Image Steganography Approaches and Their Detection

Strategies: A Survey

MEIKE HELENA KOMBRINK, Digital and Biometric Traces, Netherlands Forensic Institute, Den
Haag, Netherlands and Informatics Institute, University of Amsterdam Faculty of Science, Amsterdam,
Netherlands
ZENO JEAN MARIUS HUBERT GERADTS, Digital and Biometric Traces, Netherlands Forensic
Institute, Den Haag, Netherlands and Informatics Institute, University of Amsterdam Faculty of Science,
Amsterdam, Netherlands
MARCEL WORRING, Informatics Institute, University of Amsterdam Faculty of Science, Amsterdam,
Netherlands

Steganography is the art and science of hidden (or covered) communication. In digital steganography, the
bits of image, video, audio and text files are tweaked to represent the information to hide. This article cov-
ers the current methods for hiding information in images, alongside steganalysis methods that aim to detect
the presence of steganography. By reviewing 456 references, this article discusses the different approaches
that can be taken toward steganography and its much less widely studied counterpart. Currently in research
older steganography approaches are more widely used than newer methods even though these show greater
potential. New methods do have flaws; therefore, more research is needed to make these practically applica-
ble. For steganalysis one of the greatest challenges is the generalisability. Often one scheme can detect the
presence of one specific hiding method. More research is needed to combine current schemes and/or create
new generalisable schemes. To allow readers to compare results between different papers in our work, per-
formance indications of all steganalysis methods are outlined and a comparison of performance is included.
This comparison is given using ‘topological sorting’ graphs, which compares detection results from all papers
(as stated in the papers themselves) on different steganographic schemes.
CCS Concepts: • Security and privacy → Information-theoretic techniques; Pseudonymity,
anonymity and untraceability; Information flow control;
Additional Key Words and Phrases: Steganography, steganalysis, forensic image analysis
ACM Reference Format:
Meike Helena Kombrink, Zeno Jean Marius Hubert Geradts, and Marcel Worring. 2024. Image Steganography
Approaches and Their Detection Strategies: A Survey. ACM Comput. Surv. 57, 2, Article 33 (October 2024), 40
pages. https://doi.org/10.1145/3694965

The work presented in this article received funding from the European Union’s Horizon 2020 research and innovation
program under grant agreement No. 101021687 (project “UNCOVER”).
Authors’ Contact Information: Meike Helena Kombrink, Digital and Biometric Traces, Netherlands Forensic Institute,
Den Haag, Zuid Holland, Netherlands and Informatics Institute, University of Amsterdam Faculty of Science, Amster-
dam, Noord-Holland, Netherlands; e-mail: [email protected]; Zeno Jean Marius Hubert Geradts, Digital and Biometric
Traces, Netherlands Forensic Institute, Den Haag, Zuid Holland, Netherlands and Informatics Institute, University of Am-
sterdam Faculty of Science, Amsterdam, Noord-Holland, Netherlands; e-mail: [email protected]; Marcel Worring,
Informatics Institute, University of Amsterdam Faculty of Science, Amsterdam, Noord-Holland, Netherlands; e-mail:
[email protected].

This work is licensed under a Creative Commons Attribution International 4.0 License.
© 2024 Copyright held by the owner/author(s).
ACM 0360-0300/2024/10-ART33
https://doi.org/10.1145/3694965

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:2 M. H. Kombrink et al.

1 Introduction
The word ‘steganography’ is a combination of two words, ‘steganos’ (cover or protection) and
“graphy” (to write). Thus, steganography literally means covered writing [167, 258]. Digital
steganography is concerned with hiding any type of data in any type of cover medium (image,
audio, video, text) in such a way that no third party will suspect the existence of it [74]. It is in this
sense that steganography is different from cryptography, which aims to protect the content of a
message rather than its existence [167, 352]. For steganography one needs a cover medium (e.g.,
an image), a secret message to hide inside this cover, and a hiding procedure. Additionally, to in-
crease security, one can encrypt the hidden information. Steganalysis is concerned with detection
of the presence of hidden information. It does not go further than binary classification based on
a presented image telling the user whether steganography was used or not. Forensic steganalysis
goes a step further and is concerned with derivation of any extra information about the hidden
message (e.g., message length, embedding scheme used, and ultimately its content). See Figure 1
for a graphical summary of what makes up the process of steganography and the process of
(forensic) steganalysis.
Steganography can be applied in
many situations, including secure trans-
mission of classified documents and se-
curing online banking [74, 260]. But
steganography can also be used for
criminal purposes. In the non-digital
era the Nazis wrote over cover objects
with invisible ink [49, 135, 200]. In more
recent years digital steganography ex-
amples of use include the preparations
of the 9/11 attacks (as US officials and Fig. 1. A graphical representation of what steganography
many articles claim), a child pornog- and steganalysis include. For steganography, a sender and
raphy network, and communication of a receiver of hidden information are displayed in the blocks
10 Russian spies in America (claims on the left. Steganalysis, the block on the right, looks into
the FBI) [74]. Given its malicious use, the question of whether a message is hidden, while foren-
it is important to develop steganalysis sic steganalysis aims to discover more information on the
methods to actively look for the hidden hidden message.
messages within digital media.
Every medium (video, audio, text, and image files) has particular steganography methods.
Images are most popular [169] as they possess a high degree of redundancy [352] and have a
widespread dissemination [92]. This review will restrict itself to image steganography and image
steganalysis.
Current reviews within steganography and steganalysis have several shortcomings because of
which they do not provide a clear and complete overview. Many were published in or before 2014,
10 years ago; therefore, recent developments in the fields are not covered [22, 41, 49, 80, 120, 155,
238, 297]. Other reviews have fewer than 60 references, which cannot provide a complete overview
[64, 76, 80, 158, 303]. Other reviews have shortcomings too. First, they often go into individual
implementations of methodologies, which fails to give the reader an overview of the different
types of implementations within each approach [89, 92, 171, 262, 318, 330, 354]. Additionally, many
review papers only cover some approaches (e.g., only spatial domain or only one approach from
each domain), which inherently fails to give an overview of the entire workfield [74, 76, 89, 92,
157, 187, 203, 330, 354, 387]. Lastly, very few papers provide a full overview of both steganography
and (forensic) steganalysis, failing to show how these two are connected and influence each other

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:3

Fig. 2. Structural difference of the approaches that can be taken to steganography (depicted from left to
right) and steganalysis (depicted from right to left).

[76, 82, 92, 157, 166, 169, 171, 187, 203, 244, 271, 314, 318, 354, 387]. Our article aims to solve all
of the above shortcomings. By providing the reader an overview of the complete steganography
and the complete steganalysis field of research. It gives a comprehensive review by including many
(though not all) sources about both steganography and steganalysis and linking them conceptually
throughout the text and empirically through the performance-based topological sorting graphs.
Figure 2 shows our structural division of image steganography and steganalysis approaches. We
divide steganography into two general approaches: adaptive approaches and static approaches.
Adaptive approaches distinguish themselves through an additional statistical analysis to deter-
mine the location of the embedding. Both general approaches mostly use one of three different
approaches toward embedding (hereforth called domains). The spatial domain approaches (Least
Significant Bit (LSB), Pixel Value Differencing (PVD) and Bit-Plane Complexity Segmen-
tation (BPCS)) hide messages directly in the intensity values of pixels, while transform domain
approaches (Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT),
Integer Wavelet Transform (IWT) and Discrete Fourier Transform (DFT)) first transform
the image and then hide the message in the calculated coefficients. Deep learning approaches
(Convolutional Neural Network (CNN) and Generative Adversarial Network (GAN)) use
self-learned ways to embed a message. Steganalysis approaches, on the other hand, can be differ-
entiated by whether they are steganography method specific or whether they can generalise over
multiple steganography methods. Approaches that are not method specific are visual steganalysis,
where one visually analyses an image, and universal steganalysis, where several methods
are detected by one algorithm. Method-specific approaches are termed as targeted steganalysis,
designed to detect one method, and signature steganalysis, which looks for signature-like patterns
that can appear in images after embedding. This article is structured in accordance with Figure 2,
where we will first go into the static steganography approaches, after which the statistical analysis
of adaptive approaches will be discussed. The steganalysis part of this review is ordered by com-
plexity (from least to most complex). Hence, visual steganalysis will first be discussed, and then
the article will go into signature-based steganalysis, targeted steganalysis and finally universal
steganalysis.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:4 M. H. Kombrink et al.

2 Methodology
Several tactics were used for gathering the collection of papers. First, a search on ‘steganography’
and ‘steganalysis’ combined with ‘survey’ and ‘review’ was conducted. Then, the words ‘steganog-
raphy’ and ‘steganalysis’ were combined with the individual approaches defined in Figure 2: LSB,
PVD, BPCS, DCT, DWT, IWT, DFT, CNN and GAN. Subsequently, citations within all resulting
papers were used to deepen the search and ensure completeness. The word ‘detecting’ was also
combined with all approaches because it stood out that many papers used this word in their title
rather than ‘steganalysis’. The relevance of all papers was assessed before they were selected for
inclusion. This was done in three steps, where if a paper satisfied the criteria in the first step, the
next two were not deemed necessary, and a paper that satisfied the second criteria did not undergo
the third.
(1) First, any paper whose source (journal/conference) is topic specific to steganography
and/or steganalysis was deemed relevant to the review. This step was included because
these journals and conferences often score relatively low on ranking systems. The lower
score some of these journals and conferences received can be explained by the small size
of the steganography community, which results in lower h-scores and fewer citations in
general (factors often used for determining ranks).
(2) Second, non-topic-specific sources were submitted to a ranking system. For journals this
ranking system was twofold, where either one can determine if a source goes through to
the last step.
(a) Scopus: Any journal that was not in Scopus was judged as a journal of lesser quality
and the paper used from these sources would undergo step 3. Any journal that was
within Scopus was deemed of high enough quality to keep in the review, unless the
next journal check deemed it not to be.
(b) Scimago Journal and Country Rank: All journals were looked up within this ranking
system, which uses quartiles within specific fields. It was determined that any source
within the lowest quartile was of lesser quality and should therefore undergo step 3.
Otherwise, the journal was judged to be of good quality and sources used from this
journal could be kept within the review.
For conferences a similar tactic was determined. However, conference ranking systems
are a lot less straightforward when compared to journal ranking systems. A total of five
ranking systems were used within this review, where if more than half of the scores given
to a conference were satisfactory, a conference was deemed to be of high enough quality
to keep within the review:
(a) ERA (2010), which gives conferences scores ranging from A to C; this information was
taken from conferenceranks.1 Conferences that scored an A or B were deemed to be
good enough to keep in the review.
(b) Qualis (2012), which gives conferences scores A1, A2, B1, B2, B3, B4 or B5; this infor-
mation was also taken from conferenceranks.2 Conferences that scored a A1, A2, B1,
B2 or B3 were deemed to be of high enough quality to include in the review.
(c) CORE, whose scores were composed of A++, A+, A, A−, B, B− and C by the GII-GRIN-
SCIE (GGS) Conference Rating.3 Here a C signifies two things: a conference either
is not implemented in the ranking yet or is of the lowest quality. Conferences were
included from when they were scored a B− or higher.

1 http://www.conferenceranks.com/#data
2 Please refer to footnote 1.
3 https://scie.lcc.uma.es/gii- grin- scie- rating/ratingSearch.jsf

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:5

(d) LiveSHINE, whose scores were composed of A++, A+, A, A−, B, B− and C by the GGS
Conference Rating.4 Here a C signifies two things: a conference either is not imple-
mented in the ranking yet or is of the lowest quality. Conferences were included from
when they were scored a B− or higher.
(e) Microsoft Academic, whose scores were composed of A++, A+, A, A−, B, B− and C by
the GGS Conference Rating.5 Here a C signifies two things: a conference either is not
implemented in the ranking yet or is of the lowest quality. Conferences were included
from when they were scored a B− or higher.
(3) If an individual source was not deemed to be of relevant quality based on its journal and/or
conference, the source was individually checked for its relevance within the field. This step
was included because sometimes highly relevant papers are published within sources of
lesser quality. To check the relevance of an individual paper, the average number of cita-
tions per year for each paper was computed. For this, the number of citations as depicted
by Google Scholar was used. The authors are aware that this is most often an overstate-
ment compared to the actual number of citations. However, not all websites from indi-
vidual sources include citation scores and no other metric has been found that could be
applied to every single source within our review. Therefore, despite its drawback, Google
Scholar was deemed most useful since it was the only one that allowed for equal compar-
ison between sources. It was decided that if a source scored at least 10 citations per year
according to Google Scholar, the source was highly relevant to the field and should thus
not be excluded based on the quality of its journals or conference. Any source that scored
below 10 citations per year was excluded from the review.

3 Steganography Methods
For steganography, four characteristics are used to evaluate a method [19, 42, 43, 74, 157, 297, 352].
These parameters are:
—Embedding capacity: the amount of data that can be hidden inside an image expressed in
bits per pixel [171], where higher is better.
—Robustness: the amount of alterations to an image that a secret message can survive, where
the higher this parameter, the better the hiding scheme.
—Security: the resistance, tested through an attack resistance ability measure, a stegano-
graphic image shows towards steganalysis techniques, where higher resistance signifies
higher security.
—Imperceptibility: the ease with which the existence of a hidden message is detected based
on inspection of the quality of the steganographic image alone, where the higher the better.
This criterion is often measured through PSNR scores and/or SSIM scores.
Throughout sources, imperceptibility is often measured through the means of a PSNR score,
though SSIM scores have more recently been used as well. Research suggests that SSIM is a better
measure for imperceptibility than PSNR [329].
Security is often measured by applying different steganalysis schemes to the stego content pro-
duced by a specific steganographic scheme. If the scheme is not detected (often) by the steganalysis
algorithm, it is deemed to be secure.

4 Please refer to footnote 3.

5 Please refer to footnote 3.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:6 M. H. Kombrink et al.

All steganography approaches discussed in this section have the possibility to increase security
by encrypting the secret information [258]. However, this research focuses on steganography and
steganalysis only and will not consider cryptography.

