Image encryption based on three-dimensional bit matrix permutation
Introduction
The development of Internet technology and Social Networking Services (SNS) has led to an increase in online sharing of digital images and videos. Meanwhile, unauthorized access to or modification of private multimedia information has become a serious issue in the digital world. Although problems of this nature can be addressed by using information encryption, traditional encryption schemes, e.g., DES, and AES, are inappropriate for multimedia information encryption applications due to the intrinsic features of this kind of information, including a high correlation among the data in adjacent pixels, and the need for real time transformation [1]. Therefore, a new encryption mechanism capable of addressing the special demands of multimedia information is required. In this regard, recent years have seen significant developments in image encryption technology that use chaos theory together with permutation–diffusion architecture.
In his masterpiece, Shannon first introduced the idea of using permutation and diffusion to protect digital information [11]. In 1997, Fridrich proposed a similar architecture that uses two-dimensional (2D) dynamic systems to facilitate image encryption, thereby achieving a good balance between security and efficiency [12]. A number of image encryption algorithms have subsequently been proposed [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], and these algorithms can be classified into two categories according to the means by which pseudorandom numbers are generated, namely algorithms that are based either on chaos theory or on non-chaos theory. The first of these two categories, chaos-based image encryption, is based on the similarity between chaos dynamic systems and cryptosystems, including pseudorandom behavior, sensitivity to the initial condition, and ergodicity. In this category, one-dimensional (1D), high-dimensional, or hyper-chaotic systems are employed to generate pseudorandom number sequences to enable symmetric cryptosystems to be constructed. For example, in [6], one logistic map and one sine map are used to form a newly coupled 2D chaotic map in terms of which 2D-SLMM is employed to shuffle the plain-image, which can achieve good confusion and diffusion effects. In [14], a new 1D chaotic system is obtained by coupling the tent map and the sine map, which is then utilized to realize a pixel insert algorithm to permute the plain-image. A high-dimensional chaotic map has a more complex structure and pseudorandom properties, such as those used in Lorenz and Chen systems, and they are usually used to change the locations and values of the pixel set in image encryption algorithms. The Chen system, as applied in [10], is used to confuse and diffuse the three-color channel of the image, whereas in [9], the Chen system is employed as a pseudorandom number generator, and a temporary value feedback strategy, based on both confusion and diffusion, is used to encrypt an image with 256 Gy levels. The second of the two image encryption categories, non-chaos-based encryption, involves methods in which a non-chaotic dynamic system is used to generate a pseudorandom number sequence, which is then used to encrypt the image. For example, in [15], a new strategy based on balanced cellular automata is used to facilitate image encryption, whereas in [21], quantum Fourier transformation and a double random-phase encoding algorithm are used for color image encryption.
Bit level permutation (BLP) is a new approach that is employed at the confusion stage of image encryption and was first proposed in 2011 [1]. BLP considers an image to consist of a bit matrix to which all the encryption effects are applied; hence, the cipher-image should ultimately reflect the bit distribution of these encryption effects. In contrast, previous image encryption algorithms performed the encryption process at the pixel level, whereas BLP operates directly on each bit in the plain-image, rather than on a bit group at the pixel level. Recent times have seen the publication of a number of image encryption algorithms, most of which perform several rounds of 2D permutations among the different bit planes of the plain- image, following which the different permuted bit planes are combined to produce the confused (encrypted) image. For example, in [1], the four higher bit planes are permuted independently and the four lower bit planes are permuted together. In this particular case, the statistical information for each bit plane was not modified after the bit levels were confused. However, bit distribution plays a significant role in determining the extent to which an image encryption algorithm is able to secure an image. As illustrated in [2], there is a strong correlation between adjacent bit planes, and in this case, the BLP algorithm was used to restrict the destination location of one of the bits after confusion, which means that, one bit would only be able to move to another position within the same bit plane. The authors addressed this issue by employing a new “Expand and shrink” approach to partly eliminate this restriction, by allowing a bit that originally belonged to an even-numbered bit plane, to move to a destination that is located on any other even-numbered bit plane and this was also applied to bits located in odd-numbered planes.
However, the problem with this issue is that the bit matrix of an image is a natural 3D matrix, because an image with 256 Gy levels with a size of M×N can be considered a M×N×8 bit matrix, on the basis of which most existing algorithms try to use the combination of several 2D permutations to permute the 3D matrix. Furthermore, permutations using 2D chaotic maps or sorting algorithms experience repeated pattern problems, and the increasing number of elements in the sorting sequence results in a rapid increase in the execution time of the sorting algorithm.
In this paper, a new 3D permutation algorithm based on a coupled Chen system and a 3D chaotic map is proposed in an attempt to address the abovementioned problems. The new algorithm eliminates all the restrictions associated with bit level moving and satisfies the ideal situation of bit level permutation proposed in [2]. In the new permutation algorithm, not only the bit location but also the weight of each bit can be simultaneously modified. Furthermore, a new random visiting mechanism toward the 3D plain-image based on Chen system is employed in the first permutation stage, and after that 3D cat map govern the second the permutation stage. Simulations have been carried out and compared with three other image encryption algorithms, which indicating the security and efficiency of proposed scheme.
The remainder of this paper is organized as follows. Section 2 summarizes recently published permutation algorithms and the problems associated with the different approaches are highlighted and analyzed. In Section 3, the concept of a permutation over a 3D bit matrix is introduced and the superior permutation feature reported. Section 4 describes the diffusion phase. Section 5 summarizes the cryptosystem. Section 6 reports the simulation results of the proposed scheme and the comparable algorithm, following which a conclusion is drawn in the last section.
