Journal of Information Security and Applications 0 0 0 (2019) 102361

Contents lists available at ScienceDirect

Journal of Information Security and Applications

journal homepage: www.elsevier.com/locate/jisa

When an attacker meets a cipher-image in 2018: A year in review

Chengqing Li a,∗, Yun Zhang a, Eric Yong Xie b
a
College of Computer Science and Electronic Engineering, Hunan University, Changsha 410082, China
b
College of Information Engineering, Xiangtan University, Xiangtan 411105, Hunan, China

a r t i c l e i n f o a b s t r a c t

Article history: This paper aims to review the encountered technical contradictions when an attacker meets the cipher-
Available online xxx images encrypted by the image encryption schemes (algorithms) proposed in 2018 from the viewpoint of
an image cryptanalyst. The most representative works among them are selected and classified according
Keywords:
Image encryption to their essential structures. Almost all image cryptanalysis works published in 2018 are surveyed due to
Cryptanalysis their small number. The challenging problems on design and analysis of image encryption schemes are
Plaintext attack summarized to receive the attentions of both designers and attackers (cryptanalysts) of image encryption
Image privacy schemes, which may promote solving scenario-oriented image security problems with new technologies.
Multimedia content protection
© 2019 Published by Elsevier Ltd.

1. Introduction  denotes an arithmetic operation, g is a fixed nonlinear func-

tion, and H = {h(i )} is a pseudo-random number sequence (PRNS).
Security and privacy protection of image data is an everlasting Only small number of chaotic image encryption schemes use S-box
challenge in the cyberspace. Designing image encryption schemes [6–8]. There are some papers focusing on designing S-box with
(algorithms, techniques, methods) and further processing in the chaotic maps [9–12]. In [6,13–20], the analyzed image encryption
domain of the cipher-images received intensive attention from re- schemes are fixed to withstand the proposed attacking methods.
searchers in the field of security, signal processing, nonlinear sci- Due to the seemingly similar properties of a chaotic system and
ence, and etc [1]. From Web of Science Core Collection, we found a secure pseudo-random number generator (PRNG), a large num-
about 10 0 0 publication records published in the year 2018 by in- ber of image encryption schemes use various chaotic maps to gen-
putting “image” and “cryptograph∗ or encrypt∗ ” in the field of erate PRNS, which is used to control combination of some basic
“Topic”. Some more were published in some top conferences on encryption operations, e.g. Logistic map [21,22], Tent map [12,23],
nonlinear sciences and signal processing, e.g. ISCAS and ICME. Cat map [24], PWLCM (piecewise linear chaotic map) [25,26], and
In cryptography, substitution-permutation network is the struc- coupled map lattice [27]. Rigorously, only the encryption schemes
ture of some modern block ciphers such as AES (Advanced En- using special chaos methodologies can be classified as chaos-based
cryption Standard), where an S-box (substitution-box) is used to encryption schemes, e.g. chaos synchronization [28,29]. As the op-
substitute the input of the S-box by another block of bits. Follow- posite of image cryptography, some impressive image cryptanaly-
ing the framework of Fridrich’s image encryption scheme crypt- sis works were developed in 2018 [30–35]. Especially, all image
analyzed in [2], a large number of image encryption schemes encryption schemes published in the journal Nonlinear Dynamics
adopting permutation-then-diffusion without substitution-box (PT- before 2018 are critically reviewed in [35].
DWOS) have been proposed since 1998 [3–5], where position per- This paper selected and classified the representative methods
mutation can be represented as of protecting and disclosing image data proposed in 2018. Some
challenges existing in the two sides are concluded from the per-
I∗ (w(i )) = I(i ), (1)
spective of a senior cryptanalyst. For example, only a few image
permutation vector W = {w(i )} is a bijective map on D, the support- encryption schemes focus on real specific applications, e.g. wire-
ing domain of the plain-image, i ∈ D; value diffusion is performed less communication [36], surveillance [37]. This paper aims to let
via substitution function designers and cryptanalysts of image encryption (and privacy pro-
tection) schemes glimpse the annual progress on confrontation be-
I (i ) = I∗ (i )  g(I (i − 1 ))  h(i ), (2) tween attackers and protectors of image data.
The remainder of the paper is organized as follows.
∗
Corresponding author. Section 2 reviews the design of image encryption schemes pro-
E-mail address: chengqing.li@my.cityu.edu.hk (C. Li). posed in 2018. Then, Section 3 presents a survey on cryptanalytic

https://doi.org/10.1016/j.jisa.2019.102361
2214-2126/© 2019 Published by Elsevier Ltd.
2 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361

works on image security given in the last year. Section 4 summa- system is constructed as PRNG to control the XOR operation and
rizes the challenges on the two sides on image security. The last position permutation on a medical image. As only one round of the
section concludes the paper. two basic operations are used, the encryption schemes designed in
[37,52] are both vulnerable to chosen-plaintext attack.
2. Survey on image cryptography in 2018 In [53], Ping et al. encrypt two adjacent pixels with position
permutation controlled by 2-D Hénon map and modulo addition.
2.1. Image encryption schemes based on chaos In [54], Ping et al. use 2-D Logistic-adjusted-Sine map to imple-
ment the permutation part and 2-D Cellular Automata (CA) as the
According to essential structure and used methodologies, the source of PRNS to control the part on modulo addition. In [55], a
encryption schemes using chaotic system are coarsely classified in PRNG based on a 3-D chaotic map is designed. The framework of
the following five sub-sections. the image encryption scheme is the same as that of [54]. In [56],
Ping et al. propose a method of digit-level permutation, which can
2.1.1. Randomness-oriented chaos enhancement change the histogram of the permuted image. In [22], 2-D Logistic
To counteract dynamics degradation of chaotic maps in digi- map is used to generate PRNG to control combination of position
tal computer and enhance randomness of PRNS generated by it- permutation and diffusion with XOR.
erating a chaotic system [38], various strategies are proposed, e.g. In [57], nine special pixels in the plain-image are selected to
cascading two existing chaotic maps [39]; iteratively expanding a generate initial conditions of Lorenz system with Secure Hash Al-
parametric 2-D Cat matrix to any higher dimension [24]; arbi- gorithm SHA-256. However, the sensitivity mechanism can be eas-
trarily combining six basic nonlinear operations [40]; anti-control ily canceled as only nine selected pixels have no influence on other
[41]; constructing hyper-chaotic system [42]. In [39], Hua et al. ones. In [25], a self-adapting colour image encryption scheme
design a sine-transform-based chaotic system of generating one- based on chaotic maps and the interactions among three channels
dimensional (1-D) chaotic maps by performing a sine transform to is proposed. Due to the commutative property of modulo addition,
the combination of the outputs of two maps. In [43], Hua et al. de- the sensitivity on the plain-image can be easily canceled as well.
sign a two-dimensional (2-D) Logistic-Sine-coupling map and use
it as a source of PRNS controlling basic encryption operations of an 2.1.3. Design of multiple round PTDWOS
image encryption scheme, whose structure follows PTDWOS. The Sheela et al.[58] propose an image encryption scheme based on
same strategy is used in [44], where a spatiotemporal chaotic sys- two rounds of position permutation and XOR operation, which are
tem is coupled with Tent-Sine system to generate PRNS. In [45], an controlled by PRNSs generated by 2-D modified Henon map and
encryption scheme of protecting medical images with two rounds Sine map, respectively. Rajagopalan et al. [59] adopt the same en-
of position permutation and diffusion, where Logistic-Sine system cryption structure as [58] with hardware implementation, where
is used as PRNG. Lorenz chaotic system, Lü’s chaotic system and CA are used as the
In [46], the periodicity of Logistic map and its variants over fi- source of PRNS instead. In [60], multiple rounds of PTDWOS are
nite field Z3e is analyzed. Based on comprehensive analysis of ran- implemented in the level of bit plane, where Skew Tent map is
domness of the PRNS generated by iterating a map, one variant of used as source of PRNS. In [4], 2-D Sine-Logistic modulation map
Logistic map is selected in [46] as a PRNG. In [47], a 6-D discrete (2D-SLMM) is used to control the two basic part of a scheme fol-
chaotic system is constructed with some sine and cosine functions, lowing PTDWOS, where modular addition with a fixed matrix is
which is then used as a source of PRNS used in a stream cipher. As adopted to enhance the diffusion effect. In [61], a bit-level image
analyzed in [21], the coupled logistic map can demonstrate com- encryption algorithm is designed by operating multiple rounds of
plex bifurcation dynamics in a continuous domain. Actually, the bitwise shift operation and XOR in two directions, where a hyper-
real structure of the final chaotic systems obtained with any above chaotic map is constructed as the source of PRNG by cascading sine
way in a digital computer should be analyzed as [48] with the map and Logistic map.
methodology of state-mapping network. Following the basic structure of AES, a robust image symmet-
ric cryptosystem is proposed in [62], where Arnold’s cat map
2.1.2. Design of single round PTDWOS is used to generate a permutation matrix permuting the entire
Among the encryption schemes composing of only one round plain-image. In [63], an image encryption algorithm based on 1-
of basic operations, only [12] adopt S-box, where tent map is used D PWLCM and least squares approximation is proposed, where
as PRNG to control some basic encryption operations on image: PWLCM is used to determine the permutation relationship among
modular addition, S-box and binary shift operation. the rows and columns of the plain-image and the intermediate
In [49], 1-round PTDWOS is proposed, where the algorithm on matrixes. The method based on least squares approximation is
traversing a graph with the strategy of breadth-first search is used adopted to generate PRNS to mask the permuted intermediate ma-
to implement position permutation, and a hyper-chaotic system trixes. In [26], PWLCM is used as PRNG to control the combination
is used as PRNG to realize the second permutation and diffusion of two rounds of position permutation and modular addition. In
with XOR and modular addition. In[50], an encryption function on [64], a 2-D partitioned CA is designed to generate PRNS satisfying
image blocks with variable sizes is proposed, which is composed the global Strict Avalanche Criterion, which is used to determine
of position permutation and modulo addition. The two basic op- the combination of some basic operations including substitution
erations are controlled by Arnold map and integer Logistic map, with an S-box.
respectively. In [51], the proposed image encryption scheme per-
forms only one round of position permutation and bitwise exclu- 2.1.4. Encrypting multiple plain-images simultaneously
sive OR (XOR) operation, which are both controlled by PRNS gen- In [3], a double colour image encryptionDCIE) scheme using 3-
erated by 2-D Baker’s map. The critical frame of the surveillance D Brownian motion is presented, where two colour images are en-
video for IoT systems is encrypted the two steps in [37], where crypted simultaneously. DCIE follows the structure of PTDWOS and
2-D logistic-adjusted-sine map (LASM) is used as the source of uses 3-D Brownian motion, a 3-D autonomous chaotic system and
PRNS. The dependence of PRNS on the plain-image can be canceled 2D-LASM as PRNG to control its basic components. In [65], another
with a non-negligible probability due to the commutative princi- DCIE using Gyrator transform and Hénon map is proposed. In [66],
ple of summation and quantization error of calculating function in the four least significant bit-planes of two images are encrypted
a finite-precision computer. In [52], a simple fourth-order chaotic by position permutation and value confusion, which are controlled
 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361 3