3.1 Spatial Domain Approaches

A spatial domain approach is one that directly alters the intensity of pixels. The most fundamental
method is LSB steganography, but since its discovery more complex and adaptive models have
been developed [65, 157]. Spatial domain approaches have high embedding capacity [155] as well
as low computational complexity [17]. The major drawbacks of spatial domain approaches are a
lack in robustness [17, 155] as well as limited security [17].
The three approaches that were found most often in the literature were LSB, PVD and BPCS.
Though other approaches for hiding information in the spatial domain exist (e.g., Spread spectrum
[100, 247]), they will not be discussed in this article, as their impact is limited.
3.1.1 LSB. LSB steganography is a message-hiding approach that directly changes the bits that
are least significant to the colour of a pixel: the last bit(s). More specifically, it replaces the values of
existing bits with the binary value of the message [372]. The LSB approach is the most conventional
[235, 330] and easiest to implement [166, 200, 330] steganographic approach. However, LSB has
low robustness [166, 330] and the lowest security of all approaches [166, 330].
Differences in LSB approaches exist and are mainly concerned with which bits’ value to alter.
The most ‘basic’ method—LSB replacement (LSBR)—changes the values of the bits in the leftmost
top corner and goes right and down [120]. Alternatively, Stego-key-directed LSB (sometimes called
Random LSB, though it is far from random) uses a user-specified stego key to distribute the message
over the image [248, 259]. Another alteration is called LSB matching (LSBM), which aims to
‘match’ the binary information to the pixels, so that as few pixels as possible are altered [94, 216,
233, 251, 321, 335]. Edge LSB methods hide more of the secret message in the (sharpest) edges of
the image [46, 57, 159, 385, 413]. Other methods determine the number of bits to be hidden in a
pixel based on its grayscale value, which increases the embedding capacity [220]. Some methods
allow the user of a tool control over where the embedding takes place [168].
To be more imperceptible, Optimal Pixel Adjustment Process (OPAP) was added to basic
methods, which checks whether the change to a pixel falls within ranges that are considered to not
decrease imperceptibility [40]. Using a weight matrix was found to increase embedding capacity,
security and imperceptibility [45, 275, 375, 376], while directing embedding based on characteristic
values per block was found to not decrease imperceptibility [54]. Alternatively, shuffling [398] or
upscaling [164] an image before embedding increases imperceptibility [398]. Altering the image
after LSB embedding was found to increase security [117], while ensuring that the Local Binary
Pattern of an image is preserved [39] improves robustness.
For colour images, clever analysis of the use of Red, Green and Blue (RGB) can help to increase
imperceptibility [343]. Alternatively, the use of one of the RGB values to encode whether a hidden
message is present in the other two colour channels was implemented [133]. Colour space conver-
sion (to Hue Saturation Intesity (HSI)) before hiding a message was also found useful [260].
3.1.2 PVD. To embed using the PVD approach, a cover image is partitioned into blocks of
two consecutive non-overlapping pixels [396]. For each block, the difference between the two
pixels (the difference value (DV)) is calculated and then classified into a specific set of ranges.
Subsequently, the DV is replaced by part of the secret message, where the DV range dictates
how many bits are hidden, ensuring more information is hidden in edges [412]. Results are more
imperceptible than LSB [157, 166, 330], show a larger embedding capacity [157, 166, 330] and
provide higher security [330].

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:7

Later alterations were mainly made to the selection of blocks. Allowing difference calculation
between different directional neighbours (e.g., horizontal, vertical and/or diagonal) was found ef-
fective [47, 134, 224, 291, 349, 359, 412]. Changing the size of the (sometimes overlapping) PVD
blocks was shown to effectively improve results [53, 225, 342]. Alternatively, ensuring the falling-
off-boundary problem was no longer present [46, 319], which is one of the major flaws of PVD,
improves security [53, 322]. Combining PVD with LSB approaches was also found effective [156,
189, 397], where some use both methods within the blocks [156, 189], while others use the DV
to select which method to apply [397]. Similarly, changing the embedding process based on local
contrast of the pixel pair can optimise security and imperceptibility [278]. Improving the selec-
tion of how to change the remainder of the PVD was also tried [382]. Preservation of the natural
histogram of an image increases security [53, 446]. Others combine PVD with different methods
to reduce distortion in the steganographic image [320]. PVD has been paired with IWT [349] and
LSB [156, 189, 397].
3.1.3 BPCS. BPCS was inspired by the human visual system, which cannot perceive any shape
information if it is in a very complex binary pattern [173]. For BPCS, an image is divided into bit-
planes [266, 351]. A segmentation of each bit-plane is used and its complexity is computed, which
determines the noisy blocks [269]. These noisy blocks are then used to hide the secret information.
For BPCS, the information to be hidden should be a complex binary pattern; if it is not it needs to
be transformed into one [269] or risk being easily detected. BPCS shows good embedding capacity
[351] and provides high imperceptibility [351] but low security [269, 357].
Increasing the robustness of BPCS with respect to lossy compression is possible [273, 351].
Enhancing imperceptibility (and in some cases security) is possible by finding the most suitable
method to categorize regions as noisy [266, 267, 357, 358]. Security can be improved by using all bit-
planes [267]. Alternatively, maintaining the complexity distribution of the cover, so that the BPCS
signature is not embedded, was found to increase security [268]. BPCS was paired with IWT [309].

3.2 Transform Domain-based Approaches

Transform domain approaches are those that encode the message in the transform coefficients.
These approaches are more complex than the spatial ones. Alterations made in the transform
domain are generally more robust [157] and secure [17]. However, the embedding capacity is lower
[157]. Computationally, these approaches are a lot more complex than spatial ones [17, 74].
The four most used transforms to hide information in an image are DCT, DWT, DFT and IWT.
Though there are more transforms that can be used for information hiding, these will not be dis-
cussed in this article.
3.2.1 DCT. DCT transforms a signal into its elementary frequency components [74]. It essen-
tially breaks an image down into a summation of sinusoids of varying magnitudes and frequencies.
DCT is the most frequently used transform domain method [74]. For DCT embedding, one first
breaks the image into blocks (usually 8x8); afterwards each block is centered around 0 by substract-
ing 128 [380]. DCT is applied to each block, followed by a quantization table-based compression
of the block. Alterations to this quantization table can then hide a message. In general, DCT-based
approaches have higher imperceptibility [166] and robustness [330] than spatial approaches but
less embedding capacity [166, 330].
Flexibility was added to DCT embedding by hiding different amounts depending on the values
of the quantized DCT image [15, 17, 90, 201, 264, 274, 305, 454]. Adding a (pseudo)randomness to
the order of embedding was also found effective [75, 264, 295, 392]. Ensuring that hiding starts with
the least significant coefficients, working up to the most significant ones, was found effective [264,
323, 345]. Having variable block sizes or randomly selecting an 8x8 block was found to optimize

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:8 M. H. Kombrink et al.

DCT embedding [305, 346]. Clustering innocent (majority) and guilty (minority) actors together
to determine where to use stego was found effective [181].
Security can be increased in many ways. An example is opting to model the cover and then
adhering to rules from this cover model [67, 68, 118, 325, 326]. Another example is applying Syn-
drome Trellis Coding (STC) to find the optimal embedding solution for each DCT block [381] or,
alternatively, through the use of hypothesis testing [34]. Security can also be increased through
ensuring feature vectors do not change because of embedding [192] or through distortion min-
imization procedures for DCT [151], sometimes with side information [190, 317]. Additionally,
using matrix encoding to hide information in the non-zero coefficients of the quantization table
can increase security [392]. Another example is aiming for the lowest amount of modifications
while keeping flexibility in embedding locations [295]. Another popular approach (Steghide [142])
increases security by swapping DCT coefficients [112]. Finally, security can be increased by em-
bedding one message into multiple images [182].
Optimization of imperceptibility can also be done in several ways. Examples are optimizing the
segmentation process [264] and embedding the message during quantization, not afterwards [272].
Finally, one can postprocess the steganographic image to increase imperceptibility [348].
To optimize security and robustness, [106] opted to embed during compression, where the
rounding of only the most uncertain (close to 0.5) coefficients are altered [106]. Security and em-
bedding capacity were improved by embedding in the difference between two DCT coefficients
[9]. DCT has been paired with LSB [307].
3.2.2 D WT. D WT is a hierarchical decomposition of an image, which is based on small waves
(wavelets) of limited duration and varying frequency. It allows for the detection of the areas that are
most effective for embedding [166]. To create a DWT, horizontally neighbouring pixels’ sum and
difference value are stored; all sums are stored in the left half of the image, while all differences are
stored in the right half [120]. Vertically the same principle is applied. The result is called the first-
order 2-D Haar-DWT [120]. Four sub-bands are created, which can be used to determine where in
the image information could be effectively hidden [12]. Compared to DCT, DWT is more robust
to compression [166] and has higher security [166] but has a lower embedding capacity [90].
Different approaches use different DWT coefficients and differ in how much they hide in each
[56]. Security can be increased by using Huffman encoding—which optimally encodes one symbol
into one word, and one intensity value to one binary value—on the image [263] and/or the secret
message [263, 306]. Different numbers of wavelets have been used; some embed in the first-order
2-D Haar-wavelet [56, 132], while others use more or less than two wavelet transforms [2, 202,
290, 373]. Alternatively, only particular wavelets can be processed multiple times [12]. Impercep-
tibility was increased by using an SVD on each of the sub-bands [88, 132]. Some use the adjacent
DWT coefficient differences or the difference value between blocks to embed in an image [202].
Imperceptibility was also increased by using Diamond Encoding to dictate where to embed [8].
Only embedding in the DWT of skintones in an image was found to increase imperceptibility
[48]. DWT approaches have been paired with DCT [124, 308, 327] and LSB [21, 163].
3.2.3 IWT. IWT approaches use the same transform as DWT approaches. The difference be-
tween the two is that IWT transformation maps integers to integers instead of continuous values
[254]. Due to this discrete method, IWT avoids the floating-point precision problems that DWT
suffers from [166]. When comparing imperceptibility using IWT and DWT, it was shown that IWT
is superior [261, 312]; IWT also shows high security [166].
As for DWT approaches, differences exist in which sub-bands are used. Some use only one
predefined sub-band (mostly HH) [377], and some decide which to use adaptively based on the
length of the message [254]. Imperceptibility was increased by matching blocks of the image to

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:9

the message so that as few pixels as possible are altered [261]. Edge-based IWT embedding was
also tried [1]. Minimisation of the embedding error after embedding, using the Optimum Pixel
Adjustment (OPA) algorithm, was also tried [261]. Security can be improved by using chaotic
maps before embedding [377]. Others used a metaheuristic optimization algorithm to determine
the most efficient pixels to hide information in, which was found to increase security [138]. Alter-
natively, security and imperceptibility can be increased by using a Genetic Algorithm (GA) to
determine which coefficients to alter so that the highest imperceptibility is achieved [292, 316]. Di-
amond Encoding to dictate where to hide in an image [436] was shown to improve imperceptibility
and embedding capacity. IWT has been paired with LSB [91, 377], BPCS [309] and PVD [349].
3.2.4 DFT. DFT is the most complex computation of the transform approaches [92]. In DFT an
image is decomposed into its sine and cosine components [172]—or the phase and magnitude rep-
resentation [136]. The magnitude components are most often used, as spatial shifts do not affect
them [136]. DFT steganography is the least used transform domain approach, due to a round-
ing error that decreases security [135]. Embedding information in the DFT domain gives lesser
imperceptibility [166], security [166] and embedding capacity [166].
The addition of randomization to the embedding location increases imperceptibility [3, 59].
Adding obfuscation was found to increase security [310]. DFT has been combined with DCT [136]

3.3 Deep Learning Approaches

Deep learning approaches use a deep network to hide information. In general, these approaches
are more robust [166] and imperceptible [166]. Currently, the most used DL methods for image
steganography are CNN [33], though GANs are also used. The main advantage of GANs is that
they can generate images, which eliminates the need for a cover image. Other approaches (e.g.,
Monte Carlo Tree Search [255]) are also used but will not be further discussed in this review.
Currently one of the factors that halts further improvement of these schemes is the lack of
realistic datasets. BOSSbase is currently the most commonly used [126], yet this dataset does not
go beyond grayscale images [314] and is created by one steganography scheme, HUGO [280], with
the same embedding rate for all images. Though a newer database (ALASKA) was created, it is also
not optimal as only three embedding schemes were used, all of which are (adaptive) transform
methods [66]. It was ensured that all images have one ‘steganalysis difficulty’ level by altering the
embedding rate [66].
3.3.1 CNN. CNNs are particularly suitable for images. Use of CNNs can be more powerful than
other methods because they do not require expert knowledge on steganography [33]. Another
advantage is that they can be retrained without needing to be redesigned, e.g., for a new tool.
CNNs generally show high security [33].
The major difference between methods lies in the architecture used. Some use vanilla CNNs
[13, 14, 239, 250, 334]; others use residual blocks [399]. Up-convolutional and deconvolutional
functionalities have also been added [84, 435], as well as adversarial embedding [20, 226, 367].
Others opted to increase the reversibility of neural-network-based schemes by employing Invert-
ible Neural Networks (INNs) [232, 409], while exploiting autoencoders has also been researched
[29, 353]. Others use a Neural Style Transfer (NST) network for embedding in the style features
of an image [116]. However, not everyone discloses details of their model architecture [364].
Some of these methods lack in one of the evaluation criteria: robustness [13, 84], embedding
capacity [250], security [13, 364] or imperceptibility [13, 364, 399, 435]. Others triumph in these
evaluation criteria: robustness [29, 239, 250, 353, 364, 399, 409, 435], embedding capacity [14, 84,
232, 239, 353, 409], security [14, 20, 116, 226, 239, 250, 334, 353, 367, 399, 435] or imperceptibility
[84, 116, 232, 250, 353, 409].

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:10 M. H. Kombrink et al.

3.3.2 GAN. GANs combine more than one neural network, where the idea is that networks
have to compete. A generator creates something (a steganographic image), while the discrimina-
tor tries to detect whether it is genuine [123]. For steganography the discriminator network is
given steganographic images and cover images, which it must classify as such [33]. The use of a
discriminator should lead to more imperceptible and secure steganography, as the training process
stops once it can do no better than random guessing [33]. All advantages of CNNs over classical
approaches previously mentioned apply to GANs as well.
Different GAN architectures were tried, where a number of algorithms need a cover image for
embedding [44, 140, 214, 246, 249, 361, 370, 415, 439, 440, 453], while others generate their own
[149, 213, 337, 379, 390, 425, 451, 452]. Changes to the GAN model were made for steganography.
For example, inception modules [440] were used to solve computational complexity and over-
fitting. Alternatively, use of the Regular-Singular Method (RS method) realizes lossless data
embedding through an invertible noise addition and a discrimination function [44]. Others have
adopted a U-Net-based GAN architecture [415]. Another extension is the use of a CycleGAN ar-
chitecture [249], which essentially uses two complete GANs instead of one. Some exploit a Deep
Convolutional GAN (DCGAN) [149], use more than one discriminator [337] or add a noise layer
network to the architecture [453]. Another way the basic GAN structure was improved is through
the use of both residual connections and dense connections [439]. Alternatively, addition of a chan-
nel attention module [361] or of a mutual information mechanism alongside a hierarchical gradient
decay [390] was also tried. Some use a GAN to hide information into contours rather than full im-
ages, to later (through another GAN) create a full steganographic image [451]. Others made use of
existing diffusion models [425] or existing generative models [452] for their embedding schemes.
A difference between methods is their performance on the four criteria. With regards to robust-
ness, some methods performed very well [214, 249, 425, 440, 453]. Some triumphed for the security
criteria [140, 149, 213, 214, 246, 249, 337, 361, 379, 390, 415, 425, 439, 440, 451, 453], while [370] did
not. For the embedding capacity criteria, [44, 361, 390, 439, 452] performed very well as opposed to
[140, 149, 453]. For imperceptibility [44, 249, 337, 361, 390, 453] perform well, while [213] did not.
It should be considered that many DL approaches are not (completely) reversible. One example
[440] created a hiding scheme that did not result in the exact same information. Other methods
were developed that don’t have an extractor algorithm implemented [214, 337]. All of these are
most often not usable to communicate secret information.