Section snippets
Problems in permutation algorithms
Permutation algorithms play a significant role in image encryption, which always forms the first step of a cryptosystem and aims to relocate all the pixels. Permutation algorithms can be classified into two categories or levels depending on whether pixels or bits are moved when the image is encrypted. Bit level permutation algorithms redefine the basic bit composition of each pixel and therefore lead to completely different statistical information on the pixel level. The idea of using BLP for
Hybrid permutation architecture based on 3D bit matrix
This section describes our proposed new bit level permutation architecture based on the fact that an image can naturally be considered a 3D bit matrix. The new permutation architecture utilizes a combination of sorting-based and Cat map-based permutation. The 3D matrix is illustrated in Fig. 2.
As mentioned above, an image with 256 Gy levels with a size of 512×512 can be considered a 3D matrix with the size 512×512×8, and in this case, r1, r2, and r3 in Fig. 2 would be equal to 512, 512, and 8,
Diffusion phase of the proposed scheme
Diffusion operations are performed at the pixel level. The first step involves the transformation of the confused bit matrix into the 2D pixel matrix, which is subsequently transformed into a 1D pixel array. Eq. (10) governs the diffusion phase.
In Eq. (10), pm[i] indicates the ith pixel in the permuted image, where , while cph[i] is the ith pixel in the cipher-image. The variables tmp1 and tmp2 are two temporary values used during
The proposed scheme
The proposed scheme is a typical Fridrich architecture cryptosystem, where one encryption round includes one confusion round and one diffusion round.
- (1)
The plain-image is first transformed from 2D pixel matrix to the 3D bit matrix;
- (2)
The 3D bit matrix is permuted based on the hybrid 3D permutation algorithm introduced in Section3, where Chen system and 3D cat map govern the process.
- a.
Three random sequences qxi, qyi, qzi are generated by Chen system.
- b.
The three sequences are sorted and compared with the
Simulation results
This section describes simulations of the proposed scheme and the three comparable image encryption algorithms. The simulation results and security analysis of the four image encryption algorithms are also provided. All the simulations were performed on a personal computer with the following hardware environment: 3.2 GHz CPU, 8 GHz memory, and a 240 GB hard drive. The software environment is Windows 7 Ultimate, and the compiler platform is Visual C++ 6.0.
The three comparable image encryption
Conclusions
This paper presented a brief summary of the permutation algorithms in cryptosystems that have been proposed in recent years, in addition to a detailed analysis of their merits and drawbacks. The drawbacks were addressed by proposing a new permutation mechanism based on a 3D bit matrix, in which the Chen system was combined with sorting-based algorithms to achieve a new random means of visiting the plain-image rather than using ordinary sequential visiting. Furthermore, with the help of a 3D Cat
Acknowledgements
The work described in this paper was supported by the National Natural Science Foundation of China (Grant no. 61402092), and the Fundamental Research Funds for the Central Universities (Grant no. N130417004).
References (24)
- et al.
A chaos-based symmetric image encryption scheme using a bit-level permutation
Inf. Sci.
(2011) - et al.
A symmetric color image encryption algorithm using the intrinsic features of bit distributions
Commun. Nonlinear Sci. Numer. Simul.
(2013) - et al.
A novel color image encryption algorithm based on chaos
Signal Process.
(2012) - et al.
Image encrypting using binary bitplane
Signal Process.
(2014) - et al.
A RGB image encryption algorithm based on total plain image characteristics and chaos
Signal Process.
(2015) - et al.
2D sine logistic modulation map for image encryption
Inf. Sci.
(2015) - et al.
A symmetric image encryption algorithm based on mixed linear-nonlinear coupled map lattice
Inf. Sci.
(2014) - et al.
An image encryption scheme using reverse 2-dimensional chaotic map and dependent diffusion reverse cat map
Commun. Nonlinear Sci. Numer. Simul.
(2013) - et al.
Color image encryption using spatial bit level permutation and high dimension chaotic system
Opt. Commun.
(2011) - et al.
A new optical image encryption method based on multi-beams interference and vector composition
Opt. Laser Technol.
(2015)
A new 1D chaotic system for image encryption
Signal Process.
Image encryption based on non-affine and balanced cellular automata
Signal Process.
Cited by (200)
A bit-plane encryption algorithm for RGB image based on modulo negabinary code and chaotic system
2023, Digital Signal Processing: A Review JournalVisually meaningful image encryption algorithm based on digital signature
2023, Digital Communications and NetworksCitation Excerpt :A two-dimensional modular chaotification system [36] was suggested to improve the chaotic complexity of chaotic system. Similarly, multi-dimensional chaotic system [30] was designed. Nevertheless, it was shown to be insecure against chosen plaintext attack [37].
A parallel image encryption algorithm using intra bitplane scrambling
2023, Mathematics and Computers in SimulationA chaotic image encryption scheme based on multi-directional confusion and diffusion operations
2022, Journal of Information Security and ApplicationsA novel exponential one-dimensional chaotic map enhancer and its application in an image encryption scheme using modified ZigZag transform
2022, Journal of Information Security and ApplicationsDesigning a double-way spread permutation framework utilizing chaos and S-box for symmetric image encryption
2022, Optics Communications