by Logistic map and CA, respectively. In [67], four plain-images are In [81], a scheme encrypting optical image is designed with
encrypted simultaneously using 2D-LASM. In [68], a sequence of the same strategy as that in [78,80]. In [82], an image encryption
images are simultaneously encrypted and compressed using Tensor scheme using pixel-level scrambling, bit-level scrambling, and DNA
Compressive Sensing, which is controlled by PRNS generated by a encoding is proposed, where a 5-D hyperchaotic system is used as
3-D discrete Lorenz system. In [69], a multiple optical colour image the source of PRNS. In [83], 2-D Hénon-Sine map is constructed
encryption scheme based on phase retrieval in quaternion gyrator as a PRNG to control the involved DNA rules, DNA operation and
domain is proposed, nonlinear quaternion correlation is developed the position permutation. In [84], the saliency detection of a plain-
to perform authentication. image is detected with a given model and encrypted with a chaos-
In [70], an image encryption scheme using beta chaotic map, based encryption function. Then, the encrypted result is embedded
NSCT (Nonsubsampled Contourlet Transform), and GA (genetic al- into DCT domain of another meaningful natural image as an invis-
gorithm) is proposed. The plain-image is decomposed into three ible watermark.
subbands. Then, GA is used to select the optimized initial parame-
ters of the beta chaotic map for a multi-objective fitness function. 2.3. Encrypting image in transform domain
Finally, the inverse of NSCT is applied on the encrypted subbands
with modulo addition to produce the cipher-image. In [71], teach- A series of encryption schemes protecting optical images is pro-
ing learning based optimization method and gravitational search posed in 2018. But, most of them are simulated with digital images
algorithm are used to optimize selection of the parameters. How- due to the limit of optical devices. In [85], a monospectral syn-
ever, such optimization algorithms cost additional unknown com- thetic aperture integral imaging system is designed to capture 2-
putational load. In [72], GA is used to perform parallel permutation D optical elemental images, which are then XORed with a PRNS
and substitution on multiple bitplanes of the plain-image. generated by CA and digitally encoded by Fresnel transform. In
[86], the original image is first scrambled, and then multiplied
2.1.5. Encrypting images with compressing techniques by the random-phase mask function. A further phase-truncated
In [73], an image encryption scheme based on compressive discrete multiple-parameter FrFT is implemented to realize asym-
sensing and chaotic map is proposed. The 2D-SLMM is used in the metric encryption. In [87], a holographic encryption scheme based
three basic parts of the scheme: constructing measurement ma- on interleaving operation of computer-generated holograms is pro-
trix in compression operation; generating PRNG for diffusion oper- posed. Using the quaternion algebra, Chen et al. define a multiple-
ations; producing permutation matrix. In [74], a plain-image is first parameter fractional quaternion Fourier transform as a general
decomposed by wavelet packet transform, whose coefficients are version of the conventional multiple-parameter fractional Fourier
classified in terms of the average value and Shannon entropy. The transform [88]. Detailed experimental results demonstrate that
coefficients containing principal energy are encrypted by position MPFrQFT is superior to other eight existing algorithms on multiple
permutation, modular subtraction and XOR, which are controlled metrics: key space, key sensitivity, statistical analysis and robust-
by Logistic map, Chen system and Arnold map, respectively. Other ness.
nonzero coefficients are compressed by compressive sensing (CS) In [89], an optical hyperspectral image cryptosystem is pro-
with a secret measurement matrix constructed from a Hadamard posed using improved Chirikov mapping in the gyrator transform
matrix controlled by Cat map. The encryption scheme is based domain, the original hyperspectral image is converted into binary
on 2-D CS and fractional Fourier transform (FrFT). In [75], Liao format and then extended into a 1-D array, then the scrambled
et al. employ Kronecker product to generate higher dimensional and exchanged image is transformed using the gyrator transform.
measurement matrices. In [76], compressive sensing technique is In [90], quick response (QR) code-based non-linear technique for
combined with a stream cipher to simultaneously compress and image encryption using shearlet transform and spiral phase trans-
encrypt image and video files, where the Measurement Matrix is form is proposed. The input image is first converted into a QR code
generated using a stream cipher. Only the four most significant bits and then scrambled via the Arnold transform.
of compressed samples are encrypted, which may cause some and In [91], a DCT-domain image encryption framework is proposed
even all visual information of the encrypted object revealed. with block-wise permutation and XOR operation, which is imple-
mented with a stream cipher. In [92], a bitstream-based JPEG im-
2.2. Designing image encryption schemes based on DNA encoding age encryption scheme with negligible size expansion is proposed,
which cascades four operations: permuting the groups of succes-
With the development of DNA computing, designing image en- sive DC codes encoding the differences of quantized DC coefficient
cryption schemes based on DNA encoding received intensive atten- with the same sign; swapping the two half parts of a group of con-
tions in the past decade [77]. In 2018, about twenty technical pa- secutive DC codes; scrambling the AC codes falling the same cate-
pers on the topic were published. Here, we review representative gory; randomly shuffling all minimum coded units.
works among them.
In [78], a new heterogeneous chaotic neural network generator 2.4. Signal processing in the encrypted domain
is proposed as a source of PRNS to control the three basic encryp-
tion operations on image data: pixel position permutation; DNA- In [93], the fast methods of implementing real and complex
based bit substitution; DNA-based bit permutation. In [79], a ro- Walsh-Hadamard Transform in the cipher-images are proposed. To
bust color image encryption system using Lorenz-Rossler chaotic further reduce the computational complexity of the homomorphic
map and DNA encoding is proposed, where Lorenz-Rossler chaotic encryption operations, two parallelization strategies are given to
map is used as the source for producing PRNS. The three chan- accelerate the transform. Using strong correlation among neigh-
nels of color plain-image and PRNS are transformed into the form bouring pixels in a natural image, an effective high-capacity re-
of DNA strands. After performing subtraction and addition opera- versible data hiding scheme for encrypted images based on MSB
tions, the results are transformed back to the binary form as the (most significant bit) prediction is proposed in [94]. Interestingly,
cipher-image. In [80], an image encryption scheme based on DNA [94] adopts PWLCM as the source producing pseudo-random num-
encoding is designed, using a 2D-LASM and Single Neuron Model ber sequence, whose randomness is seriously comprised in a digi-
as sources of PRNS. First, the plain-image is encrypted by position tal computer [48].
permutation and XOR. Then, the intermediate image is further en- Reversible data hiding in encrypted image (RDH-EI) embeds
crypted in the DNA domain with XOR operation. data into a cipher-image in cloud meanwhile the corresponding
4 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361