3.4 Adaptive Approaches

Adaptive steganography approaches adaptively apply a spatial or transform domain approach.
Adaptability means that the cover image features are used in the embedding process [166]. It
generally takes statistical features into account for embedding [49], where the statistical analysis
dictates where to embed. The method is characterized by selection of which pixels should be used
to embed into, based on the statistical features [157]. Typically a cost function is used to change
each pixel, where the total cost is minimized [328]. Adaptive approaches show good robustness
[49] as well as imperceptibility. Steganalysis struggles with adaptive methods, as they confine the
modifications to the hard-to-model regions of the cover [110], and thus security is generally high.
One difference between adaptive approaches is the method they use after the statistical analysis,
where most methods limit themselves to a domain and not a single method. Examples are spatial
[96, 134, 143, 156, 188, 189, 211, 212, 218, 220, 235, 237, 257, 259, 261, 265, 266, 280, 311, 315, 328,
333, 359, 366, 386, 404, 413, 447] and transform [23, 58, 81, 131, 144, 150, 165, 261, 276, 293, 304, 305,
323, 448]. Some can be applied to both domains: [97, 148, 219, 250].
With regard to the statistical analysis of which pixels to embed into, differences exist. Some use
the difference value between blocks of pixels (2x2, 3x1, etc.) to determine where and how much to

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:11

embed [134, 156, 188, 189, 235, 359, 413]. Adaptive Phase Modulation (APM) [58] hides more or
all information in edges—and thus noisy parts [204, 385]—or uses adaptive sizes for blocks within
an image [304, 305, 416]. Conversion of the secret message to symbols, to then hide this in busier
areas, was found to increase security [447]. Others decide the amounts to embed based on analysis
of blocks or bit-planes [237, 266]. Model-based methods—modelling the cover image to adhere to
its rules for embedding—have been found to increase security [328, 333].
Some well-known algorithms focus on distortion minimization (DM). HUGO [280] does so by
creating a weighted norm between higher-order statistics of pixel differences, with higher weights
given to sparsely populated bins. UNIWARD [148] computes costs in the wavelet domain and
bases it on the relative changes of the wavelet coefficients in the HH subband [148], while WOW
[143] keeps cost low by assessing post-embedding changes in all directions. Other approaches that
focus on (sometimes asymmetric) distortion optimization have also been tried [31, 96, 97, 144, 256].
Though critical review of DM methods has shown them to suffer from security flaws, a solution
was given in [196].
Some [129–131] implemented uniform embedding, which is based on the idea that by uniformly
spreading the embedding alterations to the DCT coefficients, statistical changes can be minimized,
which increases security [131]. Additions to [131, 148] spread the payload between the luminance
and chrominance components to increase security [360]. Using the intrinsic energy of the image to
direct embedding [150] improves security. Colour channel correlation preservation has increased
security also [366, 386]. Imperceptibility was increased by calculating error ratios for different hid-
ing patterns and choosing the lowest error rate’s hiding pattern for embedding [315]. It was shown
that determining less predictable hiding places [211] or clustering modification directions during
embedding enhances security [212]. Alternatively, using the first colour channel and the stego key
to direct embedding was found to be effective [259]. Security was found to increase when divid-
ing a message over several images based on image texture features [219]. Letting the modifying
magnitude be determined by the quality factors of JPEG compression [448] increases security. In-
creasing robustness so that steganographic messages can survive JPEG compression was effective
[449]. Employing the wet paper model to minimize the channel error rate was found to increase
security and massively increase robustness [434]. It was opted to ‘reset’ the last few DCT coeffi-
cients to randomly hide bits in these reset coefficients [323]. Using gray code bit-planes rather than
natural ones increases security [265]. Clustering embedding positions to then smooth the image
increases security [404]. Machine learning was also used to determine embedding positions [165],
amongst which are CNNs [250, 311], GA [23, 24, 168, 253, 276, 293, 294, 331, 383, 388, 426], rein-
forcement learning [256, 365], fuzzy neural networks [81], and Particle Swarm Optimization
(PSO) [81, 218, 257, 261].

4 Steganalysis Methods
Steganalysis is concerned with detection of hidden information (i.e., binary classification), whereas
forensic steganalysis aims to provide further information on the hidden message. Steganalysis has
received significantly less attention from academia than steganography [74]. One of the biggest
challenges within steganalysis is the Cover Source Mismatch (CSM), which says that when a
steganalysis scheme is trained on material from one source, its accuracy will drop on another
[119, 198]. Papers have aimed to detect parts of the image preprocessing pipelines to aid in re-
solving CSM [284]. It is important to know whether the CSM exists when interpreting results of
individual papers.
In order to ensure an overview of different approaches, no separate sections will be used to
discuss forensic steganalysis approaches. This was decided because most forensic steganalysis
approaches are similar to those taken for steganalysis, which would yield doubling of information

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:12 M. H. Kombrink et al.

within the review and provide the reader with less overview of the possible approaches. However,
to allow differentiation between the two research areas, all forensic steganalysis approaches will
also be mentioned separately.
Within this review an overview of the performance of steganalysis methods is given, where
the reported performance in the literature is summarised in tables, which can be found in the
supplementary material. A summary of all results is given in topological sorting graphs.

4.1 Visual Image Inspection

This is the simplest form of steganalysis [99], where one visually inspects the image with the naked
eye [169]. As most new embedding schemes do not result in perceptible quality degradation, visual
inspection has a questionable reliability whilst also being hard to automate [99]. Visual inspection
has gotten less reliable [169]. However, when the parts of the image with no embedded informa-
tion are removed, visual inspection can detect steganographic content [169]. For visual inspection
several features can be used: file size differences, number of unique colours or the distribution of
1s and 0s [169]. However, it only works for the most simple steganography algorithms; moving
from basic LSB to a slightly improved LSB makes visual inspection ineffective [169, 395].

4.2 Signature Detection

In signature-based steganalysis the tool itself is analysed to see whether it would create a specific
signature, which is then looked for in suspected images. An advantage of signature detection
is its simplicity, alongside the promising results for sequentially embedded messages. However,
[271] showed that signature-based steganalysis is hard to automate and that the reliability of
these schemes is often questionable. When a signature is found, it does not guarantee the usage
of steganography nor of a specific tool [18] due to the existence of many false positives. For
signature steganalysis a signature can be two different things.
First, it can be a piece of information that is added by a tool [169]; this is usually easily detected
and provides information on the used tool. Examples include Jpegx, who adds 5B 3B 31 53 00
[139, 169], and Hiderman, who hides the string CDN alongside the hidden information [169]. Other
possible signatures, though less significant because they are naturally more prevalent, are changes
in metadata [5]. Examples include Outguess6 and VSL,7 which always set the QF to 75 and 100,
respectively [5].
Alternatively, a signature can be a pattern within the image that is repeated by a stego tool
during embedding [169, 170]. When found, this pattern can help to detect steganography while
also identifying the tool used [162]. This type of signature differs from the previous type in that
it was not purposefully hidden but occurs due to the degradation or unusual characteristics that
appear during embedding. To find this pattern, it is necessary to create numerous cover–stego
pairs and compare these until a pattern emerges [161]. One example is Hide and Seek, whose
steganographic image palette entries are all divisible by 4 [162, 169], while Gifshuffle produces
images that contain a randomized palette as the signature [99]. YASS [346] has a signature because
the embedding host block location was not completely randomized [210]. A signature specific to
BPCS steganography is present in the complexity histogram where a valley is found [267].
Through the usage of fuzzy rules signatures were found for [324] YASS [346], MB [325] and PQ
[106]. Another signature is the close colour pair (CCP), where after LSB an image holds more
CCP [111]. An alternative is the count of the neighbouring colours rather than the CCP [393]. F5
[392] was shown to add a characteristic header field [18]. Embedding in JPEG images through

6 https://www.provos.org/p/outguess- and- stegdetect- downloads/

7 https://vsl.sourceforge.net/

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:13

spatial methods also leaves a signature as the quantization table will differ from the standard
one [99, 103]. You can also find a signature within the features of the colour spaces of an image
[126]. Alternatively, some have aimed to generalise the search for signatures so that extraction of
signatures was automated [18].

4.3 Targeted Steganalysis

Targeted steganalysis uses probabilistic models to differentiate between steganographic and cover
images [169]. Extensive analysis of methods shows which statistical properties change due to em-
bedding [169]; therefore, one needs an in-depth understanding of the hiding scheme [169]. Thus,
targeted steganalysis is method specific in nature. These method-specific schemes typically result
in accurate detection [271].
4.3.1 LSB. One of the most used approaches for LSB steganalysis is histogram analysis, which
constructs a histogram of the questioned (difference) image [35, 36, 60, 113, 115, 174, 217, 228,
402, 403, 438, 444, 445]. Another approach investigates the changes that occur to colour pairs after
LSB embedding, i.e., two consecutive colours, where one colour is present in two pairs [105, 175,
177, 363]. More general analysis of differences between neighbouring pixels was also implemented
[111, 393, 442]. Alternatively, one can base detection on a Weighted Stego (WS) image with three
components: cover pixel prediction, least-squares weighting, and bias correction [26, 101, 180, 429].
Some have used similarity weights (SWs) between pixels (and channels) for detection [340, 341].
The seventh and eighth bit-planes, where statistical differences occur after embedding, were also
found helpful [178, 179]. Use of higher-order statistics to derive a detection equation improved
results [85]. In [450] a relation was found between the gradient energy forms and the embedded
message’s length. Additionally, combining several separate methods into one can strengthen de-
tection [25]. It was also shown that entropy and joint entropy can be combined for successful LSB
steganalysis [62].
One of the benchmarking methods is RS steganalysis. It divides image pixel groups based on
whether noise increases or decreases after flipping the LSB; the occurrence of both situations can
then be used to detect LSB [102]. RS was improved by using different (sometimes dynamic) masks
[175, 177]. Inspired by RS, [87] found that LSB heavily influences the statistics of sample pairs of
signals [87], which was further improved in [176, 231]. Additionally, cardinality changes within
subsets of pixels can be used for steganalysis [86]. It was shown that embedding in stego and cover
images gives different alteration rates [152].
The search for an optimal feature set for steganalysis having the richest features with the low-
est dimensions—so-called Rich Models (RMs)—has been extensively researched. For RMs some of
the foundational work can be found in [78, 107], though many different feature sets were exploited
[6, 28, 38, 63]. Some feature extraction methods and domains are more popular, e.g., using infor-
mation from residuals [121, 424] or, alternatively, feature extraction from regions that are more
likely to be used for embedding [368, 369]. Some RM methods do not focus on the extraction of
features but on the optimization of the classification afterwards [208, 234]. One alternative to RM
is Model-Based (MB) detection, where cover images are modelled to test the suspected images
against this model [69, 70, 72, 95].
Some have opted to use machine learning LSB steganalysis. These methods use different ar-
chitectures and types of machine learning, amongst which are vanilla neural networks [252] and
CNNs with different specifications [7, 79, 298–300, 313, 374, 391, 400, 401, 405, 408, 414, 417–419,
427, 441], dynamic evolving neural fuzzy inference system (DENFIS) [229], support vector
machine recursive feature elimination [229], a Siamese CNN [419], residual blocks [405], federated
learning [414] and ensemble classifiers [108, 407].

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:14 M. H. Kombrink et al.

Forensic LSB steganalysis has been applied by [26, 85, 101, 102, 113, 176, 180, 231, 363, 427, 429,
444, 445, 450]. Additional information in the form of message length (or embedding rate) [26, 85,
101, 102, 113, 176, 180, 231, 429, 444, 445, 450], pixel-level detection [427] and threshold used for
embedding [363] has been investigated.
4.3.2 PVD. Most PVD steganalysis methods are histogram-based methods, where the specific
histogram each one uses differs [215, 431]. Inspired by this, a scatter plot based on theoretical
and observed frequencies of the DV between pixel pairs was used [223] to show correlations. WS
image has been applied as well [437].
Within PVD steganalysis, additional information has also been gathered from the images that
are analysed. Forensic PVD steganalysis in the form of calculating an embedded message length
(or embedding rate) has been applied by [431, 437].
4.3.3 BPCS. For BPCS a wavelet characteristic function has been implemented [230]. An al-
ternative effective feature for steganalysis is the bit-plane-block complexity difference sequence
[362].
4.3.4 DCT. Stegdetect [296], a commercial tool, can detect DCT embedding by exploiting a
version of Pairs of Values (POVs) [395]. Similarly, histogram-based methods [104, 160, 205, 206,
430, 443] and the adjacency matrix of DCT coefficients can be used for steganalysis [199]. Model-
based methods, which opt to model the cover medium used to analyse differences between this
model and a suspected image, have been applied [5, 30, 301, 302, 371, 455]. [394] adapted several
higher-order statistical LSB steganalysis methods [87, 101, 102, 105] to the DCT domain.
RMs have also been applied to DCT steganalysis [4, 147, 193, 285, 288, 347]. Many RM methods
extract the features through use of a Markov process [51, 114, 282, 339], while others derive them
(amongst others) from histograms [52, 77], the wavelet domain [241, 242, 456] or the cosine domain
[98, 146, 195, 227, 245, 281]. Some RM-based methods do not focus on feature extraction but on the
optimization of the classification method afterwards [128, 183, 197, 286]. Additionally, determining
the most dominant features to reduce dimensionality has also been applied [243].
Recently, machine learning has been applied to DCT steganalysis. These methods include CNNs
with different specifications [55, 153, 406, 423, 432], applying transfer learning to existing archi-
tectures [420, 422], different detectors for different quality factors [421] and PSO [336].
Forensic DCT steganalysis has been applied by [5, 104, 199, 281, 371, 394, 430, 443]. Additional
information in the form of message length (or embedding rate) [104, 371, 394, 430, 443], which
embedding method was used [5, 281] and the change rate [199] has been investigated.
4.3.5 Other Approach-specific Steganalysis Methods. No methods were found that are targeted
only at DWT-, DFT-, IWT-, CNN- or GAN-based steganography methods.

