that low power can
be obtained from the encrypted image compared with the decrypted image, proving the validity and
robustness of the proposed method.

Table 5. PSNR test for encrypted and decrypted images based on the proposed method
Image PSNR for encrypted image PSNR for decrypted image
Lenna 9.5857 52.81
Peppers 7.7337 19.67
Barbara 8.495 34.65

6.5. Differential attack

The NPCR and unified average-changing intensity (UACI) are two essential parameters to determine
the differential attack. The NPCR and UACI can be calculated according to the following equations [34]:
𝐷(𝑖,𝑗)
𝑁𝑃𝐶𝑅 = ∑𝑖,𝑗 × 100 (19)
𝑀×𝑁
1 |𝑐1 (𝑖,𝑗)−𝑐2 (𝑖,𝑗)|
𝑈𝐴𝐶𝐼 = [∑𝑖,𝑗 ] × 100 (20)
𝑀×𝑁 255

where 𝑐1 and 𝑐2 are two images whose size of 𝑀 × 𝑁. If 𝑐1 (𝑖, 𝑗) = 𝑐2 (𝑖, 𝑗), then 𝐷(𝑖, 𝑗) = 0.
Otherwise 𝐷(𝑖, 𝑗) = 1. The ideal value of NPCR is 0.996 and UACI is 0.334 [35]. Table 6 shows a
comparison of NPCR and UACI for the proposed algorithm using different colour images. It is clear that the
quantities of NPCR and UACI are near to the ideal quantities. Table 7 shows a comparison of NPCR and
UACI of the suggested algorithm with other image cryptography techniques using Lena's image. The results
demonstrate that the proposed image encryption technique is close to ideal values and that our proposed
image encryption is more robust to differential attacks.

Table 6. A comparison of NPCR and UACI of the proposed algorithm using different colour images
Images Channels NPCR (%) UACI (%)
Lena Red 99.6724 33.4985
Green 99.6992 33.4562
Blue 99.6682 33.5086
Peppers Red 99.7853 33.3946
Green 99.7838 33.5541
Blue 99.6877 33.3252
Barbara Red 99.6917 33.6271
Green 99.6544 33.4530
Blue 99.6785 33.5630

Table 7. A comparison of NPCR and UACI of the suggested algorithm with other image cryptography
techniques using Lena image
Methods Channels Channels UACI (%)
Ours Red 99.6724 33.4985
Green 99.6992 33.4562
Blue 99.6682 33.5086
[23] Red 99.6243 33.4686
Green 99.6433 33.5020
Blue 99.6029 33.4155
[36] Red 99.6479 33.4390
Green 99.6579 33.4799
Blue 99.6288 33.4833
[35] Red 99.6531 33.4572
Green 99.6522 33.4715
Blue 99.6518 33.4384
[32] Red 99.6052 33.4280
Green 99.6060 33.4966
Blue 99.6113 33.3779

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1617

6.6. Anti-shear attack

The analysis of anti-shear attack for the proposed method is shown in Figure 5. Figure 5(a)
illustrates the original encrypted image. To check the ability of the anti-shear attack for the proposed
algorithm, the encrypted image is cut-off by 40×40 in the middle and replacing the cut-off pixels by zero
values, as illustrated in Figure 5(b). Figure 5(c) illustrates the decrypted image of the original encrypted
image. The decrypted cut image is illustrated in Figure 5(d). In comparison, the results demonstrated that
only pixels cut from the encrypted image were affected and the reaming decrypted image has not changed by
the cutting.

(a) (b) (c) (d)

Figure 5. The analysis of the anti-shear attack (a) before cutting, (b) after cutting, (c) decryption before
cutting, and (d) decrypting after cutting

7. KEY SENSITIVITY ANALYSIS

A slight modification in the key will have a significant impact on the ciphertext; such a definition is
known as key sensitivity. In this context, this experiment used different images and which were given by the
first level of the initial key (K4(0)=[0.234545, 0.98791245]) , and sensitivity key (K4(1)=[0.234545000000001,
0.98791245]) for 2D-HCM. Table 8 shows key sensitivity of the proposed algorithm for different images. It
can be seen that the suggested algorithm has outstanding PSNR values besides the entropy measures,
confirming that the proposed algorithm has adequate key sensitivity.