plain-image can be perfectly reconstructed by the authorized re- a linear programming problem without sensitive information of
ceiver. In [95], two RDH-EI schemes are proposed with private- the searched image.
key homomorphism and public-key homomorphism, respectively. • Quantum image encryption
In [96], an enhanced RDH-EI embedding two bits in each cipher- The development of quantum cryptography and quantum chaos
pixel with minimum distortion of stego-pixel is achieved through attracts image presentation in quantum world and study on
homomorphic encryption. In contrast to many data hiding schemes how to utilize quantum properties for protecting image data
focusing on LSB, an efficient MSB prediction-based method for [109,110]. In [111], a quantum representation model is pro-
RDH-EI is proposed in [97], where vacating embedding room af- posed to let quantum hardware encrypt any number of plain-
ter encryption and reserving embedding room before encryption images simultaneously. Then, the quantum circuit of quan-
are discussed, respectively. To cope with the compressed images, tum 3D Arnold transform is defined. Numerical simulations are
[98] proposes a separable RDH for encrypted JPEG bitstreams, given to show performance of a multi-image encryption scheme
where the original bitstream is losslessly reconstructed using an combining quantum 3D Arnold transform, quantum XOR oper-
iterative recovery algorithm. In [99], a RDH-EI is designed using ations and scaled Zhongtang chaotic system.
homomorphic multiplication and probabilistic properties of Paillier • Permutation against attacks based on “jigsaw puzzle solver”
cryptosystem, where the embedding room is reserved before en- Due to simplicity and obvious security enhancing effect of posi-
cryption. tion permutation, it is widely used in image encryption scheme
In [100], Wang et al. formulate lossy compression on low- [92,112]. The ciphertext-only attack on any block-wise permu-
frequency wavelet coefficients as a problem of weighted rate- tation scheme can be attributed to solving a jigsaw puzzle.
distortion optimization and solve it by incorporating empirical Recently, Kiya et al. published a series of works on encrypt-
characteristic of their distribution. In [101], Wang et al. propose an ing each image block with marginal influence on statistics and
iterative reconstruction scheme for compression file of encrypted permuting the encrypted blocks among the whole image. Such
binary images using the Markov Random Field to characterize the encryption-then-compression framework does not seriously in-
corresponding plain-images in the spatial domain. The associate fluence the further compression and can withstand the attacks
Markov Random Field, decryption and decompression based on based on conventional jigsaw puzzle solvers [113]. But, the cor-
LDPC are all represented with the methodology of factor graph, relation among blocks in cipher-images may still facilitate op-
a bipartite graph representing the factorization of a function into timizing the attacking on the block-wise permutation schemes
some subfunctions. analyzed in [114,115] with the advanced jigsaw puzzle solvers.

3. Survey on image cryptanalysis in 2018

2.5. Generating cipher-images in other application scenarios
In the year 2018, about 30 papers on image cryptanalysis were
• Visual secret sharing published. Among them, the most striking work is depreciating
The (k, n)-visual secret sharing (VSS) scheme encode a secret motivation of most chaos-based encryption schemes in [116], refut-
image into n shares, such that any k shares can be superim- ing any superiority of chaos-based encryption schemes than sev-
posed to reveal the secret image to human’s eyes but any k − 1 eral variants of AES in terms of computational effort and security
or less ones can leak nothing. In [102], three efficient and flex- performance, which may deserve argument with rigorous theoret-
ible XOR-based (k, n)-VSS schemes is constructed, such that ical analyse or convincing engineering application. In [116], four
more than k shares can also reveal the secret image. As for obviously insecure encryption schemes are subtly constructed to
progressive (k, n) secret image sharing (PSIS) scheme, it is re- question credibility of the security assessment metrics used in the
quired that k to n shares can reconstruct the secret image pro- field of chaotic cryptography widely, e.g. correlation, entropy, his-
gressively. In [103], three PSIS schemes are constructed using togram variance, Number of Pixel Change Rate (NPCR), Unified Av-
XOR operation, a variant of Hamming code and modified ver- erage Changing Intensity (UACI), key sensitivity, robustness against
sion of shorten Hamming code, respectively. In [104], an image differential attack, sequence test, and NIST randomness test. Con-
secret sharing with three decoding options, lossless recovery, sidering the quantitative test results of the four special schemes on
grayscale stacking recovery and visual previewing, is designed the metrics, it is concluded that the metrics are far ineffective for
based on random grid-based VSS and Chinese remainder theo- security analysis, which are further qualitatively analyzed the inef-
rem. ficiency of the metrics in [1]. Other image cryptanalysis works are
• Privacy-preserving application of image data separately reviewed in the following three subsections.
In [105], Yuan et al. use similarity degree among encrypted
higher-dimensional profile vectors of users images as a metric 3.1. Cryptanalysis of single round PTDWOS
to search and recommend friends in social media. In [106], a
privacy-preserving biometrics-based authentication solution is In [117], security of a block-wise image encryption scheme us-
proposed to authenticate the users to remote service providers ing an ECG signal and the plain-image to determine the used PRNG
via zero-knowledge proof of knowledge on two secrets: an is analyzed comprehensively. As the sensitivity mechanism on the
identity token encoding the biometric identifier of the user’s plain-image is built via a modulo addition, it can be canceled with
image and a secret owned by the user. In [107], a searchable a non-negligible probability of 1/256. Meanwhile, no permutation
symmetric encryption over encrypted multimedia data is pro- operation is adopted in the scheme, causing an equivalent secret
posed by considering the search criteria as a high-dimensional key can be derived from one pair of a known plain-image and its
feature vector, which is further mapped by locality sensitive corresponding cipher-image. As the insecurity of sole permutation
hashing. The inverted file identifier vectors are encrypted by analyzed in [118], position permutation is still a simple but effi-
an additive homomorphic encryption or pseudo-random posi- cient measure to improve the security level of the whole encryp-
tion permutations. In [108], a privacy-preserving content-based tion scheme.
image retrieval (CBIR) scheme is proposed to allow the data Four research teams independently studied security analysis of
owner outsource the image database and CBIR operation to the an image encryption scheme using the combination of some 1-D
cloud server without disclosing its actual content. The proposed chaotic maps proposed by Pak et al. in 2017, whose structure is
scheme let the cloud server can solve earth movers distance as one-round PTDWOS with an additional shift permutation. Ignoring
 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361 5