4.4 Universal Steganalysis

Within this review, a universal (or blind) approach will be categorized as such when it is able to
detect schemes from at least two different domains from spatial, transform and deep learning. A
method that detects, for example, different DWT steganographic schemes has been categorized
as targeted. Note that this sometimes differs from other sources’ classification of universal ste-
ganalysis approaches. Consequently, some research that was titled towards universal steganalysis
is regarded as targeted for this review.
One of the first universal steganalysis approaches [395] evolved around POVs, which are pixel
values, quantized DCT coefficients or palette indices that get mapped to each other through LSB
flipping. This distribution can then be used for steganalysis [395]. A more general analysis of the
differences between neighbouring pixels has also been found effective [127, 207]. Since then, many

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:15

other approaches have been implemented. Histogram analysis was found useful [137], and the
seventh and eighth bit-planes, where statistical differences occur between stego and cover images,
have also been exploited [10]. Using compressive sensing to distinguish stego from cover has also
been implemented [277]. Fusing several individual methods has shown its merit as well [185].
RMs are the most researched universal detection scheme [11, 37, 125, 145, 194, 221, 384, 411],
where some used (amongst others) features from histogram analysis [61, 83, 209], from wavelet
domain [93, 122, 236, 240, 338, 410] and from Markov chains [141, 332, 355]. Analysis of the number
of changes to a file based on extracted features was successful [287].
Recently machine learning has been applied to universal steganalysis. Different architectures
were implemented: regression trees [378], CNNs [389] and a CNN with residual blocks [27, 344],
which was further investigated in [32].
Some research has not focused on the construction of new methods but on further testing of
existing methods [175, 184, 186].
Forensic universal steganalysis has been applied by few researchers [287, 344, 378]; additional
information was looked for in the form of number of embedding changes [287, 378] and detecting
the embedding method that was used [344].

4.5 Summary of Steganalysis Performance per Steganography Method

In this section, topological sorting diagrams are shown to depict the relative performance of dif-
ferent steganalysis approaches previously discussed on some of the most-targeted steganography
methods. The performance of all methods is taken from their publication; tables showing the re-
porting of these results in the literature can be found in the supplementary materials. It is impor-
tant to realize that these diagrams do not depict numerical differences but show a hierarchy based
on reported results. That is, when one method is shown to outperform another, it will be placed
above it, but the distance between two methods does not depict a numerical difference between
their performance. Caution should always be taken towards these results, as ordering is purely
based on the results reported. This means that not all results are formulated based on the same
dataset. Most studies created their own steganography dataset and then reported results on it. This
means that reported results can vary for different datasets, which has the possibility to alter the
topological sorting diagrams. Still the diagrams give a good impression of relative performance of
different steganalysis schemes.
Within all topological sorting depictions in this section, three different types of relationships are
distinguished: method A can outperform method B, method A can be comparable to method B and
method A can be outperformed by (underperform) method B. Interesting to note is the difference
in the way results are reported, where in some cases one method claims to be comparable to
several others, even though these others were found to outperform or underperform each other.
This would suggest that being comparable to all of them would not be possible. This could be due
to differences in datasets but also due to the interpretability of ‘comparable’ and ‘outperforms’.
Some researchers call an 0.05% accuracy increase ‘outperforms’, while others will call this a
‘comparable’ result.
The spatial domain steganography methods that have the most reported results (HILL,
S-UNIWARD, HUGO and WOW) are shown in Figure 3. The transform-based steganography meth-
ods with the most reported results (JSteg,8 JPHide&Seek,9 Outguess, nsF5, MB, F5, J-UNIWARD,
StegHide and UED) are shown in Figures 4 and 5.
A few notable observations can be made from the different topological sorting graphs. First of
all, the references that mention their performance (and sometimes that of others) differ greatly per
8 https://packetstormsecurity.com/files/21618/jsteg.txt.html
9 http://linux01.gwdg.de/∼alatham/stego.html

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:16 M. H. Kombrink et al.

Fig. 3. Figure depicting the relative performance of different methods on four adaptive spatial domain
steganography methods: HILL [211], S-UNIWARD [148], HUGO [280] and WOW [143]. Black arrows with
an ‘o’ in them signify a method that outperforms another, a green ‘c’ signifies one method was found to be
comparable to another, and a red ‘u’ signifies that one method is outperformed by (underperforms) another.
If a method is not compared to others by the authors themselves, the reference will have no outgoing arrows.

method, as does their place in the ranking (which is rarely the same). From this, we can conclude
that no steganalysis method exists that is applicable to the majority of the steganography methods.
Another observation is the number of methods that do not compare their performance with any
other (in the figures these have no outgoing arrows). Therefore, these methods cannot be compared
and their relative performance is unknown. Thus, it is strongly recommended for future research
to always compare new schemes with previous ones. By doing so, relative performance can be
more readily judged, which increases the potential usefulness of a new publication.
It is very noticeable that for some steganography scheme papers, they compare their perfor-
mance to many other steganalysis schemes, which results in a form of ranking when observing the
topological sorting diagrams. However, for some schemes (e.g., StegHide or JPHide), comparison

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:17

Fig. 4. Figure depicting the relative performance of different methods on four transform domain approaches:
J-UNIWARD [148], UED [130], StegHide [142] and nsF5 [112], with the same interpretation of arrows and
colours as Figure 3. If a method is shown twice in one ordering, it denotes that the method’s relation differs
depending on the circumstances (e.g., outperforms for low embedding rate but is outperformed on high
embedding rate).

in papers is more often one on one, with no later additional comparisons. This makes comparison
of relative performance difficult and results in topological sorting diagrams that are not very
informative.
Lastly, it is interesting to note that no contradictions within the orderings were found. That is, no
method specified itself to outperform another was later found to not outperform that method at all.
This indicates there was no difference in relative performance of the methods, despite differences
in settings. This strengthens the results depicted in the topological sorting diagrams.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:18 M. H. Kombrink et al.

Fig. 5. Figure depicting the relative performance of different methods on four transform domain approaches:
JPHide, MB [325], Outguess, F5 [392] and Jsteg, with same interpretation of arrows and colours as Figure 4.

5 Discussion and Practical Implications

Steganography has been researched much more widely than its counterpart, steganalysis. The
most popular steganography methods are, as can be seen from the topological sorting graphs,
mostly (adaptive) transform domain methods. This is likely because these methods score a lot

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:19

better on the four evaluation criteria for steganography than older and more basic schemes like
LSBR and LSBM. However, newer steganography methods, especially the deep-learning-based
methods, show greater potential. At the same time, these newer methods suffer from one great
downside compared to the more often used methods: the deep-learning-based methods are often
non-reversible. This ensures that despite the good performance of these methods, they are in their
current form not at all usable in practice (if no one can extract the hidden information, there is no
use for the steganographic method). Therefore, steganography research should focus on enabling
the practical use of their developed schemes by ensuring reversibility at all times. Without this re-
versibility, the potential that new schemes have cannot be fully used, thus stopping these schemes
from finding practical applications.
For steganalysis most effort has been put into the construction of Rich Models. Though these
models have achieved good performance on the steganography scheme they were designed for,
they cannot easily generalise to more methods (as can be seen in Figures 3 through 5). Machine-
learning-based steganalysis has the potential to overcome this generalisation problem; however,
more research needs to be conducted for it to reach this potential. This future research area could
take steganalysis from a mainly academic study area to it reaching practical applications, for ex-
ample, within law enforcement agencies.
For practical applications of steganalysis it is of importance to focus on two separate goals.
First, universal detection methods that truly detect the presence of most steganography methods
are needed. Currently, steganalysis methods are shown to not generalise to more methods, as can
be seen from Figures 3 through 5, where no single method comes out on top in most of the topo-
logical sorting graphs. Second, forensic steganalysis should move further towards the extraction
of a message than is currently aimed for, as the mere detection of the presence of steganography
does not answer most of the questions a practical application of steganalysis would ask. The im-
plications of not taking these next steps are that steganalysis will continue to lack connection to
the institutions and people that would be able to use breakthroughs in steganalysis in their work
and practically apply this research. When these next steps are taken, it can greatly improve the
practical usability of (forensic) steganalysis (e.g., within law enforcement).
With this survey a clear basis for future research into steganography and steganalysis is
provided.

References
[1] Hayat Al-Dmour and Ahmed Al-Ani. 2016. A steganography embedding method based on edge identification and
XOR coding. Expert Systems with Applications 46 (2016), 293–306. DOI:https://doi.org/10.1016/j.eswa.2015.10.024
[2] H. TS. ALRikabi and Hussein Tuama Hazim. 2021. Enhanced data security of communication system using combined
encryption and steganography. International Journal of Interactive Mobile Technologies (iJIM) 15, 16 (2021), 145.
[3] F. Alturki and R. Mersereau. 2001. Secure blind image steganographic technique using discrete Fourier transfor-
mation. In Proceedings 2001 International Conference on Image Processing (Cat. No. 01CH37205), Vol. 2. 542–545.
DOI:https://doi.org/10.1109/ICIP.2001.958548
[4] Hossein Amirkhani and Mohammad Rahmati. 2011. New framework for using image contents in blind steganalysis
systems. Journal of Electronic Imaging 20, 1 (2011), 013016. DOI:https://doi.org/10.1117/1.3554413
[5] Panagiotis Andriotis, George Oikonomou, and Theo Tryfonas. 2013. JPEG steganography detection with Benford’s
Law. Digital Investigation 9, 3–4 (2013), 246–257.
[6] S. Arivazhagan, E. Amrutha, and W. Sylvia Lilly Jebarani. 2022. Universal steganalysis of spatial content-independent
and content-adaptive steganographic algorithms using normalized feature derived from empirical mode
decomposed components. Signal Processing: Image Communication 101 (2022), 116567. DOI:https://doi.org/10.1016/
j.image.2021.116567
[7] S. Arivazhagan, E. Amrutha, W. Sylvia Lilly Jebarani, and S. T. Veena. 2021. Hybrid convolutional neural network
architecture driven by residual features for steganalysis of spatial steganographic algorithms. Neural Computing and
Applications 33, 17 (2021), 11465–11485.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:20 M. H. Kombrink et al.

[8] Samer Atawneh, Ammar Almomani, Hussein Al Bazar, Putra Sumari, and Brij Gupta. 2017. Secure and impercep-
tible digital image steganographic algorithm based on diamond encoding in DWT domain. Multimedia Tools and
Applications 76, 18 (2017), 18451–18472.
[9] Abdelhamid Awad Attaby, Mona F. M. Mursi Ahmed, and Abdelwahab K. Alsammak. 2018. Data hiding inside JPEG
images with high resistance to steganalysis using a novel technique: DCT-M3. Ain Shams Engineering Journal 9,
4 (2018), 1965–1974. DOI:https://doi.org/10.1016/j.asej.2017.02.003
[10] İsmail Avcıbaş, Mehdi Kharrazi, Nasir Memon, and Bülent Sankur. 2005. Image steganalysis with binary similarity
measures. EURASIP Journal on Advances in Signal Processing 2005, 17 (2005), 1–9.
[11] I. Avcibas, N. Memon, and B. Sankur. 2003. Steganalysis using image quality metrics. IEEE Transactions on Image
Processing 12, 2 (2003), 221–229. DOI:https://doi.org/10.1109/TIP.2002.807363
[12] Della Baby, Jitha Thomas, Gisny Augustine, Elsa George, and Neenu Rosia Michael. 2015. A novel DWT based image
securing method using steganography. International Conference on Advances in Information and Communication
Technologies 46 (2015), 612–618. DOI:https://doi.org/10.1016/j.procs.2015.02.105
[13] Shumeet Baluja. 2017. Hiding images in plain sight: Deep steganography. Advances in Neural Information Processing
Systems 30 (2017), 2069–2079.
[14] Shumeet Baluja. 2020. Hiding images within images. IEEE Transactions on Pattern Analysis and Machine Intelligence
42, 7 (2020), 1685–1697. DOI:https://doi.org/10.1109/TPAMI.2019.2901877
[15] Deepika Bansal and Rita Chhikara. 2014. An improved DCT based steganography technique. International Journal
of Computer Applications 102, 14 (2014), 46–49.
[16] Patrick Bas, Tomáš Filler, and Tomáš Pevnỳ. 2011. “Break our steganographic system”: the ins and outs of organizing
BOSS. In International workshop on information hiding. Springer, 59–70.
[17] Mohammed Baziyad, Tamer Rabie, Ibrahim Kamel, and Mahdi Benkhelifa. 2022. Polynomial fitting: Enhancing the
stego quality of DCT-based Steganography schemes. Multimedia Tools and Applications 81 (2022), 1–21.
[18] Graeme Bell and Yeuan-Kuen Lee. 2010. A method for automatic identification of signatures of steganography soft-
ware. IEEE Transactions on Information Forensics and Security 5, 2 (2010), 354–358. DOI:https://doi.org/10.1109/TIFS.
2010.2046985
[19] W. Bender, D. Gruhl, N. Morimoto, and A. Lu. 1996. Techniques for data hiding. IBM Systems Journal 35, 3.4 (1996),
313–336. DOI:https://doi.org/10.1147/sj.353.0313
[20] Solène Bernard, Patrick Bas, John Klein, and Tomáš Pevný. 2022. Backpack: A backpropagable adversarial embedding
scheme. IEEE Transactions on Information Forensics and Security 17 (2022), 3539–3554. DOI:https://doi.org/10.1109/
TIFS.2022.3204218
[21] Swati Bhargava and Manish Mukhija. 2019. Hide image and text using LSB, DWT and RSA based on image steganog-
raphy. ICTACT Journal on Image & Video Processing 9, 3 (2019), 1940–1946.
[22] Souvik Bhattacharyya. 2011. A survey of steganography and steganalysis technique in image, text, audio and video
as cover carrier. Journal of Global Research in Computer Science 2, 4 (2011), 1–16.
[23] Rajib Biswas and Samir Kumar Bandyapadhay. 2019. Random selection based GA optimization in 2D-DCT domain
color image steganography. Multimedia Tools and Applications 79 (2019), 1–20.
[24] Rajib Biswas and Samir Kumar Bandyapadhay. 2020. Random selection based GA optimization in 2D-DCT domain
color image steganography. Multimedia Tools and Applications 79, 11 (2020), 7101–7120.
[25] Benedikt Boehm. 2014. StegExpose—A Tool for Detecting LSB Steganography. (2014). DOI:https://doi.org/10.48550/
ARXIV.1410.6656
[26] Rainer Böhme. 2008. Weighted stego-image steganalysis for JPEG covers. In International Workshop on Information
Hiding, Kaushal Solanki, Kenneth Sullivan, and Upamanyu Madhow (Eds.). Springer, Berlin, 178–194.
[27] Mehdi Boroumand, Mo Chen, and Jessica Fridrich. 2019. Deep residual network for steganalysis of digital images.
IEEE Transactions on Information Forensics and Security 14, 5 (2019), 1181–1193. DOI:https://doi.org/10.1109/TIFS.
2018.2871749
[28] Mehdi Boroumand and Jessica Fridrich. 2017. Nonlinear feature normalization in steganalysis. In Proceedings of
the 5th ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’17). Association for Computing
Machinery, New York, NY, USA, 45–54. DOI:https://doi.org/10.1145/3082031.3083239
[29] Tu Bui, Shruti Agarwal, Ning Yu, and John Collomosse. 2023. RoSteALS: Robust steganography using autoencoder
latent space. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops.
933–942.
[30] Jan Butora and Jessica Fridrich. 2020. Reverse JPEG compatibility attack. IEEE Transactions on Information Forensics
and Security 15 (2020), 1444–1454. DOI:https://doi.org/10.1109/TIFS.2019.2940904
[31] Jan Butora and Jessica Fridrich. 2020. Steganography and its detection in JPEG images obtained with the “trunc”
quantizer. In 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP’20). IEEE, 2762–
2766.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:21