Table 8. Key sensitivity analysis for different images

Image PSNR of decrypted PSNR of decrypted image with Entropy of decrypted Entropy of decrypted image
image key sensitivity image with key sensitivity
Lenna 52.81 9.5469 7.0994 7.9969
Barbara 34.00 8.5065 7.5505 7.997
Peppers 19.07 7.7245 7.2181 7.998

8. CONCLUSION
The transmission of images represents an important issue in network security. Attackers could
obtain transmitted images through the network. Therefore, it is important to employ a robust image
encryption technique to secure transmitted images. In this paper, we proposed an image encryption technique
based on hyperchaotic maps based on three scenarios. Firstly, the plain image is permuted using the 2D-
CCM. Then, the scrambled image is XORed with a pseudo-random bit generator. The pseudo-random bit
generator was designed using a hybrid of chaotic maps. Also, the public key Chebyshev polynomial was
applied as the final stage of encryption. Simulation findings demonstrated the validity of the suggested image
encryption technique in transforming the plain image into indistinguishable noise ones. The security results
show that the suggested method can combat different varieties of offensives. In future work, we will apply
DNA coding and anti-noise attack analysis with the proposed image encryption technology to further
enhance and investigate the performance of image encryption systems.

REFERENCES
[1] A. A. Karawia and Y. A. Elmasry, “New encryption algorithm using bit-level permutation and non-invertible chaotic map,” IEEE
Access, vol. 9, pp. 101357–101368, 2021, doi: 10.1109/ACCESS.2021.3096995.
[2] J. Zheng and T. Lv, “Image encryption algorithm based on cascaded chaotic map and improved Zigzag transform,” IET Image
Processing, vol. 16, no. 14, pp. 3863–3875, Dec. 2022, doi: 10.1049/ipr2.12600.
[3] M. A. Mokhtar, N. M. Sadek, and A. G. Mohamed, “Design of image encryption algorithm based on different chaotic mapping,”
in 2017 34th National Radio Science Conference (NRSC), Mar. 2017, pp. 197–204, doi: 10.1109/NRSC.2017.7893504.
[4] H. Xiang and L. Liu, “An improved digital logistic map and its application in image encryption,” Multimedia Tools and
Applications, vol. 79, no. 41, pp. 30329–30355, Nov. 2020, doi: 10.1007/s11042-020-09595-x.
Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)
1618  ISSN: 2302-9285