the fact that permutation and modulo addition are not commuta- how the changes of the cipher-pixels influence that of the plain-
tive, Wang et al. simplifies Pak’s scheme by cascading two non- pixels in the scenario of chosen-ciphertext attack [126]. In [6], two
neighbouring permutations together and break it with a chosen- chosen plain-images are constructed to recover an equivalent ver-
plaintext attack in [16]. The enhanced version of Pak’s scheme sion of S-box and PRNS of a chaos-based image encryption scheme
suggested in [16] is found still vulnerable against chosen-plaintext using S-box.
attack [17]. Assuming a parameter seed is known, the parameter
determining the shift permutation of Pak’s scheme is disclosed in 3.3. Other image cryptanalysis
[18] by its internal properties. Then, Pak’s scheme is degenerated
into the version analyzed in [16]. Based on the preliminary crypt- According to the scenario or theory used by an encryption
analysis works given in [16] and [18], both the equivalent secret- scheme, a special cryptanalysis method may work very well.
key of Pak’s scheme and all its unknown parameters are recovered In [127], a ciphertext-only attack on a double-image optical
by a little more pairs of chosen plain-image and the correspond- encryption technique based on an asymmetric (public-key en-
ing cipher-image in [19]. In [119], Zhu and Sun propose a chosen- cryption) algorithm is proposed using phase retrieval algorithm
plaintext attack on another one-round PTDWOS, where diffusion is with median filtering and normalization operation. As for an opti-
implemented by modulo addition and XOR. The equivalent secret cal cryptosystem based on phase-truncated Fourier transform and
key of such diffusion mechanism can be easily recovered [120]. nonlinear operations, the same research group presented a specific
In [15], Li et al. propose a chosen-plaintext attack on an im- attack based on phase retrieval algorithm with normalization and
age encryption scheme based on Ikeda map, where the Ikeda map a bilateral filter in [128]. In [129], a joint compression and encryp-
is used to generate PRNS to control the mask operation XOR. In tion scheme based on elliptic curve cryptography (ECC) is crypt-
[121], Li et al. further present two chosen-plaintext attack meth- analyzed. Due to the used parameter of ECC is too small, three
ods on a colour image encryption using chaotic APFM (Amplitude known attacking methods on solving elliptic curve discrete loga-
Phase Frequency Model) nonlinear adaptive filter, which is com- rithmic problem are used to recover the private key from the pub-
posed of three parts: position permutation; nonlinear substitution; lic key.
XOR operation. In [122], Li et al. recover the permutation rela- In [130], a large number of plaintexts and the corresponding
tion of image encryption scheme using PTDWOS with some cho- ciphertexts encrypted by any optical encryption technique based
sen plain-images. In [123], Li et al. recover the equivalent secret on random phase encoding are used as training samples for opti-
key of an encryption algorithm using one-round PTDWOS with 512 mizing the parameters of a deep neural network, which can re-
pairs of chosen plain-images and the corresponding cipher-images, construct any plain-image from its corresponding cipher-image.
where hybrid hyper-chaos and CA are used as the source of PRNS Actually, deep learning can also be used to test security of an en-
to control the permutation and mask operation XOR. In [31], an- cryption scheme via adversarial training.
other encryption scheme using PTDWOS is cryptanalyzed. Similar In [33], Luo et al. evaluates the security of chaotic encryption
to the strategy given in [117], the sensitivity mechanism of PRNG schemes by using side-channel attack. Some intermediate values
on the plain-image can be easily canceled by some special plain- relating to the plaintext and the round keys can be revealed by
images. observing the power consumption. In [131], influences of differ-
In [32], security defects and chosen-plaintext attack on an im- ent operations on resistance of chaotic image encryption schemes
age encryption schemes for body area network system are an- combining position permutation and a mask operation against
alyzed. Furthermore, some additional operations are adopted to chosen-plaintext attacks are discussed. Among some reversible and
withstand the proposed attack: using SHA-512 to build linkage be- irreversible gates, including XOR, Toffoli and Fredkin gates, only
tween the plain-image and PRNS; updating initial conditions and Fredkin gate can help the supported encryption scheme to with-
control parameters of the chaotic system; adding more complex stand chosen-ciphertext attack.
encryption operations. However, the incurred computational load About eighty publications on “reversible data hiding in the
is not considered for the required security level. cipher-image (encrypted domain)” were published in the past
decade. Many of them use simple stream cipher XOR to generate
3.2. Cryptanalysis of some image encryption schemes with complex cipher-images. Using the strong correlation among neighboring bits
structure in every bit-plane in a natural image, a cipher-text only attack is
proposed in [34] to recover the approximate version of the original
Once the basic encryption operations are repeated some rounds plain-image with good visual quality.
or S-box is used, the structure of the whole encryption scheme be- In [132], a chosen-ciphertext attack on a chaotic stream ci-
come much more complex, which dramatically increases the hard- pher using self-synchronization and closed-loop feedback is pro-
ship of the corresponding cryptanalysis. posed. Some parameters of the used 3-D discrete time chaotic
In [13], two chosen plain-images of fixed value are constructed system can be estimated. In [133], Lin et al. further derive coeffi-
to recover the equivalent secret key of an “efficient image cryp- cients of the nominal matrix of a stream cipher based on 8-D self-
tographic algorithm” reusing the permutation matrix generated by synchronous chaotic system in the scenarios of known-plaintext
Cat map or Baker map dynamically. But, how to perform the attack attack and chosen-ciphertext attack, respectively.
on the PTDWOS with multiple rounds are not theoretically proved
4. The challenges on design and analysis of image encryption
in [13].
schemes
In [124], some special plain-images are selected to narrow the
scope of some subkeys of a PTDWOS with less than or equal
Based on the above review on two sides of image cryptology,
to 3 rounds. In [125], the usability of a 4-round diffusion-then-
we summarize some critical challenges on design and security
permutation without substitution-box is questioned, which can be
analysis of image encryption schemes as follows.
considered as a special form of 5-round PTDWOS. Furthermore,
the linear relationship between plain-image and the corresponding • No specific application scenario is targeted in the design pro-
cipher-image is derived as the adopted permutation is a static rota- cess of most image encryption schemes. Under such precon-
tion of 90 degrees, which is used to support the proposed differen- dition, the balancing point among computation load, the real
tial attack. Similar to the strategy introduced in [2], the permuta- required security level and economic cost for a potential at-
tion relation of a multiple-round PTDWOS is recovered by checking tack cannot be considered at all. Only a few image encryption
6 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361