[32] Jan Butora, Yassine Yousfi, and Jessica Fridrich. 2021. How to pretrain for steganalysis. In Proceedings of the 2021 ACM
Workshop on Information Hiding and Multimedia Security (IH&MMSec’21). Association for Computing Machinery,
New York, NY, USA, 143–148. DOI:https://doi.org/10.1145/3437880.3460395
[33] Olivia Byrnes, Wendy La, Hu Wang, Congbo Ma, Minhui Xue, and Qi Wu. 2021. Data hiding with deep learning:
A survey unifying digital watermarking and steganography. CoRR abs/2107.09287 (2021). arXiv:2107.09287 https:
//arxiv.org/abs/2107.09287
[34] Christian Cachin. 1998. An information-theoretic model for Steganography. In International Workshop on Informa-
tion Hiding, David Aucsmith (Ed.). Springer, Berlin, 306–318.
[35] Kaiwei Cai, Xiaolong Li, Tieyong Zeng, Bin Yang, and Xiaoqing Lu. 2010. Reliable histogram features for detecting
LSB matching. In 2010 IEEE International Conference on Image Processing. 1761–1764. DOI:https://doi.org/10.1109/
ICIP.2010.5651567
[36] Giacomo Cancelli, Gwenael Doerr, Ingemar J. Cox, and Mauro Barni. 2008. Detection of ±1 LSB steganography
based on the amplitude of histogram local extrema. In 2008 15th IEEE International Conference on Image Processing.
1288–1291. DOI:https://doi.org/10.1109/ICIP.2008.4711998
[37] Ji cang Lu, Fen lin Liu, and Xiang yang Luo. 2014. Selection of image features for steganalysis based on the Fisher
criterion. Digital Investigation 11, 1 (2014), 57–66. DOI:https://doi.org/10.1016/j.diin.2013.12.001
[38] Mehmet U. Celik, Gaurav Sharma, and A. Murat Tekalp. 2004. Universal image steganalysis using rate-distortion
curves. In Security, Steganography, and Watermarking of Multimedia Contents VI, Edward J. Delp III and Ping W.
Wong (Eds.), Vol. 5306. International Society for Optics and Photonics, SPIE, 467–476. DOI:https://doi.org/10.1117/
12.531359
[39] Soumendu Chakraborty and Anand Singh Jalal. 2020. A novel local binary pattern based blind feature image
steganography. Multimedia Tools and Applications 79, 27 (2020), 19561–19574.
[40] Chi-Kwong Chan and Lee-Ming Cheng. 2004. Hiding data in images by simple LSB substitution. Pattern Recognition
37, 3 (2004), 469–474.
[41] Rajarathnam Chandramouli, Mehdi Kharrazi, and Nasir Memon. 2004. Image steganography and steganalysis: Con-
cepts and practice. In Digital Watermarking, Ton Kalker, Ingemar Cox, and Yong Man Ro (Eds.). Springer, Berlin,
35–49.
[42] R. Chandramouli and N. Memon. 2001. Analysis of LSB based image steganography techniques. In Proceedings 2001
International Conference on Image Processing (Cat. No. 01CH37205), Vol. 3. 1019–1022. DOI:https://doi.org/10.1109/
ICIP.2001.958299
[43] Rajarathnam Chandramouli and Nasir D. Memon. 2003. Steganography capacity: A steganalysis perspective. In
Security and Watermarking of Multimedia Contents V, Edward J. Delp III and Ping Wah Wong (Eds.), Vol. 5020.
International Society for Optics and Photonics, SPIE, 173–177. DOI:https://doi.org/10.1117/12.479732
[44] Ching-Chun Chang. 2020. Adversarial learning for invertible steganography. IEEE Access 8 (2020), 198425–198435.
DOI:https://doi.org/10.1109/ACCESS.2020.3034936
[45] Chin-Chen Chang, Chun-Sen Tseng, and Chia-Chen Lin. 2005. Hiding data in binary images. In Information Security
Practice and Experience, Robert H. Deng, Feng Bao, HweeHwa Pang, and Jianying Zhou (Eds.). Springer, Berlin,
338–349.
[46] Chin-Chen Chang and Hsien-Wen Tseng. 2004. A steganographic method for digital images using side match. Pattern
Recognition Letters 25, 12 (2004), 1431–1437. DOI:https://doi.org/10.1016/j.patrec.2004.05.006
[47] Ko-Chin Chang, Chien-Ping Chang, Ping S. Huang, and Te-Ming Tu. 2008. A novel image steganographic method
using tri-way pixel-value differencing. Journal of Multimedia 3, 2 (2008), 37–44.
[48] Abbas Cheddad, Joan Condell, Kevin Curran, and Paul Mc Kevitt. 2008. Biometric inspired digital image steganog-
raphy. In 15th Annual IEEE International Conference and Workshop on the Engineering of Computer Based Systems
(ECBS’08). 159–168. DOI:https://doi.org/10.1109/ECBS.2008.11
[49] Abbas Cheddad, Joan Condell, Kevin Curran, and Paul McKevitt. 2010. Digital image steganography: Survey and
analysis of current methods. Signal Processing 90, 3 (2010), 727–752.
[50] B. Chen and G. Wornell. 1998. Digital watermarking and information embedding using dither modulation. In 1998
IEEE Second Workshop on Multimedia Signal Processing (Cat. No. 98EX175). 273–278. https://doi.org/10.1109/MMSP.
1998.738946
[51] Chunhua Chen and Yun Q. Shi. 2008. JPEG image steganalysis utilizing both intrablock and interblock correlations.
In 2008 IEEE International Symposium on Circuits and Systems (ISCAS’08). 3029–3032. DOI:https://doi.org/10.1109/
ISCAS.2008.4542096
[52] Chunhua Chen, Yun Q. Shi, Wen Chen, and Guorong Xuan. 2006. Statistical moments based universal steganalysis
using JPEG 2-D array and 2-D characteristic function. In 2006 International Conference on Image Processing. 105–108.
DOI:https://doi.org/10.1109/ICIP.2006.312383

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:22 M. H. Kombrink et al.

[53] Jeanne Chen. 2014. A PVD-based data hiding method with histogram preserving using pixel pair matching. Signal
Processing: Image Communication 29, 3 (2014), 375–384. DOI:https://doi.org/10.1016/j.image.2014.01.003
[54] Jeanne Chen, Tung-Shou Chen, and Meng-Wen Cheng. 2003. A new data hiding method in binary image. In Pro-
ceedings of the 5th International Symposium on Multimedia Software Engineering, 2003. 88–93. DOI:https://doi.org/
10.1109/MMSE.2003.1254426
[55] Mo Chen, Vahid Sedighi, Mehdi Boroumand, and Jessica Fridrich. 2017. JPEG-phase-aware convolutional neural
network for steganalysis of JPEG images. In Proceedings of the 5th ACM Workshop on Information Hiding and Mul-
timedia Security (IH&MMSec’17). Association for Computing Machinery, New York, NY, USA, 75–84. DOI:https:
//doi.org/10.1145/3082031.3083248
[56] Po-Yueh Chen and Hung-Ju Lin. 2006. A DWT based approach for image steganography. International Journal of
Applied Science and Engineering 4, 3 (2006), 275–290.
[57] Wen-Jan Chen, Chin-Chen Chang, and T. Hoang Ngan Le. 2010. High payload steganography mechanism using
hybrid edge detector. Expert Systems with Applications 37, 4 (2010), 3292–3301. DOI:https://doi.org/10.1016/j.eswa.
2009.09.050
[58] Wen-Yuan Chen. 2007. Color image steganography scheme using set partitioning in hierarchical trees coding, digital
Fourier transform and adaptive phase modulation. Applied Mathematics and Computation 185, 1 (2007), 432–448.
DOI:https://doi.org/10.1016/j.amc.2006.07.041
[59] Wen-Yuan Chen. 2008. Color image steganography scheme using DFT, SPIHT codec, and modified differential phase-
shift keying techniques. Applied Mathematics and Computation 196, 1 (2008), 40–54. DOI:https://doi.org/10.1016/j.
amc.2007.05.063
[60] Xianyi Chen, Guangyong Gao, Dandan Liu, and Zhihua Xia. 2016. Steganalysis of LSB matching using characteristic
function moment of pixel differences. China Communications 13, 7 (2016), 66–73. DOI:https://doi.org/10.1109/CC.
2016.7559077
[61] Xiaochuan Chen, Yunhong Wang, Tieniu Tan, and Lei Guo. 2006. Blind image steganalysis based on statistical
analysis of empirical matrix. In 18th International Conference on Pattern Recognition (ICPR’06), Vol. 3. 1107–1110.
DOI:https://doi.org/10.1109/ICPR.2006.332
[62] Sonam Chhikara and Rajeev Kumar. 2020. An information theoretic image steganalysis for LSB steganography. Acta
Cybernetica 24, 4 (Jul. 2020), 593–612. DOI:https://doi.org/10.14232/actacyb.279174
[63] Kang Leng Chiew and Josef Pieprzyk. 2010. Blind steganalysis: A countermeasure for binary image steganography.
In 2010 International Conference on Availability, Reliability and Security. 653–658. DOI:https://doi.org/10.1109/ARES.
2010.66
[64] J. Anita Christaline, R. Ramesh, and D. Vaishali. 2015. Critical review of image steganalysis techniques. International
Journal of Advanced Intelligence Paradigms 7, 3–4 (2015), 368–381.
[65] Shaveta Chutani and Anjali Goyal. 2019. A review of forensic approaches to digital image Steganalysis. Multimedia
Tools and Applications 78, 13 (2019), 18169–18204.
[66] Rémi Cogranne, Quentin Giboulot, and Patrick Bas. 2019. The ALASKA steganalysis challenge: A first step towards
steganalysis. In Proceedings of the ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’19).
Association for Computing Machinery, New York, NY, USA, 125–137. DOI:https://doi.org/10.1145/3335203.
3335726
[67] Rémi Cogranne, Quentin Giboulot, and Patrick Bas. 2020. Steganography by minimizing statistical detectability: The
cases of JPEG and color images. In Proceedings of the 2020 ACM Workshop on Information Hiding and Multimedia
Security (IH&MMSec’20). Association for Computing Machinery, New York, NY, USA, 161–167. DOI:https://doi.org/
10.1145/3369412.3395075
[68] Rémi Cogranne, Quentin Giboulot, and Patrick Bas. 2022. Efficient steganography in JPEG images by minimiz-
ing performance of optimal detector. IEEE Transactions on Information Forensics and Security 17 (2022), 1328–1343.
DOI:https://doi.org/10.1109/TIFS.2021.3111713
[69] Rémi Cogranne and Florent Retraint. 2013. An asymptotically uniformly most powerful test for LSB matching de-
tection. IEEE Transactions on Information Forensics and Security 8, 3 (2013), 464–476. DOI:https://doi.org/10.1109/
TIFS.2013.2238232
[70] Rémi Cogranne, Cathel Zitzmann, Lionel Fillatre, Florent Retraint, Igor Nikiforov, and Philippe Cornu. 2011. Statisti-
cal decision by using quantized observations. In Proceedings of the 2011 IEEE International Symposium on Information
Theory. 1210–1214. DOI:https://doi.org/10.1109/ISIT.2011.6033727
[71] Rémi Cogranne, Cathel Zitzmann, Florent Retraint, Igor Nikiforov, Lionel Fillatre, and Philippe Cornu. 2013. Statis-
tical Detection of LSB Matching Using Hypothesis Testing Theory. In International Workshop on Information Hiding,
Matthias Kirchner and Dipak Ghosal (Eds.). Springer, Berlin, Heidelberg, 46–62.
[72] Rémi Cogranne, Cathel Zitzmann, Florent Retraint, Igor V. Nikiforov, Philippe Cornu, and Lionel Fillatre. 2014. A
local adaptive model of natural images for almost optimal detection of hidden data. Signal Processing 100 (2014),
169–185. DOI:https://doi.org/10.1016/j.sigpro.2014.01.027