[5] S. K.U. and A. Mohamed, “A novel image encryption scheme using both pixel level and bit level permutation with chaotic map,”
Applied Soft Computing, vol. 90, pp. 1–17, May 2020, doi: 10.1016/j.asoc.2020.106162.
[6] M. S. Lee, “Improved cryptanalysis of a knapsack-based probabilistic encryption scheme,” Information Sciences, vol. 222, pp.
779–783, Feb. 2013, doi: 10.1016/j.ins.2012.07.063.
[7] H. Fujita, “Quantum McEliece public-key cryptosystem,” Quantum Information and Computation, vol. 12, no. 3–4, pp. 181–203,
Mar. 2012, doi: 10.26421/QIC12.3-4-1.
[8] V. Lyubashevsky, “Lattice signatures without trapdoors,” in Annual International Conference on the Theory and Applications of
Cryptographic Techniques, 2012, pp. 738–755, doi: 10.1007/978-3-642-29011-4_43.
[9] L. Kocarev, J. Makraduli, and P. Amato, “Public-key encryption based on Chebyshev polynomials,” Circuits, Systems and Signal
Processing, vol. 24, no. 5, pp. 497–517, Oct. 2005, doi: 10.1007/s00034-005-2403-x.
[10] Z. Li, C. Peng, W. Tan, and L. Li, “A novel chaos-based color image encryption scheme using bit-level permutation,” Symmetry,
vol. 12, no. 9, pp. 1–17, Sep. 2020, doi: 10.3390/sym12091497.
[11] X. Wang and N. Guan, “A novel chaotic image encryption algorithm based on extended Zigzag confusion and RNA operation,”
Optics & Laser Technology, vol. 131, pp. 1–17, Nov. 2020, doi: 10.1016/j.optlastec.2020.106366.
[12] J. Sun, “A 3D image encryption algorithm based on chaos and random cross diffusion,” Modern Physics Letters B, vol. 35, no. 30,
p. 2150465, Oct. 2021, doi: 10.1142/S0217984921504650.
[13] S. Sun and Y. Guo, “A new hyperchaotic image encryption algorithm based on stochastic signals,” IEEE Access, vol. 9, pp.
144035–144045, 2021, doi: 10.1109/ACCESS.2021.3121588.
[14] Z. A. Abduljabbar et al., “Provably secure and fast color image encryption algorithm based on s-boxes and hyperchaotic map,”
IEEE Access, vol. 10, pp. 26257–26270, 2022, doi: 10.1109/ACCESS.2022.3151174.
[15] D. Herbadji, A. Belmeguenai, N. Derouiche, and H. Liu, “Colour image encryption scheme based on enhanced quadratic chaotic
map,” IET Image Processing, vol. 14, no. 1, pp. 40–52, Jan. 2020, doi: 10.1049/iet-ipr.2019.0123.
[16] B. Li, X. Liao, and Y. Jiang, “A novel image encryption scheme based on logistic map and dynatomic modular curve,”
Multimedia Tools and Applications, vol. 77, no. 7, pp. 8911–8938, Apr. 2018, doi: 10.1007/s11042-017-4786-7.
[17] A. Mansouri and X. Wang, “A novel one-dimensional sine powered chaotic map and its application in a new image encryption
scheme,” Information Sciences, vol. 520, pp. 46–62, May 2020, doi: 10.1016/j.ins.2020.02.008.
[18] C. Li, T. Xie, Q. Liu, and G. Cheng, “Cryptanalyzing image encryption using chaotic logistic map,” Nonlinear Dynamics, vol. 78,
no. 2, pp. 1545–1551, Oct. 2014, doi: 10.1007/s11071-014-1533-8.
[19] S. Zhu and C. Zhu, “Plaintext-related image encryption algorithm based on block structure and five-dimensional chaotic map,”
IEEE Access, vol. 7, pp. 147106–147118, 2019, doi: 10.1109/ACCESS.2019.2946208.
[20] J. C. Sprott, Chaos and time-series analysis. Oxford: Oxford University Press, 2003.
[21] S. Wang, Q. Peng, and B. Du, “Chaotic color image encryption based on 4D chaotic maps and DNA sequence,” Optics & Laser
Technology, vol. 148, p. 107753, Apr. 2022, doi: 10.1016/j.optlastec.2021.107753.
[22] Z. Zhang, J. Tang, F. Zhang, H. Ni, J. Chen, and Z. Huang, “Color image encryption using 2D sine-cosine coupling map,” IEEE
Access, vol. 10, pp. 67669–67685, 2022, doi: 10.1109/ACCESS.2022.3185229.
[23] L. Teng, X. Wang, F. Yang, and Y. Xian, “Color image encryption based on cross 2D hyperchaotic map using combined cycle
shift scrambling and selecting diffusion,” Nonlinear Dynamics, vol. 105, no. 2, pp. 1859–1876, Jul. 2021, doi: 10.1007/s11071-
021-06663-1.
[24] T. Gopalakrishnan and S. Ramakrishnan, “Image encryption using hyperchaotic map for permutation and diffusion by multiple
hyper-chaotic maps,” Wireless Personal Communications, vol. 109, no. 1, pp. 437–454, Nov. 2019, doi: 10.1007/s11277-019-
06573-x.
[25] J. Wu, X. Liao, and B. Yang, “Image encryption using 2D Hénon-Sine map and DNA approach,” Signal Processing, vol. 153, pp.
11–23, Dec. 2018, doi: 10.1016/j.sigpro.2018.06.008.
[26] H. Ning, Y. Liu, and D. He, “Public key encryption algorithm based on Chebyshev polynomials over finite fields,” in 2006 8th
international Conference on Signal Processing, 2006, pp. 1–4, doi: 10.1109/ICOSP.2006.345958.
[27] H. M. Al-Mashhadi and I. Q. Abduljaleel, “Color image encryption using chaotic maps, triangular scrambling, with DNA
sequences,” in 2017 International Conference on Current Research in Computer Science and Information Technology (ICCIT),
Apr. 2017, pp. 93–98, doi: 10.1109/CRCSIT.2017.7965540.
[28] J. Hao, H. Li, H. Yan, and J. Mou, “A new fractional chaotic system and its application in image encryption with DNA mutation,”
IEEE Access, vol. 9, pp. 52364–52377, 2021, doi: 10.1109/ACCESS.2021.3069977.
[29] K. Xuejing and G. Zihui, “A new color image encryption scheme based on DNA encoding and spatiotemporal chaotic system,”
Signal Processing: Image Communication, vol. 80, pp. 1–11, Feb. 2020, doi: 10.1016/j.image.2019.115670.
[30] X. Wu, K. Wang, X. Wang, H. Kan, and J. Kurths, “Color image DNA encryption using NCA map-based CML and one-time
keys,” Signal Processing, vol. 148, pp. 272–287, Jul. 2018, doi: 10.1016/j.sigpro.2018.02.028.
[31] J. Chen, J. Tang, F. Zhang, H. Ni, and Y. Tang, “A novel digital color image encryption algorithm based on a new 4-D hyper-
chaotic system and an improved S-box,” International Journal of Innovative Computing, Information and Control, vol. 18, no. 1,
pp. 73–92, 2022, doi: 10.24507/ijicic.18.01.73.
[32] X. Wu, J. Kurths, and H. Kan, “A robust and lossless DNA encryption scheme for color images,” Multimedia Tools and
Applications, vol. 77, no. 10, pp. 12349–12376, May 2018, doi: 10.1007/s11042-017-4885-5.
[33] A. ur Rehman, X. Liao, R. Ashraf, S. Ullah, and H. Wang, “A color image encryption technique using exclusive-OR with DNA
complementary rules based on chaos theory and SHA-2,” Optik, vol. 159, pp. 348–367, Apr. 2018, doi:
10.1016/j.ijleo.2018.01.064.
[34] Y. Wu, J. P. Noonan, and S. Agaian, “NPCR and UACI randomness tests for image encryption,” Cyber journals:
multidisciplinary journals in science and technology, Journal of Selected Areas in Telecommunications (JSAT), vol. 1, no. 2, pp.
31–38, 2011.
[35] X. Wang, S. Gao, L. Yu, Y. Sun, and H. Sun, “Chaotic image encryption algorithm based on bit-combination scrambling in
decimal system and dynamic diffusion,” IEEE Access, vol. 7, pp. 103662–103677, 2019, doi: 10.1109/ACCESS.2019.2931052.
[36] Z. Hua, Z. Zhu, S. Yi, Z. Zhang, and H. Huang, “Cross-plane colour image encryption using a two-dimensional logistic tent
modular map,” Information Sciences, vol. 546, pp. 1063–1083, Feb. 2021, doi: 10.1016/j.ins.2020.09.032.