schemes are designed for solving security challenges in specific nior image cryptanalyst. The representative works on the two sides
application scenarios: surveillance [37], IIoT (industrial Internet were classified and some creative ideas and methods were high-
of Things) [29], social media [91]. lighted. More importantly, the challenges on design and analysis
• Special formats of image data are not considered. A large num- of image encryption schemes were briefly summarized to promote
ber of image encryption schemes just treat plain-image as an the progress of protection level of image data in cyberspace.
ordinary textual data [5]. Due to the large size of conventional
image files, compression and encryption should be combined Declaration of interests
well [92,134]. As [135], only ROI (region of interest) of DICOM
(Digital Imaging and Communications in Medicine) medical im- The authors declare that they have no known competing finan-
ages are encrypted with one round of permutation and substi- cial interests or personal relationships that could have appeared to
tution to satisfy real-time processing requirement. influence the work reported in this paper.
• Importance of S-box is seriously neglected. As a lookup table, S-
box is an efficient way to obscure the relationship between the Acknowledgements
secret key and the cipher-image. Strangely enough, it is seldom
adopted in image encryption schemes. This research was supported by the Natural Science Founda-
• The underlying weak randomness of chaos-based PRNG is tion of China (No. 61772447, 61532020), the Joint Funds of the
widely omitted. The periods of PRNS obtained by iterating Lo- National Natural Science Foundation of China, China General Tech-
gistic map and skew tent map in digital computer are rigor- nology Research Institute (No. U1736113) and the Fundamental Re-
ously analyzed in [48]. Much more theoretical results on dy- search Funds for the Central Universities (No. 531118010348).
namical properties of chaotic maps, e.g. 2D-LASM, Hénon map,
Lorenz system in limited-precision arithmetic domain should be Supplementary material
further investigated by extending the analysis in the field Z2 to
Z2e using advanced theoretical analysis tools, e.g. spectral test Supplementary material associated with this article can be
[136], Hensel’s lifting lemma, where e > 1. found, in the online version, at doi:10.1016/j.jisa.2019.102361.
• The security test metrics are widely misused. As shown in
[116,117], the security assessment metrics used in most chaos- References
based image encryption schemes are unconvincing. So the con-
vincing methodologies for assessing security of image encryp- [1] Li C, Lin D, Feng B, Lü J, Hao F. Cryptanalysis of a chaotic image encryp-
tion schemes need to be established urgently. tion algorithm based on information entropy. IEEE Access 2018;6:75834–42.
doi:10.1109/ACCESS.2018.2883690.
• The previous lessons drawn by tough cryptanlaysis are ignored [2] Xie EY, Li C, Yu S, Lü J. On the cryptanalysis of Fridrich’s chaotic image en-
by the designers of new image encryption schemes. For ex- cryption scheme. Signal Process 2017;132:150–4. doi:10.1016/j.sigpro.2016.10.
ample, the sensitivity mechanisms of PRNS on the plain-image 002.
[3] Gan Z, Chai X, Zhang M, Lu Y. A double color image encryption
built in [27,80,92,119] can be easily canceled by constructing scheme based on three-dimensional Brownian motion. Multimed Tools Appl
some special plain-images, which has been reported in many 2018;77(21):27919–53. doi:10.1007/s11042- 018- 5974- 9.
cryptanalysis papers such as [1,117]. [4] Ye G, Pan C, Huang X, Mei Q. An efficient pixel-level chaotic im-
age encryption algorithm. Nonlinear Dyn 2018;94(1):745–56. doi:10.1007/
• Prejudice on the value of cryptanalysis obstructs the progress of s11071- 018- 4391- y.
image security. As emphasized in the abstract of the special re- [5] dai Liu D, Zhang W, Yu H, liang Zhu Z. An image encryption scheme using
port on PRNG test suite of NIST, “statistical testing cannot serve self-adaptive selective permutation and inter-intra-block feedback diffusion.
Signal Process 2018;151:130–43. doi:10.1016/j.sigpro.2018.05.008.
as a substitute for cryptanalysis”. It is unreasonable to require
[6] Zhu C, Wang G, Sun K. Cryptanalysis and improvement on an image en-
a cryptanalyst to propose modifications to withstand the pro- cryption algorithm design using a novel chaos based S-box. Symmetry-Basel
posed attacks. The work of cryptanalysis should focus on the 2018;10(9). art. no. 399. doi: 10.3390/sym10090399.
[7] Liu H, Kadir A, Sun X, Li Y. Chaos based adaptive double-image en-
conditions of required plain-image; computational cost of the
cryption scheme using hash function and S-boxes. Multimed Tools Appl
attack; space complexity. In 2018, improvement (enhancement, 2018;77(1):1391–407. doi:10.1007/s11042- 016- 4288- z.
remedy) methods were proposed to fix the reported security [8] Ullah A, Jamal SS, Shah T. A novel scheme for image encryption using sub-
defects in a number of papers on image cryptanalysis as ref- stitution box and chaotic system. Nonlinear Dyn 2018;91(1):359–70. doi:10.
1007/s11071- 017- 3874- 6.
erences [6,13–20,30,32,119,131]. In fact, the proposed improve- [9] Aboytes-Gonzalez JA, Murguia JS, Mejia-Carlos M, Gonzalez-Aguilar H,
ment methods fail to obtain the expected object due to the fol- Ramirez-Torres MT. Design of a strong S-box based on a matrix approach.
lowing subjective and objective reasons: Nonlinear Dyn 2018;94(3):2003–12. doi:10.1007/s11071- 018- 4471- z.
[10] Khan M, Asghar Z. A novel construction of substitution box for image en-
– The authors proposing a successful attack, e.g. chosen- cryption applications with Gingerbreadman chaotic map and s8 permutation.
plaintext attack in [30], on a given image encryption scheme Neural Comput. Appl. 2018;29(4):993–9. doi:10.10 07/s0 0521- 016- 2511- 5.
may not own comprehensive knowledge on designing se- [11] Alzaidi AA, Ahmad M, Doja MN, Solami EA, Beg MMS. A new 1d chaotic
map and beta-hill climbing for generating substitution-boxes. IEEE Access
cure image encryption schemes. 2018;6:55405–18. doi:10.1109/ACCESS.2018.2871557.
– Many image encryption schemes are not designed and pre- [12] Palacios-Luengas L, Delgado-Gutierrez G, Diaz-Mendez JA, Vazquez-Medina R.
sented following the basic rules and general guidelines Symmetric cryptosystem based on skew tent map. Multimed Tools Appl
2018;77(2):2739–70. doi:10.1007/s11042- 017- 4375- 9.
given in [35,137], which makes essential improvement on
[13] Diab H, El-semary AM. Cryptanalysis and improvement of the im-
the security level of the analyzed schemes very hard. age cryptosystem reusing permutation matrix dynamically. Signal Process
• To cryptanalyze the seemingly complex image encryption 2018;148:172–92. doi:10.1016/j.sigpro.2018.02.011.
[14] Wu J, Liao X, Yang B. Cryptanalysis and enhancements of image encryp-
scheme (e.g. multi-round PTDWOS), the modern advanced at-
tion based on three-dimensional bit matrix permutation. Signal Process
tacking methods in classic text cryptanalysis should be utilized, 2018;142:292–300. doi:10.1016/j.sigpro.2017.06.014.
especially that published in the top security conferences re- [15] Li M, Fan H, Xiang Y, Li Y, Zhang Y. Cryptanalysis and improvement of a
cently. chaotic image encryption by first-order time-delay system. IEEE Multimed.
2018;25(3):92–101. doi:10.1109/MMUL.2018.112142439.
[16] Wang H, Xiao D, Chen X, Huang H. Cryptanalysis and enhancements of im-
5. Conclusion age encryption using combination of the 1D chaotic map. Signal Process
2018;144:444–52. doi:10.1016/j.sigpro.2017.11.005.
[17] Panwar K, Purwar RK, Jain A. Cryptanalysis and improvement of an image en-
This paper reviewed the representative works on protecting and cryption scheme using combination of one-dimensional chaotic maps. J Elec-
attacking image data proposed in 2018 from the viewpoint of a se- tron Imaging 2018;27(5). art. no. 053037. doi: 10.1117/1.JEI.27.5.053037.
 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361 7