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:23

[73] I. Cox, J. Kilian, F. Leighton, and T. Shamoon. 1997. Secure spread spectrum watermarking for multimedia. IEEE
Transactions on Image Processing 6, 12 (1997), 1673–1687. https://doi.org/10.1109/83.650120
[74] Mukesh Dalal and Mamta Juneja. 2021. Steganography and steganalysis (in digital forensics): A cybersecurity guide.
Multimedia Tools and Applications 80, 4 (2021), 5723–5771.
[75] Ajit Danti and Preethi Acharya. 2010. Randomized embedding scheme based on DCT coefficients for image
steganography. International Journal of Computer Applications 2010 (2010), 97–103.
[76] D. Darwis, N. B. Pamungkas, and Wamiliana. 2021. Comparison of least significant bit, pixel value differencing, and
modulus function on Steganography to measure image quality, storage capacity, and robustness. Journal of Physics:
Conference Series 1751, 1 (Jan. 2021), 012039. DOI:https://doi.org/10.1088/1742-6596/1751/1/012039
[77] Tomáš Denemark, Mehdi Boroumand, and Jessica Fridrich. 2016. Steganalysis features for content-adaptive JPEG
Steganography. IEEE Transactions on Information Forensics and Security 11, 8 (2016), 1736–1746. DOI:https://doi.org/
10.1109/TIFS.2016.2555281
[78] Tomas Denemark, Vahid Sedighi, Vojtech Holub, Remi Cogranne, and Jessica Fridrich. 2014. Selection-channel-
aware rich model for steganalysis of digital images. In 2014 IEEE International Workshop on Information Forensics
and Security (WIFS’14). 48–53. DOI:https://doi.org/10.1109/WIFS.2014.7084302
[79] Xiaoqing Deng, Bolin Chen, Weiqi Luo, and Da Luo. 2019. Fast and effective global covariance pooling net-
work for image steganalysis. In Proceedings of the ACM Workshop on Information Hiding and Multimedia Security
(IH&MMSec’19). Association for Computing Machinery, New York, NY, USA, 230–234. DOI:https://doi.org/10.1145/
3335203.3335739
[80] Madhavi B. Desai and S. V. Patel. 2014. Survey on universal image steganalysis. International Journal of Computer
Science and Information Technologies 5, 3 (2014), 4752–4759.
[81] Sachin Dhawan, Chinmay Chakraborty, Jaroslav Frnda, Rashmi Gupta, Arun Kumar Rana, and Subhendu Kumar
Pani. 2021. SSII: Secured and high-quality steganography using intelligent hybrid optimization algorithms for IoT.
IEEE Access 9 (2021), 87563–87578. DOI:https://doi.org/10.1109/ACCESS.2021.3089357
[82] Sachin Dhawan and Rashmi Gupta. 2021. Analysis of various data security techniques of steganography: A survey.
Information Security Journal: A Global Perspective 30, 2 (2021), 63–87.
[83] Jing Dong and Tieniu Tan. 2008. Blind image steganalysis based on run-length histogram analysis. In 2008 15th IEEE
International Conference on Image Processing. 2064–2067. DOI:https://doi.org/10.1109/ICIP.2008.4712192
[84] Xintao Duan, Kai Jia, Baoxia Li, Daidou Guo, En Zhang, and Chuan Qin. 2019. Reversible image steganography
scheme based on a U-Net structure. IEEE Access 7 (2019), 9314–9323. DOI:https://doi.org/10.1109/ACCESS.2019.
2891247
[85] S. Dumitrescu and Xiaolin Wu. 2005. A new framework of LSB steganalysis of digital media. IEEE Transactions on
Signal Processing 53, 10 (2005), 3936–3947. DOI:https://doi.org/10.1109/TSP.2005.855078
[86] S. Dumitrescu, Xiaolin Wu, and N. Memon. 2002. On steganalysis of random LSB embedding in continuous-tone
images. In Proceedings of the International Conference on Image Processing, Vol. 3. 641–644. DOI:https://doi.org/10.
1109/ICIP.2002.1039052
[87] Sorina Dumitrescu, Xiaolin Wu, and Zhe Wang. 2003. Detection of LSB steganography via sample pair analysis. In
International Workshop on Information Hiding, Fabien A. P. Petitcolas (Ed.). Springer, Berlin, 355–372.
[88] Asha Durafe and Vinod Patidar. 2022. Development and analysis of IWT-SVD and DWT-SVD steganography using
fractal cover. Journal of King Saud University, Computer and Information Sciences 34, 7 (2022), 4483–4498. DOI:https:
//doi.org/10.1016/j.jksuci.2020.10.008
[89] Wafa M. Eid, Sarah S. Alotaibi, Hasna M. Alqahtani, and Sahar Q. Saleh. 2022. Digital image steganalysis: Cur-
rent methodologies and future challenges. IEEE Access 10 (2022), 92321–92336. DOI:https://doi.org/10.1109/ACCESS.
2022.3202905
[90] Sahar A. El_Rahman. 2018. A comparative analysis of image steganography based on DCT algorithm and steganog-
raphy tool to hide nuclear reactors confidential information. Computers & Electrical Engineering 70 (2018), 380–399.
DOI:https://doi.org/10.1016/j.compeleceng.2016.09.001
[91] Elshazly Emad, Abdelwahab Safey, Abouzaid Refaat, Zahran Osama, Elaraby Sayed, and Elkordy Mohamed. 2018.
A secure image steganography algorithm based on least significant bit and integer wavelet transform. Journal of
Systems Engineering and Electronics 29, 3 (2018), 639–649.
[92] Oleg Evsutin, Anna Melman, and Roman Meshcheryakov. 2020. Digital steganography and watermarking for digital
images: A review of current research directions. IEEE Access 8 (2020), 166589–166611. DOI:https://doi.org/10.1109/
ACCESS.2020.3022779
[93] H. Farid. 2002. Detecting hidden messages using higher-order statistical models. In Proceedings of the International
Conference on Image Processing, Vol. 2. II–II. DOI:https://doi.org/10.1109/ICIP.2002.1040098
[94] Mansoor Fateh, Mohsen Rezvani, and Yasser Irani. 2021. A new method of coding for steganography based on LSB
matching revisited. Security and Communication Networks 2021 (2021), 1–15.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:24 M. H. Kombrink et al.

[95] Lionel Fillatre. 2012. Adaptive steganalysis of least significant bit replacement in grayscale natural images. IEEE
Transactions on Signal Processing 60, 2 (2012), 556–569. DOI:https://doi.org/10.1109/TSP.2011.2174231
[96] Tomáš Filler and Jessica Fridrich. 2010. Gibbs construction in steganography. IEEE Transactions on Information Foren-
sics and Security 5, 4 (2010), 705–720. DOI:https://doi.org/10.1109/TIFS.2010.2077629
[97] Tomás Filler and Jessica Fridrich. 2011. Design of adaptive steganographic schemes for digital images. In Media
Watermarking, Security, and Forensics III, Nasir D. Memon, Jana Dittmann, Adnan M. Alattar, and Edward J. Delp III
(Eds.), Vol. 7880. International Society for Optics and Photonics, SPIE, 78800F. DOI:https://doi.org/10.1117/12.872192
[98] Jessica Fridrich. 2005. Feature-based steganalysis for JPEG images and its implications for future design of
steganographic schemes. In International Workshop on Information Hiding, Jessica Fridrich (Ed.). Springer, Berlin,
67–81.
[99] Jessica Fridrich and Miroslav Goljan. 2002. Practical steganalysis of digital images: State of the art. In Security and
Watermarking of Multimedia Contents, Edward J. Delp III and Ping Wah Wong (Eds.), Vol. 4675. International Society
for Optics and Photonics, SPIE, 1–13. DOI:https://doi.org/10.1117/12.465263
[100] Jessica Fridrich and Miroslav Goljan. 2003. Digital image steganography using stochastic modulation. In Security
and Watermarking of Multimedia Contents V, Edward J. Delp III and Ping Wah Wong (Eds.), Vol. 5020. International
Society for Optics and Photonics, SPIE, 191–202. DOI:https://doi.org/10.1117/12.479739
[101] Jessica Fridrich and Miroslav Goljan. 2004. On estimation of secret message length in LSB steganography in spatial
domain. In Security, Steganography, and Watermarking of Multimedia Contents VI, Edward J. Delp III and Ping W.
Wong (Eds.), Vol. 5306. International Society for Optics and Photonics, SPIE, 23–34. DOI:https://doi.org/10.1117/12.
521350
[102] J. Fridrich, M. Goljan, and Rui Du. 2001. Detecting LSB steganography in color, and gray-scale images. IEEE Multi-
Media 8, 4 (2001), 22–28. DOI:https://doi.org/10.1109/93.959097
[103] Jessica Fridrich, Miroslav Goljan, and Rui Du. 2001. Steganalysis based on JPEG compatibility. In Multimedia Systems
and Applications, Andrew G. Tescher, Bhaskaran Vasudev, and V. Michael Bove Jr. (Eds.), Vol. 4518. International
Society for Optics and Photonics, SPIE, 275–280. DOI:https://doi.org/10.1117/12.448213
[104] Jessica Fridrich, Miroslav Goljan, and Dorin Hogea. 2003. Steganalysis of JPEG images: Breaking the F5 algorithm.
In International Workshop on Information Hiding, Fabien A. P. Petitcolas (Ed.). Springer, Berlin, 310–323.
[105] Jessica Fridrich, Miroslav Goljan, and David Soukal. 2003. Higher-order statistical steganalysis of palette images.
In Security and Watermarking of Multimedia Contents V, Edward J. Delp III and Ping Wah Wong (Eds.), Vol. 5020.
International Society for Optics and Photonics, SPIE, 178–190. DOI:https://doi.org/10.1117/12.473140
[106] Jessica Fridrich, Miroslav Goljan, and David Soukal. 2004. Perturbed quantization steganography with wet paper
codes. In Proceedings of the 2004 Workshop on Multimedia and Security (MM&Sec’04). Association for Computing
Machinery, New York, NY, USA, 4–15. DOI:https://doi.org/10.1145/1022431.1022435
[107] Jessica Fridrich and Jan Kodovsky. 2012. Rich models for steganalysis of digital images. IEEE Transactions on Infor-
mation Forensics and Security 7, 3 (2012), 868–882. DOI:https://doi.org/10.1109/TIFS.2012.2190402
[108] Jessica Fridrich and Jan Kodovský. 2013. Steganalysis of LSB replacement using parity-aware features. In Interna-
tional Workshop on Information Hiding, Matthias Kirchner and Dipak Ghosal (Eds.). Springer, Berlin, 31–45.
[109] Jessica Fridrich, Jan Kodovský, Vojtěch Holub, and Miroslav Goljan. 2011. Breaking HUGO – The Process Discovery.
In International Workshop on Information Hiding, Tomáš Filler, Tomáš Pevný, Scott Craver, and Andrew Ker (Eds.).
Springer, Berlin, Heidelberg, 85–101.
[110] Jessica Fridrich, Jan Kodovskỳ, Vojtěch Holub, and Miroslav Goljan. 2011. Steganalysis of content-adaptive steganog-
raphy in spatial domain. In International Workshop on Information Hiding. Springer, 102–117.
[111] J. Fridrich and M. Long. 2000. Steganalysis of LSB encoding in color images. In Proceedings of the 2000 IEEE Inter-
national Conference on Multimedia and Expo (ICME’00). Latest Advances in the Fast Changing World of Multimedia,
Vol. 3. 1279–1282. DOI:https://doi.org/10.1109/ICME.2000.871000
[112] Jessica Fridrich, Tomáš Pevný, and Jan Kodovský. 2007. Statistically undetectable Jpeg steganography: Dead ends
challenges, and opportunities. In Proceedings of the 9th Workshop on Multimedia & Security (MM&Sec’07). Association
for Computing Machinery, New York, NY, USA, 3–14. DOI:https://doi.org/10.1145/1288869.1288872
[113] Jessica Fridrich, David Soukal, and Miroslav Goljan. 2005. Maximum likelihood estimation of length of secret mes-
sage embedded using ±k steganography in spatial domain. In Security, Steganography, and Watermarking of Mul-
timedia Contents VII, Edward J. Delp III and Ping W. Wong (Eds.), Vol. 5681. International Society for Optics and
Photonics, SPIE, 595–606. DOI:https://doi.org/10.1117/12.584426
[114] Dongdong Fu, Yun Q. Shi, Dekun Zou, and Guorong Xuan. 2006. JPEG steganalysis using empirical transition matrix
in block DCT domain. In 2006 IEEE Workshop on Multimedia Signal Processing. 310–313. DOI:https://doi.org/10.1109/
MMSP.2006.285320
[115] Yunkai Gao, Xiaolong Li, Bin Yang, and Yifeng Lu. 2009. Detecting LSB matching by characterizing the amplitude of
histogram. In 2009 IEEE International Conference on Acoustics, Speech and Signal Processing. 1505–1508. DOI:https:
//doi.org/10.1109/ICASSP.2009.4959881

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:25

[116] Mallika Garg, Jagpal Singh Ubhi, and Ashwani Kumar Aggarwal. 2023. Neural style transfer for image steganogra-
phy and destylization with supervised image to image translation. Multimedia Tools and Applications 82, 4 (2023),
6271–6288.
[117] Kazem Ghazanfari, Shahrokh Ghaemmaghami, and Saeed R. Khosravi. 2011. LSB++ : An improvement to LSB+
steganography. In 2011 IEEE Region 10 Conference (TENCON’11). 364–368. DOI:https://doi.org/10.1109/TENCON.
2011.6129126
[118] Quentin Giboulot, Patrick Bas, and Rémi Cogranne. 2022. Multivariate side-informed gaussian embedding minimiz-
ing statistical detectability. IEEE Transactions on Information Forensics and Security 17 (2022), 1841–1854. DOI:https:
//doi.org/10.1109/TIFS.2022.3173184
[119] Quentin Giboulot, Rémi Cogranne, Dirk Borghys, and Patrick Bas. 2020. Effects and solutions of cover-source mis-
match in image steganalysis. Signal Processing: Image Communication 86 (2020), 115888. DOI:https://doi.org/10.1016/
j.image.2020.115888
[120] Stuti Goel, Arun Rana, and Manpreet Kaur. 2013. A review of comparison techniques of image steganography. Global
Journal of Computer Science and Technology 13 (2013), 8–14.
[121] Miroslav Goljan, Jessica Fridrich, and Rémi Cogranne. 2014. Rich model for Steganalysis of color images. In 2014
IEEE International Workshop on Information Forensics and Security (WIFS’14). 185–190. DOI:https://doi.org/10.1109/
WIFS.2014.7084325
[122] Miroslav Goljan, Jessica Fridrich, and Taras Holotyak. 2006. New blind steganalysis and its implications. In Secu-
rity, Steganography, and Watermarking of Multimedia Contents VIII, Edward J. Delp III and Ping Wah Wong (Eds.),
Vol. 6072. International Society for Optics and Photonics, SPIE, 607201. DOI:https://doi.org/10.1117/12.643254
[123] Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville,
and Yoshua Bengio. 2014. Generative adversarial nets. Advances in Neural Information Processing Systems 27 (2014),
1–9.
[124] Anuradha Goswami and Sarika Khandelwal. 2016. Hybrid DCT-DWT digital image Steganography. International
Research Journal of Engineering and Technology 5, 6 (2016), 228–233.
[125] Hongmei Gou, Ashwin Swaminathan, and Min Wu. 2007. Noise features for image tampering detection and Ste-
ganalysis. In 2007 IEEE International Conference on Image Processing, Vol. 6. VI–97–VI–100. DOI:https://doi.org/10.
1109/ICIP.2007.4379530
[126] Shreyank N. Gowda and Chun Yuan. 2021. StegColNet: Steganalysis based on an ensemble colorspace approach.
In Structural, Syntactic, and Statistical Pattern Recognition, Andrea Torsello, Luca Rossi, Marcello Pelillo, Battista
Biggio, and Antonio Robles-Kelly (Eds.). Springer International Publishing, Cham, 313–323.
[127] Qingxiao Guan, Jing Dong, and Tieniu Tan. 2011. An effective image steganalysis method based on neighborhood
information of pixels. In 2011 18th IEEE International Conference on Image Processing. 2721–2724. DOI:https://doi.
org/10.1109/ICIP.2011.6116231
[128] Nadjib Guettari, Anne Sophie Capelle-Laizé, and Philippe Carré. 2016. Blind image steganalysis based on evidential
K-Nearest Neighbors. In 2016 IEEE International Conference on Image Processing (ICIP’16). 2742–2746. DOI:https:
//doi.org/10.1109/ICIP.2016.7532858
[129] Linjie Guo, Jiangqun Ni, and Yun Qing Shi. 2012. An efficient JPEG steganographic scheme using uniform em-
bedding. In 2012 IEEE International Workshop on Information Forensics and Security (WIFS’12). 169–174. DOI:https:
//doi.org/10.1109/WIFS.2012.6412644
[130] Linjie Guo, Jiangqun Ni, and Yun Qing Shi. 2014. Uniform embedding for efficient JPEG Steganography. IEEE Trans-
actions on Information Forensics and Security 9, 5 (2014), 814–825. DOI:https://doi.org/10.1109/TIFS.2014.2312817
[131] Linjie Guo, Jiangqun Ni, Wenkang Su, Chengpei Tang, and Yun-Qing Shi. 2015. Using statistical image model for
JPEG steganography: Uniform embedding revisited. IEEE Transactions on Information Forensics and Security 10,
12 (2015), 2669–2680. DOI:https://doi.org/10.1109/TIFS.2015.2473815
[132] Adnan Gutub and Faiza Al-Shaarani. 2020. Efficient implementation of multi-image secret hiding based on LSB and
DWT steganography comparisons. Arabian Journal for Science and Engineering 45, 4 (2020), 2631–2644.
[133] Adnan Abdul-Aziz Gutub, et al. 2010. Pixel indicator technique for RGB image steganography. Journal of Emerging
Technologies in Web Intelligence 2, 1 (2010), 56–64.
[134] Mohamed Abdel Hameed, Saleh Aly, and M. Hassaballah. 2018. An efficient data hiding method based on adaptive
directional pixel value differencing (ADPVD). Multimedia Tools and Applications 77, 12 (2018), 14705–14723.
[135] Nagham Hamid, Abid Yahya, R. Badlishah Ahmad, and Osamah M. Al-Qershi. 2012. Image steganography tech-
niques: An overview. International Journal of Computer Science and Security (IJCSS) 6, 3 (2012), 168–187.
[136] Mohamed Hamidi, Mohamed El Haziti, Hocine Cherifi, and Mohammed El Hassouni. 2018. Hybrid blind robust
image watermarking technique based on DFT-DCT and Arnold transform. Multimedia Tools and Applications 77,
20 (2018), 27181–27214.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:26 M. H. Kombrink et al.