Bulletin of Electr Eng & Inf, Vol. 12, No. 3, June 2023: 1607-1619
Bulletin of Electr Eng & Inf ISSN: 2302-9285  1619

BIOGRAPHIES OF AUTHORS

Ahmed Kamil Hasan Al-Ali received the B.Sc. and M.Sc. degrees in
electrical engineering from Al-Mustansiriyah University, Baghdad, Iraq, in 2001 and 2005,
respectively, and the Ph.D. in Communication Engineering from Queensland University of
Technology, Australia in 2019. He is a lecturer at the Department of Electromechanical
Engineering, University of Technology. His research interests include digital signal
processing, speaker recognition systems, and digital communication systems. He can be
contacted at email: ahmed.k.alali@uotechnology.edu.iq.

Jafaar Mohammed Daif Alkhasraji received his B.Sc and M.Sc degrees
from Electromechanical Engineering and Laser and Optoelectronics Engineering
Departments, University of Technology, Baghdad, Iraq, in 2001, and 2008, respectively,
and PhD degree from Newcastle Newcastle Upon Tyne, United Kingdom, in 2019. Dr.
Jafaar has been a member of the syndicate of Iraqi engineers since 2002, and the IEEE
community since 2016. His research interests include optical communication, wireless
communication, DSP, MIMO, underwater optical wireless communication, coded and
precoded OFDM communications systems, and adaptive array signal processing and
estimation. He can be contacted at email: jaafar.m.dhaif@uotechnology.edu.iq.

Colour image encryption based on hybrid bit-level scrambling, ciphering … (Ahmed Kamil Hasan Al-Ali)