[18] Chen J, Han F, Qian W, Yao Y-D, liang Zhu Z. Cryptanalysis and improvement its application. Signal Process 2018;153:231–42. doi:10.1016/j.sigpro.2018.07.
in an image encryption scheme using combination of the 1D chaotic map. 011.
Nonlinear Dyn 2018;93(4):2399–413. doi:10.1007/s11071- 018- 4332- 9. [47] Zhu S, Zhu C. Image encryption algorithm with an avalanche effect
[19] Zhu C, Wang G, Sun K. Improved cryptanalysis and enhancements of an image based on a six-dimensional discrete chaotic system. Multimed Tools Appl
encryption scheme using combined 1D chaotic maps. Entropy 2018;20(11). 2018;77(21):29119–42. doi:10.1007/s11042- 018- 6078- 2.
art. no. 843. doi: 10.3390/e20110843. [48] Li C, Feng B, Li S, Kurths J, Chen G. Dynamic analysis of digital chaotic maps
[20] Feng W, He Y-G. Cryptanalysis and improvement of the hyper-chaotic image via state-mapping networks. IEEE Trans Circuits Syst I 2019;66(6):2322–35.
encryption scheme based on DNA encoding and scrambling. IEEE Photonics J doi:10.1109/TCSI.2018.2888688.
2018;10(6). art. no. 7909215. doi: 10.1109/JPHOT.2018.2880590. [49] Yin Q, Wang C. A new chaotic image encryption scheme using breadth-first
[21] Elsadany AA, Yousef AM, Elsonbaty A. Further analytical bifurcation analysis search and dynamic diffusion. Int J Bifurcation Chaos 2018;28(4). art. no.
and applications of coupled logistic maps. Appl Math Comput 2018;338:314– 1850047. doi: 10.1142/S0218127418500475.
36. doi:10.1016/j.amc.2018.06.008. [50] Liu L, Hao S, Lin J, Wang Z, Hu X, Miao S. Image block encryption algo-
[22] Huang X, Ye G. An image encryption algorithm based on irregular rithm based on chaotic maps. IET Signal Process 2018;12(1):22–30. doi:10.
wave representation. Multimed Tools Appl 2018;77(2):2611–28. doi:10.1007/ 1049/iet-spr.2016.0584.
s11042- 017- 4455- x. [51] Mondal B, Kumar P, Singh S. A chaotic permutation and diffusion based im-
[23] Ahmad J, Khan MA, Ahmed F, Khan JS. A novel image encryption scheme age encryption algorithm for secure communications. Multimed Tools Appl
based on orthogonal matrix, skew tent map, and XOR operation. Neural Com- 2018;77(23):31177–98. doi:10.1007/s11042- 018- 6214- z.
put Appl 2018;30(12):3847–57. doi:10.10 07/s0 0521- 017- 2970- 3. [52] Liu J, Ma Y, Li S, Lian J, Zhang X. A new simple chaotic system and its applica-
[24] Hua Z, Yi S, Zhou Y, Li C, Wu Y. Designing hyperchaotic cat maps with tion in medical image encryption. Multimed Tools Appl 2018;77(17):22787–
any desired number of positive Lyapunov exponents. IEEE Trans Cybern 808. doi:10.1007/s11042- 017- 5534- 8.
2018;48(2):463–73. doi:10.1109/TCYB.2016.2642166. [53] Ping P, Xu F, Mao Y, Wang Z. Designing permutation-substitution image
[25] Luo Y, Zhou R, Liu J, Qiu S, Cao Y. An efficient and self-adapting encryption networks with Henon map. Neurocomputing 2018;283:53–63.
colour-image encryption algorithm based on chaos and interactions among doi:10.1016/j.neucom.2017.12.048.
multiple layers. Multimed Tools Appl 2018;77(20):26191–217. doi:10.1007/ [54] Ping P, Wu J, Mao Y, Xu F, Fan J. Design of image cipher using life-like cellu-
s11042- 018- 5844- 5. lar automata and chaotic map. Signal Process 2018;150:233–47. doi:10.1016/j.
[26] Zhang Y, Tang Y. A plaintext-related image encryption algorithm sigpro.2018.04.018.
based on chaos. Multimed Tools Appl 2018;77(6):6647–69. doi:10.1007/ [55] Sahari ML, Boukemara I. A pseudo-random numbers generator based on a
s11042- 017- 4577- 1. novel 3D chaotic map with an application to color image encryption. Nonlin-
[27] Kumar S, Kumar M, Budhiraja R, Das MK, Singh S. A cryptographic model for ear Dyn 2018;94(1):723–44. doi:10.1007/s11071- 018- 4390- z.
better information security. J Inf Secur Appl 2018;43:123–38. doi:10.1016/j. [56] Ping P, Fan J, Mao Y, Xu F, Gao J. A chaos based image encryption scheme
jisa.2018.10.011. using digit-level permutation and block diffusion. IEEE Access 2018;6:67581–
[28] Lakshmanan S, Prakash M, Lim CP, Rakkiyappan R, Balasubramaniam P, Na- 93. doi:10.1109/ACCESS.2018.2879565.
havandi S. Synchronization of an inertial neural network with time-varying [57] Li Z, Peng C, Li L, Zhu X. A novel plaintext-related image encryption scheme
delays and its application to secure communication. IEEE Trans Neural Netw using hyper-chaotic system. Nonlinear Dyn 2018;94(2):1319–33. doi:10.1007/
Learn Syst 2018;29(1):195–207. doi:10.1109/TNNLS.2016.2619345. s11071- 018- 4426- 4.
[29] Hu B, Guan Z-H, Xiong N, Chao H-C. Intelligent impulsive synchronization of [58] Sheela SJ, Suresh KV, Tandur D. Image encryption based on modified
nonlinear interconnected neural networks for image protection. IEEE Trans Henon map using hybrid chaotic shift transform. Multimed Tools Appl
Ind Inf 2018;14(8):3775–87. doi:10.1109/TII.2018.2808966. 2018;77(19):25223–51. doi:10.1007/s11042- 018- 5782- 2.
[30] Diab H, El-semary AM. Secure image cryptosystem with unique key streams [59] Rajagopalan S, Rethinam S, Arumugham S, Upadhyay HN, Rayappan JBB,
via hyper-chaotic system. Signal Process 2018;142:53–68. doi:10.1016/j.sigpro. Amirtharajan R. Networked hardware assisted key image and chaotic
2017.06.028. attractors for secure RGB image communication. Multimed Tools Appl
[31] Fan H, Li M, Liu D, An K. Cryptanalysis of a plaintext-related chaotic RGB 2018;77(18):23449–82. doi:10.1007/s11042- 017- 5566- 0.
image encryption scheme using total plain image characteristics. Multimed [60] Teng L, Wang X, Meng J. A chaotic color image encryption using integrated
Tools Appl 2018;77(15):20103–27. doi:10.1007/s11042-017-5437-8. bit-level permutation. Multimed Tools Appl 2018;77(6):6883–96. doi:10.1007/
[32] Ahmad M, Solami EA, Wang X-Y, Doja MN, Beg MMS, Alzaidi AA. Cryptanaly- s11042- 017- 4605- 1.
sis of an image encryption algorithm based on combined chaos for a BAN sys- [61] Cao C, Sun K, Liu W. A novel bit-level image encryption algorithm based
tem, and improved scheme using SHA-512 and hyperchaos. Symmetry-Basel on 2D-LICM hyperchaotic map. Signal Process 2018;143:122–33. doi:10.1016/
2018;10(7). art. no. 266. doi: 10.3390/sym10070266. j.sigpro.2017.08.020.
[33] Luo Y, Zhang D, Liu J, Liu Y, Cao Y, Ding X. Cryptanalysis of chaos-based cryp- [62] Becheikh R, Omrani T, Rhouma R, Belghith S. RISC: a robust image symmet-
tosystem from the hardware perspective. Int J Bifurcation Chaos 2018;28(9). ric cryptosystem. Multimed Tools Appl 2018;77(19):24615–42. doi:10.1007/
art. no. 1850114. doi: 10.1142/S0218127418501146. s11042- 017- 5575- z.
[34] Khelifi F. On the security of a stream cipher in reversible data hiding schemes [63] Ghebleh M, Kanso A, Stevanovic D. A novel image encryption algorithm based
operating in the encrypted domain. Signal Process 2018;143:336–45. doi:10. on piecewise linear chaotic maps and least squares approximation. Multimed
1016/j.sigpro.2017.09.020. Tools Appl 2018;77(6):7305–26. doi:10.1007/s11042- 017- 4634- 9.
[35] Özkaynak F. Brief review on application of nonlinear dynamics in image en- [64] Wang Y, Zhao Y, Zhou Q, Lin Z. Image encryption using partitioned cellular
cryption. Nonlinear Dyn 2018;92(2):305–13. doi:10.1007/s11071- 018- 4056- x. automata. Neurocomputing 2018;275:1318–32. doi:10.1016/j.neucom.2017.09.
[36] Helmy M, El-Rabaie E-SM, Eldokany IM, El-Samie FEA. Chaotic encryption 068.
with different modes of operation based on Rubik’s cube for efficient wireless [65] Shao Z, Shang Y, Fu X, Yuan H, Shu H. Double-image cryptosystem using
communication. Multimed Tools Appl 2018;77(20):27337–61. doi:10.1007/ chaotic map and mixture amplitude-phase retrieval in gyrator domain. Mul-
s11042- 018- 5923- 7. timed Tools Appl 2018;77(1):1285–98. doi:10.1007/s11042- 016- 4279- 0.
[37] Muhammad K, Hamza R, Ahmad J, Lloret J, Wang H, Baik SW. Secure surveil- [66] Hanis S, Amutha R. Double image compression and encryption scheme us-
lance framework for IoT systems using probabilistic image encryption. IEEE ing logistic mapped convolution and cellular automata. Multimed Tools Appl
Trans Ind Inf 2018;14(8):3679–89. doi:10.1109/TII.2018.2791944. 2018;77(6):6897–912. doi:10.1007/s11042- 017- 4606- 0.
[38] Tang X, Chen G, Lu T. Some iterative properties of f-chaos in nonautonomous [67] Yu C, Li J, Li X, Ren X, Gupta BB. Four-image encryption scheme based on
discrete systems. Entropy 2018;20(3):188. doi:10.3390/e20030188. quaternion fresnel transform, chaos and computer generated hologram. Mul-
[39] Hua Z, Zhou B, Zhou Y. Sine-transform-based chaotic system with FPGA im- timed Tools Appl 2018;77(4):4585–608. doi:10.1007/s11042- 017- 4637- 6.
plementation. IEEE Trans Ind Electron 2018;65(3):2557–66. doi:10.1109/TIE. [68] Wang Q, Wei M, Chen X, Miao Z. Joint encryption and compression of
2017.2736515. 3D images based on tensor compressive sensing with non-autonomous
[40] Hua Z, Zhou Y. One-dimensional nonlinear model for producing chaos. IEEE 3D chaotic system. Multimed Tools Appl 2018;77(2):1715–34. doi:10.1007/
Trans Circ Syst I 2018;65(1):235–46. doi:10.1109/TCSI.2017.2717943. s11042- 017- 4349- y.
[41] Chen S, Yu S, Lu J, Chen G, He J. Design and FPGA-based realization of a [69] Shao Z, Shang Y, Tong Q, Ding H, Zhao X, Fu X. Multiple color image en-
chaotic secure video communication system. IEEE Trans Circuits Syst Video cryption and authentication based on phase retrieval and partial decryption
Technol 2018;28(9):2359–71. doi:10.1109/TCSVT.2017.2703946. in quaternion gyrator domain. Multimed Tools Appl 2018;77(19):25821–40.
[42] Bouslehi H, Seddik H. A new rapid hyperchaotic system for more efficient doi:10.1007/s11042- 018- 5818- 7.
2D data encryption. Multimed Tools Appl 2018;77(6):7741–62. doi:10.1007/ [70] Kaur M, Kumar V. Beta chaotic map based image encryption using genetic
s11042- 017- 4675- 0. algorithm. Int J Bifurcation Chaos 2018;28(11). art. no. 1850132. doi: 10.1142/
[43] Hua Z, Jin F, Xu B, Huang H. 2D logistic-sine-coupling map for image encryp- S0218127418501328.
tion. Signal Process 2018;149:148–61. doi:10.1016/j.sigpro.2018.03.010. [71] Noshadian S, Ebrahimzade A, Kazemitabar SJ. Optimizing chaos based im-
[44] Gayathri J, Subashini S. A spatiotemporal chaotic image encryption age encryption. Multimed Tools Appl 2018;77(19):25569–90. doi:10.1007/
scheme based on self adaptive model and dynamic keystream fetching s11042- 018- 5807- x.
technique. Multimed Tools Appl 2018;77(19):24751–87. doi:10.1007/ [72] Mozaffari S. Parallel image encryption with bitplane decomposition and ge-
s11042- 018- 5675- 4. netic algorithm. Multimed Tools Appl 2018;77(19):25799–819. doi:10.1007/
[45] Hua Z, Yi S, Zhou Y. Medical image encryption using high-speed scrambling s11042- 018- 5817- 8.
and pixel adaptive diffusion. Signal Process 2018;144:134–44. doi:10.1016/j. [73] Huang H, He X, Xiang Y, Wen W, Zhang Y. A compression-diffusion-
sigpro.2017.10.004. permutation strategy for securing image. Signal Process 2018;150:183–90.
[46] Yang B, Liao X. Some properties of the logistic map over the finite field and doi:10.1016/j.sigpro.2018.04.014.
8 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361