[137] Jeremiah Joseph Harmsen and William A. Pearlman. 2003. Steganalysis of additive-noise modelable information
hiding. In Security and Watermarking of Multimedia Contents V, Edward J. Delp III and Ping Wah Wong (Eds.),
Vol. 5020. International Society for Optics and Photonics, SPIE, 131–142. DOI:https://doi.org/10.1117/12.476813
[138] M. Hassaballah, Mohamed Abdel Hameed, Ali Ismail Awad, and Khan Muhammad. 2021. A novel image steganog-
raphy method for industrial Internet of Things security. IEEE Transactions on Industrial Informatics 17, 11 (2021),
7743–7751. DOI:https://doi.org/10.1109/TII.2021.3053595
[139] T. A. Hawi, M. A. Qutayri, and H. Barada. 2004. Steganalysis attacks on stego-images using stego-signatures and
statistical image properties. In 2004 IEEE Region 10 Conference (TENCON’04), Vol. B. 104–107. DOI:https://doi.org/
10.1109/TENCON.2004.1414542
[140] Jamie Hayes and George Danezis. 2017. Generating steganographic images via adversarial training. In Advances in
Neural Information Processing Systems, I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan,
and R. Garnett (Eds.), Vol. 30. Curran Associates, Inc. https://proceedings.neurips.cc/paper_files/paper/2017/file/
fe2d010308a6b3799a3d9c728ee74244-Paper.pdf
[141] J. Hemalatha, M. Sekar, Chandan Kumar, Adnan Gutub, and Aditya Kumar Sahu. 2023. Towards improving the
performance of blind image steganalyzer using third-order SPAM features and ensemble classifier. Journal of Infor-
mation Security and Applications 76 (2023), 103541. DOI:https://doi.org/10.1016/j.jisa.2023.103541
[142] Stefan Hetzl and Petra Mutzel. 2005. A graph–theoretic approach to steganography. In International Conference on
Communications and Multimedia Security, Jana Dittmann, Stefan Katzenbeisser, and Andreas Uhl (Eds.). Springer,
Berlin, 119–128.
[143] Vojtěch Holub and Jessica Fridrich. 2012. Designing steganographic distortion using directional filters. In 2012 IEEE
International Workshop on Information Forensics and Security (WIFS’12). 234–239. DOI:https://doi.org/10.1109/WIFS.
2012.6412655
[144] Vojtěch Holub and Jessica Fridrich. 2013. Digital image steganography using universal distortion. In Proceedings of
the 1st ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’13). Association for Computing
Machinery, New York, NY, USA, 59–68. DOI:https://doi.org/10.1145/2482513.2482514
[145] Vojtech Holub and Jessica Fridrich. 2013. Random projections of residuals for digital image steganalysis. IEEE Trans-
actions on Information Forensics and Security 8, 12 (2013), 1996–2006. DOI:https://doi.org/10.1109/TIFS.2013.2286682
[146] Vojtěch Holub and Jessica Fridrich. 2015. Low-complexity features for JPEG steganalysis using undecimated DCT.
IEEE Transactions on Information Forensics and Security 10, 2 (2015), 219–228. DOI:https://doi.org/10.1109/TIFS.2014.
2364918
[147] Vojtěch Holub and Jessica Fridrich. 2015. Phase-aware projection model for steganalysis of JPEG images. In Media
Watermarking, Security, and Forensics 2015, Adnan M. Alattar, Nasir D. Memon, and Chad D. Heitzenrater (Eds.),
Vol. 9409. International Society for Optics and Photonics, SPIE, 94090T. DOI:https://doi.org/10.1117/12.2075239
[148] Vojtěch Holub, Jessica Fridrich, and Tomáš Denemark. 2014. Universal distortion function for steganography in an
arbitrary domain. EURASIP Journal on Information Security 2014, 1 (2014), 1–13.
[149] Donghui Hu, Liang Wang, Wenjie Jiang, Shuli Zheng, and Bin Li. 2018. A novel image steganography method via
deep convolutional generative adversarial networks. IEEE Access 6 (2018), 38303–38314. DOI:https://doi.org/10.1109/
ACCESS.2018.2852771
[150] Xianglei Hu, Jiangqun Ni, Weizhe Zhang, and Jiwu Huang. 2021. Efficient JPEG batch steganography using intrinsic
energy of image contents. IEEE Transactions on Information Forensics and Security 16 (2021), 4544–4558. DOI:https:
//doi.org/10.1109/TIFS.2021.3109464
[151] Fangjun Huang, Jiwu Huang, and Yun-Qing Shi. 2012. New channel selection rule for JPEG steganography. IEEE
Transactions on Information Forensics and Security 7, 4 (2012), 1181–1191. DOI:https://doi.org/10.1109/TIFS.2012.
2198213
[152] Fangjun Huang, Bin Li, and Jiwu Huang. 2007. Attack LSB matching steganography by counting alteration rate of
the number of neighbourhood gray levels. In 2007 IEEE International Conference on Image Processing, Vol. 1. I–401–
I–404. DOI:https://doi.org/10.1109/ICIP.2007.4378976
[153] Junwen Huang, Jiangqun Ni, Linhong Wan, and Jingwen Yan. 2019. A customized convolutional neural network
with low model complexity for JPEG steganalysis. In Proceedings of the ACM Workshop on Information Hiding
and Multimedia Security (IH&MMSec’19). Association for Computing Machinery, New York, NY, USA, 198–203.
DOI:https://doi.org/10.1145/3335203.3335734
[154] Jiwu Huang and Yun Shi. 1998. Adaptive image watermarking scheme based on visual masking. Electronics Letters
34, 8 (April 1998), 748–750. https://doi.org/10.1049/el:19980545
[155] Mehdi Hussain and Mureed Hussain. 2013. A survey of image steganography techniques. International Journal of
Advanced Science and Technology 54 (2013), 113–124.
[156] Mehdi Hussain, Qaiser Riaz, Shahzad Saleem, Abdul Ghafoor, and Ki-Hyun Jung. 2021. Enhanced adaptive data
hiding method using LSB and pixel value differencing. Multimedia Tools and Applications 80, 13 (2021), 20381–20401.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:27

[157] Mehdi Hussain, Ainuddin Wahid Abdul Wahab, Yamani Idna Bin Idris, Anthony T. S. Ho, and Ki-Hyun Jung. 2018.
Image steganography in spatial domain: A survey. Signal Processing: Image Communication 65 (2018), 46–66.
[158] A. Idakwo, B. Muazu, A. Adedokun, and O. Sadiq. 2020. An extensive survey of digital image steganography: State
of the art. ATBU Journal of Science, Technology and Education 8, 2 (2020), 40–54. http://www.atbuftejoste.com/index.
php/joste/article/view/972
[159] Anastasia Ioannidou, Spyros T. Halkidis, and George Stephanides. 2012. A novel technique for image steganography
based on a high payload method and edge detection. Expert Systems with Applications 39, 14 (2012), 11517–11524.
DOI:https://doi.org/10.1016/j.eswa.2012.02.106
[160] Mao Jia-Fa, Niu Xin-Xin, Xiao Gang, Sheng Wei-Guo, and Zhang Na-Na. 2016. A steganalysis method in the DCT
domain. Multimedia Tools and Applications 75, 10 (2016), 5999–6019.
[161] N. F. Johnson and S. Jajodia. 1998. Steganalysis: The investigation of hidden information. In 1998 IEEE Information
Technology Conference, Information Environment for the Future (Cat. No. 98EX228). 113–116. DOI:https://doi.org/10.
1109/IT.1998.713394
[162] Neil F. Johnson and Sushil Jajodia. 1998. Steganalysis of images created using current steganography software. In
International Workshop on Information Hiding, David Aucsmith (Ed.). Springer, Berlin, 273–289.
[163] Saurabh V. Joshi, Ajinkya A. Bokil, Nikhil A. Jain, and Dipali Koshti. 2012. Image steganography combination of
spatial and frequency domain. International Journal of Computer Applications 53, 5 (2012), 25–29.
[164] Ki-Hyun Jung and Kee-Young Yoo. 2015. Steganographic method based on interpolation and LSB substitution of
digital images. Multimedia Tools and Applications 74, 6 (2015), 2143–2155.
[165] Inas Jawad Kadhim, Prashan Premaratne, and Peter James Vial. 2020. High capacity adaptive image steganography
with cover region selection using dual-tree complex wavelet transform. Cognitive Systems Research 60 (2020), 20–32.
DOI:https://doi.org/10.1016/j.cogsys.2019.11.002
[166] Inas Jawad Kadhim, Prashan Premaratne, Peter James Vial, and Brendan Halloran. 2019. Comprehensive survey of
image steganography: Techniques, evaluations, and trends in future research. Neurocomputing 335 (2019), 299–326.
DOI:https://doi.org/10.1016/j.neucom.2018.06.075
[167] David Kahn. 1996. The history of steganography. In International Workshop on Information Hiding, Ross Anderson
(Ed.). Springer, Berlin, 1–5.
[168] Hamidreza Rashidy Kanan and Bahram Nazeri. 2014. A novel image steganography scheme with high embed-
ding capacity and tunable visual image quality based on a genetic algorithm. Expert Systems with Applications 41,
14 (2014), 6123–6130. DOI:https://doi.org/10.1016/j.eswa.2014.04.022
[169] Konstantinos Karampidis, Ergina Kavallieratou, and Giorgos Papadourakis. 2018. A review of image steganalysis
techniques for digital forensics. Journal of Information Security and Applications 40 (2018), 217–235. DOI:https://doi.
org/10.1016/j.jisa.2018.04.005
[170] Manveer Kaur and Gagandeep Kaur. 2014. Review of various steganalysis techniques. International Journal of Com-
puter Science and Information Technologies 5, 2 (2014), 1744–1747.
[171] Sharanpreet Kaur, Surender Singh, Manjit Kaur, and Heung-No Lee. 2022. A systematic review of computational
image steganography approaches. Archives of Computational Methods in Engineering 29, 7 (2022), 4775–4797.
[172] Akanksha Kaushal and Vineeta Choudhary. 2013. Secured image steganography using different transform domain.
International Journal of Computer Applications 77, 2 (2013), 24–28.
[173] Eiji Kawaguchi and Richard O. Eason. 1999. Principles and applications of BPCS steganography. In Multimedia Sys-
tems and Applications, Andrew G. Tescher, Bhaskaran Vasudev, V. Michael Bove Jr., and Barbara Derryberry (Eds.),
Vol. 3528. International Society for Optics and Photonics, SPIE, 464–473. DOI:https://doi.org/10.1117/12.337436
[174] A. D. Ker. 2005. Steganalysis of LSB matching in grayscale images. IEEE Signal Processing Letters 12, 6 (2005), 441–444.
DOI:https://doi.org/10.1109/LSP.2005.847889
[175] Andrew David Ker. 2004. Quantitative evaluation of pairs and RS steganalysis. In Security, Steganography, and Wa-
termarking of Multimedia Contents VI, Edward J. Delp III and Ping W. Wong (Eds.), Vol. 5306. International Society
for Optics and Photonics, SPIE, 83–97. DOI:https://doi.org/10.1117/12.526720
[176] Andrew D. Ker. 2005. A general framework for structural steganalysis of LSB replacement. In International Workshop
on Information Hiding, Mauro Barni, Jordi Herrera-Joancomartí, Stefan Katzenbeisser, and Fernando Pérez-González
(Eds.). Springer, Berlin, 296–311.
[177] Andrew D. Ker. 2005. Improved detection of LSB steganography in grayscale images. In International Workshop on
Information Hiding, Jessica Fridrich (Ed.). Springer, Berlin, 97–115.
[178] Andrew D. Ker. 2007. A fusion of maximum likelihood and structural steganalysis. In International Workshop
on Information Hiding, Teddy Furon, François Cayre, Gwenaël Doërr, and Patrick Bas (Eds.). Springer, Berlin,
204–219.
[179] Andrew D. Ker. 2007. Steganalysis of embedding in two least-significant bits. IEEE Transactions on Information Foren-
sics and Security 2, 1 (2007), 46–54. DOI:https://doi.org/10.1109/TIFS.2006.890519

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:28 M. H. Kombrink et al.