[74] Lv X, Liao X, Yang B. A novel scheme for simultaneous image compression [102] Shyu SJ. XOR-based visual cryptographic schemes with monotonously increas-
and encryption based on wavelet packet transform and multi-chaotic systems. ing and flawless reconstruction properties. IEEE Trans Circ Syst Video Technol
Multimed Tools Appl 2018;77(21):28633–63. doi:10.1007/s11042- 018- 6013- 6. 2018;28(9):2397–401. doi:10.1109/TCSVT.2017.2707923.
[75] Zhang D, Liao X, Yang B, Zhang Y. A fast and efficient approach to color-image [103] Liu Y-X, Yang C-N, Wu S-Y, Chou Y-S. Progressive (k, n) secret image shar-
encryption based on compressive sensing and fractional fourier transform. ing schemes based on boolean operations and covering codes. Signal Process–
Multimed Tools Appl 2018;77(2):2191–208. doi:10.1007/s11042- 017- 4370- 1. Image Commun 2018;66:77–86. doi:10.1016/j.image.2018.05.004.
[76] Pudi V, Chattopadhyay A, Lam K-Y. Secure and lightweight compressive sens- [104] Yan X, Lu Y, Liu L, Liu J, Yang G. Chinese Remainder Theorem-based two-in-
ing using stream cipher. IEEE Trans Circ SystII 2018;65(3):371–5. doi:10.1109/ one image secret sharing with three decoding options. Digit Signal Process
TCSII.2017.2715659. 2018;82:80–90. doi:10.1016/j.dsp.2018.07.015.
[77] Wang X-Y, Li P, Zhang Y-Q, Liu L-Y, Zhang H, Wang X. A novel color image [105] Yuan X, Wang X, Wang C, Squicciarini AC, Ren K. Towards privacy-preserving
encryption scheme using DNA permutation based on the Lorenz system. Mul- and practical image-centric social discovery. IEEE Trans Depend Secure Com-
timed Tools Appl 2018;77(5):6243–65. doi:10.1007/s11042- 017- 4534- z. put 2018;15(5):868–82. doi:10.1109/TDSC.2016.2609930.
[78] Maddodi G, Awad A, Awad D, Awad M, Lee B. A new image encryp- [106] Gunasinghe H, Bertino E. Privbiomtauth: privacy preserving biometrics-based
tion algorithm based on heterogeneous chaotic neural network generator and user centric protocol for user authentication from mobile phones. IEEE
and DNA encoding. Multimed Tools Appl 2018;77(19):24701–25. doi:10.1007/ Trans Inf Forensics Secur 2018;13(4):1042–57. doi:10.1109/TIFS.2017.2777787.
s11042- 018- 5669- 2. [107] Wang Q, He M, Du M, Chow SSM, Lai RWF, Zou Q. Searchable encryption over
[79] Girdhar A, Kumar V. A RGB image encryption technique using Lorenz feature-rich data. IEEE Trans Dependable Secure Comput 2018;15(3):496–510.
and Rossler chaotic system on DNA sequences. Multimed Tools Appl doi:10.1109/TDSC.2016.2593444.
2018;77(20):27017–39. doi:10.1007/s11042- 018- 5902- z. [108] Xia Z, Zhu Y, Sun X, Qin Z, Ren K. Towards privacy-preserving content-based
[80] Slimane NB, Aouf N, Bouallegue K, Machhout M. A novel chaotic image cryp- image retrieval in cloud computing. IEEE Trans Cloud Comput 2018;6(1):276–
tosystem based on DNA sequence operations and single neuron model. Mul- 86. doi:10.1109/TCC.2015.2491933.
timed Tools Appl 2018;77(23):30993–1019. doi:10.1007/s11042- 018- 6145- 8. [109] Khan M, Waseem HM. A novel image encryption scheme based on quantum
[81] Fu X-Q, Liu B-C, Xie Y-Y, Li W, Liu Y. Image encryption-then-transmission dynamical spinning and rotations. PLoS ONE 2018;13(11). art. no. e0206460.
using DNA encryption algorithm and the double chaos. IEEE Photonics J doi: 10.1371/journal.pone.0206460.
2018;10(3). art. no. 3900515. doi: 10.1109/JPHOT.2018.2827165. [110] El-Latif AAA, Abd-El-Atty B, Talha M. Robust encryption of quantum medical
[82] Sun S. A novel hyperchaotic image encryption scheme based on DNA en- images. IEEE Access 2018;6:1073–81. doi:10.1109/ACCESS.2017.2777869.
coding, pixel-level scrambling and bit-level scrambling. IEEE Photonics J [111] Zhou N, Yan X, Liang H, Tao X, Li G. Multi-image encryption scheme based on
2018;10(2). art. no. 7201714. doi: 10.1109/JPHOT.2018.2817550. quantum 3D arnold transform and scaled Zhongtang chaotic system. Quan-
[83] Wu J, Liao X, Yang B. Image encryption using 2D Hénon-Sine map and DNA tum Inf Process 2018;17(12). art. no. 338. doi: 10.1007/s11128- 018- 2104- 6.
approach. Signal Process 2018;153:11–23. doi:10.1016/j.sigpro.2018.06.008. [112] Li C, Lin D, Lü J. Cryptanalyzing an image-scrambling encryption algorithm of
[84] Wen W, Zhang Y, Fang Y, Fang Z. Image salient regions encryption pixel bits. IEEE Multimedia 2017;3:64–71. doi:10.1109/MMUL.2017.3051512.
for generating visually meaningful ciphertext image. Neural Comput Appl [113] Chuman T, Kiya H. Security evaluation for block scrambling-based image en-
2018;29(3):653–63. doi:10.10 07/s0 0521- 016- 2490- 6. cryption including JPEG distortion against jigsaw puzzle solver attacks. IE-
[85] Li X, Zhao M, Xing Y, et al. Designing optical 3D images encryption and recon- ICE Trans Fund ElectronCommun Comput Sci 2018;E101A(12):2405–8. doi:10.
struction using monospectral synthetic aperture integral imaging. Opt Express 1587/transfun.E101.A.2405.
2018;26(9):11084–99. doi:10.1364/OE.26.011084. [114] Imaizumi S, Kiya H. A block-permutation-based encryption scheme with in-
[86] Ren G, Han J, Fu J, Shan M. Asymmetric image encryption using phase- dependent processing of RGB components. IEICE Trans Inf Syst 2018;E101–
truncated discrete multiple-parameter fractional fourier transform. Opt Rev D(12):3150–7. doi:10.1587/transinf.2018EDT0 0 02.
2018;25(6):701–7. doi:10.1007/s10043- 018- 0464- x. [115] Li C. Cracking a hierarchical chaotic image encryption algorithm based on per-
[87] Kong D, Cao L, Shen X, Zhang H, Jin G. Image encryption based on interleaved mutation. Signal Process 2016;118:203–10. doi:10.1016/j.sigpro.2015.07.008.
computer-generated holograms. IEEE Trans Ind Inf 2018;14(2):673–8. doi:10. [116] Preishuber M, Huetter T, Katzenbeisser S, Uhl A. Depreciating motivation and
1109/TII.2017.2714261. empirical security analysis of chaos-based image and video encryption. IEEE
[88] Chen B, Yu M, Tian Y, Li L, Wang D, Sun X. Multiple-parameter fractional Trans Inf Forensics Secur 2018;13(9):2137–50. doi:10.1109/TIFS.2018.2812080.
quaternion fourier transform and its application in colour image encryption. [117] Li C, Lin D, Lü J, Hao F. Cryptanalyzing an image encryption algorithm based
IET Image Proc 2018;12(12):2238–49. doi:10.1049/iet-ipr.2018.5440. on autoblocking and electrocardiography. IEEE Multimed 2018;25(4):46–56.
[89] Chen H, Tanougast C, Liu Z, Blondel W, Hao B. Optical hyperspectral image doi:10.1109/MMUL.2018.2873472.
encryption based on improved Chirikov mapping and gyrator transform. Opt [118] Zhang L, Liu Y, Wang C, Zhou J, Zhang Y, Chen G. Improved known-plaintext
Lasers Eng 2018;107:62–70. doi:10.1016/j.optlaseng.2018.03.011. attack to permutation-only multimedia ciphers. Inf Sci 2018;430–431(3):228–
[90] Kumar R, Bhaduri B, Hennelly B. QR code-based non-linear image en- 39. doi:10.1016/j.ins.2017.11.021.
cryption using Shearlet transform and spiral phase transform. J Mod Opt [119] Zhu C, Sun K. Cryptanalyzing and improving a novel color image encryption
2018;65(3):321–30. doi:10.1080/09500340.2017.1395486. algorithm using RT-enhanced chaotic tent maps. IEEE Access 2018;6:18759–
[91] Sun W, Zhou J, Zhu S, Tang YY. Robust privacy-preserving image sharing over 70. doi:10.1109/ACCESS.2018.2817600.
online social networks (OSNs). ACM Trans Multimed Comput Commun Appl [120] Zhang LY, Liu Y, Pareschi F, Zhang Y, Wong K-W, Rovatti R, et al. On the se-
2018;14(1). art. no. 14. doi: 10.1145/3165265. curity of a class of diffusion mechanisms for image encryption. IEEE Trans
[92] He J, Huang S, Tang S, Huang J. JPEG image encryption with improved Cybern 2018;48(4):1163–75. doi:10.1109/TCYB.2017.2682561.
format compatibility and file size preservation. IEEE Trans Multimedia [121] Fan H, Li M, Liu D, Zhang E. Cryptanalysis of a colour image encryption using
2018;20(10):2645–58. doi:10.1109/TMM.2018.2817065. chaotic APFM nonlinear adaptive filter. Signal Process 2018;143:28–41. doi:10.
[93] Zheng P, Huang J. Efficient encrypted images filtering and transform coding 1016/j.sigpro.2017.08.018.
with Walsh-Hadamard transform and parallelization. IEEE Trans Image Pro- [122] Li M, Guo Y, Huang J, Li Y. Cryptanalysis of a chaotic image encryption
cess 2018;27(5):2541–56. doi:10.1109/TIP.2018.2802199. scheme based on permutation-diffusion structure. Signal Process-Image Com-
[94] Puteaux P, Puech W. An efficient MSB prediction-based method for high- mun 2018;62:164–72. doi:10.1016/j.image.2018.01.002.
capacity reversible data hiding in encrypted images. IEEE Trans Inf Forensics [123] Li M, Lu D, Wen W, Ren H, Zhang Y. Cryptanalyzing a color image encryption
Secur 2018;13(7):1670–81. doi:10.1109/TIFS.2018.2799381. scheme based on hybrid hyper-chaotic system and cellular automata. IEEE Ac-
[95] Chen B, Wu X, Wei Y-S. Reversible data hiding in encrypted images with cess 2018;6:47102–11. doi:10.1109/ACCESS.2018.2867111.
private-key homomorphism and public-key homomorphism. J Vis Commun [124] Farajallah M, El Assad S, Deforges O. Cryptanalyzing an image encryption
Image Represent 2018;57:272–82. doi:10.1016/j.jvcir.2018.11.017. scheme using reverse 2-dimensional chaotic map and dependent diffusion.
[96] Bhardwaj R. Enhanced encrypted reversible data hiding algorithm with Multimed Tools Appl 2018;77(21):28225–48. doi:10.1007/s11042- 018- 6015- 4.
minimum distortion through homomorphic encryption. J Electron Imaging [125] Dhall S, Pal SK, Sharma K. Cryptanalysis of image encryption scheme based
2018;27(2). art. no. 023017. doi: 10.1117/1.JEI.27.2.023017. on a new 1D chaotic system. Signal Process 2018;146:22–32. doi:10.1016/j.
[97] Puteaux P, Puech W. An efficient MSB prediction-based method for high- sigpro.2017.12.021.
capacity reversible data hiding in encrypted images. IEEE Trans Inf Forensics [126] Hoang TM, Thanh HX. Cryptanalysis and security improvement for a symmet-
Secur 2018;13(7):1670–81. doi:10.1109/TIFS.2018.2799381. ric color image encryption algorithm. Optik 2018;155:366–83. doi:10.1016/j.
[98] Qian Z, Zhou H, Zhang X, Zhang W. Separable reversible data hid- ijleo.2017.10.072.
ing in encrypted JPEG bitstreams. IEEE Trans Dependable Secure Comput [127] Xiong Y, He A, Quan C. Security analysis of a double-image encryption tech-
2018;15(6):1055–67. doi:10.1109/TDSC.2016.2634161. nique based on an asymmetric algorithm. J Opt Soc Am A 2018;35(2):320–6.
[99] Xiang S, Luo X. Reversible data hiding in homomorphic encrypted do- doi:10.1364/JOSAA.35.0 0 0320.
main by mirroring ciphertext group. IEEE Trans Circuits Syst Video Technol [128] Xiong Y, He A, Quan C. Cryptanalysis of an optical cryptosystem based on
2018;28(11):3099–110. doi:10.1109/TCSVT.2017.2742023. phase-truncated fourier transform and nonlinear operations. Opt Commun
[100] Wang C, Xiao D, Peng H, Zhang R. A lossy compression scheme for encrypted 2018;428:120–30. doi:10.1016/j.optcom.2018.07.058.
images exploiting Cauchy distribution and weighted rate distortion optimiza- [129] Singh KM, Singh LD, Tuithung T. Cryptanalysis of multimedia encryption us-
tion. J Vis Commun Image Represent 2018;51:122–30. doi:10.1016/j.jvcir.2018. ing elliptic curve cryptography. Optik 2018;168:370–5. doi:10.1016/j.ijleo.2018.
01.007. 04.068.
[101] Wang C, Ni J, Zhang X, Huang Q. Efficient compression of encrypted bi- [130] Hai H, Liao M, Lu D, He W, Peng X. Cryptanalysis on double random phase
nary images using the Markov random field. IEEE Trans Inf Forensics Secur encoding with deep learning. Optics and Photonics for Information Processing
2018;13(5):1271–85. doi:10.1109/TIFS.2017.2784379. XII; 2018. W. doi: 10.1117/12.2319902.
 C. Li, Y. Zhang and E.Y. Xie / Journal of Information Security and Applications 000 (2019) 102361 9

[131] Annaby MH, Ayad H, Rushdi MA. On security of image ciphers based on logic [135] Noura M, Noura H, Chehab A, Mansour MM, Sleem L, Couturier R. A dynamic
circuits and chaotic permutations. Multimed Tools Appl 2018;77(16):20455– approach for a lightweight and secure cipher for medical images. Multimed
76. doi:10.1007/s11042- 017- 5439- 6. Tools Appl 2018;77(23):31397–426. doi:10.1007/s11042- 018- 6051- 0.
[132] Lin Z, Wang G, Wang X, Yu S, Lü J. Security performance analysis of [136] Entacher K. Bad subsequences of well-known linear congruential pseudoran-
a chaotic stream cipher. Nonlinear Dyn 2018;94(2):1003–17. doi:10.1007/ dom number generators. ACM Trans Model ComputSimul 1998;8(1):61–70.
s11071- 018- 4406- 8. doi:10.1145/272991.273009.
[133] Lin Z, Yu S, Feng X, Lü J. Cryptanalysis of a chaotic stream cipher and its [137] Álvarez G, Li S. Some basic cryptographic requirements for chaos-
improved scheme. Int J Bifur Chaos 2018;28(7). art. no. 1850086. doi: 10.1142/ based cryptosystems. Int J Bifur Chaos 2006;16(8):2129–51. doi:10.1142/
S0218127418500864. S0218127406015970.
[134] Li P, Lo K-T. A content-adaptive joint image compression and encryption
scheme. IEEE Trans Multimed 2018;20(8):1960–72. doi:10.1109/TMM.2017.
2786860.