[180] Andrew D. Ker and Rainer Böhme. 2008. Revisiting weighted stego-image steganalysis. In Security, Forensics,
Steganography, and Watermarking of Multimedia Contents X, Edward J. Delp III, Ping Wah Wong, Jana Dittmann,
and Nasir D. Memon (Eds.), Vol. 6819. International Society for Optics and Photonics, SPIE, 56–72. DOI:https:
//doi.org/10.1117/12.766820
[181] Andrew D. Ker and Tomás Pevný. 2011. A new paradigm for steganalysis via clustering. In Media Watermarking,
Security, and Forensics III, Nasir D. Memon, Jana Dittmann, Adnan M. Alattar, and Edward J. Delp III (Eds.), Vol. 7880.
International Society for Optics and Photonics, SPIE, 78800U. DOI:https://doi.org/10.1117/12.872888
[182] Andrew David Ker and Tomas Pevny. 2012. Batch steganography in the real world. In Proceedings of the on Mul-
timedia and Security (MM&Sec’12). Association for Computing Machinery, New York, NY, USA, 1–10. DOI:https:
//doi.org/10.1145/2361407.2361409
[183] Andrew D. Ker and Tomás Pevný. 2012. Identifying a steganographer in realistic and heterogeneous data sets. In
Media Watermarking, Security, and Forensics 2012, Nasir D. Memon, Adnan M. Alattar, and Edward J. Delp III (Eds.),
Vol. 8303. International Society for Optics and Photonics, SPIE, 83030N. DOI:https://doi.org/10.1117/12.910565
[184] Mehdi Kharrazi, Husrev T. Sencar, and Nasir Memon. 2005. Benchmarking steganographic and steganalysis tech-
niques. In Security, Steganography, and Watermarking of Multimedia Contents VII, Edward J. Delp III and Ping W.
Wong (Eds.), Vol. 5681. International Society for Optics and Photonics, SPIE, 252–263. DOI:https://doi.org/10.1117/
12.587375
[185] Mehdi Kharrazi, Husrev T. Sencar, and Nasir Memon. 2006. Improving steganalysis by fusion techniques: A case
study with image steganography. In Transactions on Data Hiding and Multimedia Security I, Yun Q. Shi (Ed.). Springer,
Berlin, 123–137.
[186] Mehdi Kharrazi, Husrev Taha Sencar, and Nasir D. Memon. 2006. Performance study of common image steganog-
raphy and steganalysis techniques. Journal of Electronic Imaging 15, 4 (2006), 041104. DOI:https://doi.org/10.1117/1.
2400672
[187] Hamza Kheddar, Mustapha Hemis, Yassine Himeur, David Megías, and Abbes Amira. 2024. Deep learning for ste-
ganalysis of diverse data types: A review of methods, taxonomy, challenges and future directions. Neurocomputing
581 (2024), 127528. DOI:https://doi.org/10.1016/j.neucom.2024.127528
[188] Masoumeh Khodaei, Bahram Sadeghi Bigham, and Karim Faez. 2016. Adaptive data hiding, using pixel-value-
differencing and LSB substitution. Cybernetics and Systems 47, 8 (2016), 617–628. DOI:https://doi.org/10.1080/
01969722.2016.1214459 arXiv:https://doi.org/10.1080/01969722.2016.1214459
[189] M. Khodaei and K. Faez. 2012. New adaptive steganographic method using least-significant-bit substitution and
pixel-value differencing. IET Image Processing 6, 6 (Aug. 2012), 677–686. https://digital-library.theiet.org/content/
journals/10.1049/iet-ipr.2011.0059
[190] Younhee Kim, Zoran Duric, and Dana Richards. 2007. Modified matrix encoding technique for minimal distortion
steganography. In International Workshop on Information Hiding, Jan L. Camenisch, Christian S. Collberg, Neil F.
Johnson, and Phil Sallee (Eds.). Springer, Berlin, 314–327.
[191] Young-Sik Kim, O-Hyung Kwon, and Rae-Hong Park. 1999. Wavelet based watermarking method for digital images
using the human visual system. In 1999 IEEE International Symposium on Circuits and Systems (ISCAS), Vol. 4. 80–83.
https://doi.org/10.1109/ISCAS.1999.779947
[192] Jan Kodovský and Jessica Fridrich. 2008. On completeness of feature spaces in blind steganalysis. In Proceedings
of the 10th ACM Workshop on Multimedia and Security (MM&Sec’08). Association for Computing Machinery, New
York, NY, USA, 123–132. DOI:https://doi.org/10.1145/1411328.1411352
[193] Jan Kodovsky and Jessica Fridrich. 2009. Calibration revisited. In Proceedings of the 11th ACM Workshop on Mul-
timedia and Security (MM&Sec’09). Association for Computing Machinery, New York, NY, USA, 63–74. DOI:https:
//doi.org/10.1145/1597817.1597830
[194] Jan Kodovsky and Jessica Fridrich. 2011. Steganalysis in high dimensions: Fusing classifiers built on random sub-
spaces. In Media Watermarking, Security, and Forensics III, Nasir D. Memon, Jana Dittmann, Adnan M. Alattar,
and Edward J. Delp III (Eds.), Vol. 7880. International Society for Optics and Photonics, SPIE, 78800L. DOI:https:
//doi.org/10.1117/12.872279
[195] Jan Kodovský and Jessica Fridrich. 2012. Steganalysis of JPEG images using rich models. In Media Watermarking,
Security, and Forensics 2012, Nasir D. Memon, Adnan M. Alattar, and Edward J. Delp III (Eds.), Vol. 8303. International
Society for Optics and Photonics, SPIE, 83030A. DOI:https://doi.org/10.1117/12.907495
[196] Jan Kodovsky, Jessica Fridrich, and Vojtech Holub. 2011. On dangers of overtraining steganography to incomplete
cover model. In Proceedings of the 13th ACM Multimedia Workshop on Multimedia and Security (MM&Sec’11). Asso-
ciation for Computing Machinery, New York, NY, USA, 69–76. DOI:https://doi.org/10.1145/2037252.2037266
[197] Jan Kodovsky, Jessica Fridrich, and Vojtěch Holub. 2012. Ensemble classifiers for steganalysis of digital media.
IEEE Transactions on Information Forensics and Security 7, 2 (2012), 432–444. DOI:https://doi.org/10.1109/TIFS.2011.
2175919

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:29

[198] Jan Kodovský, Vahid Sedighi, and Jessica Fridrich. 2014. Study of cover source mismatch in steganalysis and ways
to mitigate its impact. In Media Watermarking, Security, and Forensics 2014, Adnan M. Alattar, Nasir D. Memon,
and Chad D. Heitzenrater (Eds.), Vol. 9028. International Society for Optics and Photonics, SPIE, 90280J. DOI:https:
//doi.org/10.1117/12.2039693
[199] Jan Kodovský and Jessica Fridrich. 2010. Quantitative structural steganalysis of Jsteg. IEEE Transactions on Informa-
tion Forensics and Security 5, 4 (2010), 681–693. DOI:https://doi.org/10.1109/TIFS.2010.2056684
[200] Robert Krenn. 2004. Steganography and steganalysis. https://www.krenn.nl/univ/cry/steg/article.pdf
[201] K. B. Shiva Kumar, K. B. Raja, R. K. Chhotaray, and Sabyasachi Pattanaik. 2010. Bit length replacement steganography
based on DCT coefficients. International Journal of Engineering Science and Technology 2, 8 (2010), 3561–3570.
[202] Vijay Kumar and Dinesh Kumar. 2018. A modified DWT-based image steganography technique. Multimedia Tools
and Applications 77, 11 (2018), 13279–13308.
[203] Vijay Kumar, Sahil Sharma, Chandan Kumar, and Aditya Kumar Sahu. 2023. Latest trends in deep learning tech-
niques for image steganography. International Journal of Digital Crime and Forensics (IJDCF) 15, 1 (2023), 1–14.
[204] Debina Laishram and Themrichon Tuithung. 2021. A novel minimal distortion-based edge adaptive image steganog-
raphy scheme using local complexity. Multimedia Tools and Applications 80, 1 (2021), 831–854.
[205] Kwangsoo Lee, Andreas Westfeld, and Sangjin Lee. 2006. Category attack for LSB steganalysis of JPEG images. In
Digital Watermarking, Yun Qing Shi and Byeungwoo Jeon (Eds.). Springer, Berlin, 35–48.
[206] Kwangsoo Lee, Andreas Westfeld, and Sangjin Lee. 2007. Generalised category attack—improving histogram-based
attack on JPEG LSB embedding. In International Workshop on Information Hiding, Teddy Furon, François Cayre,
Gwenaël Doërr, and Patrick Bas (Eds.). Springer, Berlin, 378–391.
[207] Daniel Lerch-Hostalot and David Megías. 2013. LSB matching steganalysis based on patterns of pixel differences
and random embedding. Computers & Security 32 (2013), 192–206. DOI:https://doi.org/10.1016/j.cose.2012.11.005
[208] Daniel Lerch-Hostalot and David Megías. 2016. Unsupervised steganalysis based on artificial training sets. Engineer-
ing Applications of Artificial Intelligence 50 (2016), 45–59. DOI:https://doi.org/10.1016/j.engappai.2015.12.013
[209] Bin Li, Jiwu Huang, and Yun Q. Shi. 2008. Textural features based universal steganalysis. In Security, Forensics,
Steganography, and Watermarking of Multimedia Contents X, Edward J. Delp III, Ping Wah Wong, Jana Dittmann,
and Nasir D. Memon (Eds.), Vol. 6819. International Society for Optics and Photonics, SPIE, 681912. DOI:https:
//doi.org/10.1117/12.765817
[210] Bin Li, Yun Q. Shi, and Jiwu Huang. 2008. Steganalysis of YASS. In Proceedings of the 10th ACM Workshop on Multi-
media and Security (MM&Sec’08). Association for Computing Machinery, New York, NY, USA, 139–148. DOI:https:
//doi.org/10.1145/1411328.1411354
[211] Bin Li, Ming Wang, Jiwu Huang, and Xiaolong Li. 2014. A new cost function for spatial image steganography. In
2014 IEEE International Conference on Image Processing (ICIP’14). 4206–4210. DOI:https://doi.org/10.1109/ICIP.2014.
7025854
[212] Bin Li, Ming Wang, Xiaolong Li, Shunquan Tan, and Jiwu Huang. 2015. A strategy of clustering modification direc-
tions in spatial image steganography. IEEE Transactions on Information Forensics and Security 10, 9 (2015), 1905–1917.
DOI:https://doi.org/10.1109/TIFS.2015.2434600
[213] Chuanlong Li, Yumeng Jiang, and Marta Cheslyar. 2018. Embedding image through generated intermediate
medium using deep convolutional generative adversarial network. Computers, Materials & Continua 56, 2 (2018),
313–324.
[214] Fengyong Li, Zongliang Yu, and Chuan Qin. 2022. GAN-based spatial image steganography with cross feedback
mechanism. Signal Processing 190 (2022), 108341. DOI:https://doi.org/10.1016/j.sigpro.2021.108341
[215] Xiaolong Li, Bin Li, Xiangyang Luo, Bin Yang, and Ruihui Zhu. 2013. Steganalysis of a PVD-based content adaptive
image steganography. Signal Processing 93, 9 (2013), 2529–2538. DOI:https://doi.org/10.1016/j.sigpro.2013.03.029
[216] Xiaolong Li, Bin Yang, Daofang Cheng, and Tieyong Zeng. 2009. A generalization of LSB matching. IEEE Signal
Processing Letters 16, 2 (2009), 69–72.
[217] Xiaolong Li, Tieyong Zeng, and Bin Yang. 2008. Detecting LSB matching by applying calibration technique for
difference image. In Proceedings of the 10th ACM Workshop on Multimedia and Security (MM&Sec’08). Association
for Computing Machinery, New York, NY, USA, 133–138. DOI:https://doi.org/10.1145/1411328.1411353
[218] Zhaotong Li and Ying He. 2018. Steganography with pixel-value differencing and modulus function based on PSO.
Journal of Information Security and Applications 43 (2018), 47–52. DOI:https://doi.org/10.1016/j.jisa.2018.10.006
[219] Xin Liao, Jiaojiao Yin, Mingliang Chen, and Zheng Qin. 2022. Adaptive payload distribution in multiple images
steganography based on image texture features. IEEE Transactions on Dependable and Secure Computing 19, 2 (2022),
897–911. DOI:https://doi.org/10.1109/TDSC.2020.3004708
[220] Wen-Nung Lie and Li Chun Chang. 1999. Data hiding in images with adaptive numbers of least significant bits based
on the human visual system. In Proceedings 1999 International Conference on Image Processing (Cat. No. 99CH36348),
Vol. 1. 286–290. DOI:https://doi.org/10.1109/ICIP.1999.821615

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:30 M. H. Kombrink et al.

[221] Wen-Nung Lie and Guo-Shiang Lin. 2005. A feature-based classification technique for blind image steganalysis. IEEE
Transactions on Multimedia 7, 6 (2005), 1007–1020. DOI:https://doi.org/10.1109/TMM.2005.858377
[222] Wen-Nung Lie, Guo-Shiang Lin, Chih-Liang Wu, and Ta-Chun Wang. 2000. Robust image watermarking on the DCT
domain. In 2000 IEEE International Symposium on Circuits and Systems (ISCAS), Vol. 1. 228–231. https://doi.org/10.
1109/ISCAS.2000.857069
[223] Wen-Bin Lin, Tai-Hung Lai, and Ko-Chin Chang. 2021. Statistical feature-based steganalysis for pixel-value differ-
encing steganography. EURASIP Journal on Advances in Signal Processing 2021, 1 (2021), 1–18.
[224] Hsing-Han Liu and Chuan-Min Lee. 2019. High-capacity reversible image steganography based on pixel value or-
dering. EURASIP Journal on Image and Video Processing 2019, 1 (2019), 1–15.
[225] Hsing-Han Liu, Yuh-Chi Lin, and Chia-Ming Lee. 2019. A digital data hiding scheme based on pixel-value differenc-
ing and side match method. Multimedia Tools and Applications 78, 9 (2019), 12157–12181.
[226] Minglin Liu, Weiqi Luo, Peijia Zheng, and Jiwu Huang. 2021. A new adversarial embedding method for enhancing
image steganography. IEEE Transactions on Information Forensics and Security 16 (2021), 4621–4634. DOI:https://doi.
org/10.1109/TIFS.2021.3111748
[227] Qingzhong Liu. 2011. Steganalysis of DCT-embedding based adaptive steganography and YASS. In Proceedings of the
13th ACM Multimedia Workshop on Multimedia and Security (MM&Sec’11). Association for Computing Machinery,
New York, NY, USA, 77–86. DOI:https://doi.org/10.1145/2037252.2037267
[228] Qingzhong Liu, A. H. Sung, Jianyun Xu, and B. M. Ribeiro. 2006. Image complexity and feature extraction for ste-
ganalysis of LSB matching steganography. In 18th International Conference on Pattern Recognition (ICPR’06), Vol. 2.
267–270. DOI:https://doi.org/10.1109/ICPR.2006.684
[229] Qingzhong Liu, Andrew H. Sung, Zhongxue Chen, and Jianyun Xu. 2008. Feature mining and pattern classifica-
tion for steganalysis of LSB matching steganography in grayscale images. Pattern Recognition 41, 1 (2008), 56–66.
DOI:https://doi.org/10.1016/j.patcog.2007.06.005
[230] Julio Lopez-Hernandez, Raul Martinez-Noriega, Mariko Nakano-Miyatake, and Kazuhiko Yamaguchi. 2008. Detec-
tion of BPCS-steganography using SMWCF steganalysis and SVM. In 2008 International Symposium on Information
Theory and Its Applications. 1–5. DOI:https://doi.org/10.1109/ISITA.2008.4895497
[231] Peizhong Lu, Xiangyang Luo, Qingyang Tang, and Li Shen. 2005. An improved sample pairs method for detection of
LSB embedding. In International Workshop on Information Hiding, Jessica Fridrich (Ed.). Springer, Berlin, 116–127.
[232] Shao-Ping Lu, Rong Wang, Tao Zhong, and Paul L. Rosin. 2021. Large-capacity image steganography based on
invertible neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition
(CVPR’21). 10816–10825.
[233] Tzu-Chuen Lu, Chun-Ya Tseng, and Jhih-Huei Wu. 2015. Dual imaging-based reversible hiding technique using LSB
matching. Signal Processing 108 (2015), 77–89.
[234] Ivans Lubenko and Andrew D. Ker. 2011. Steganalysis using logistic regression. In Media Watermarking, Security, and
Forensics III, Nasir D. Memon, Jana Dittmann, Adnan M. Alattar, and Edward J. Delp III (Eds.), Vol. 7880. International
Society for Optics and Photonics, SPIE, 78800K. DOI:https://doi.org/10.1117/12.872245
[235] Weiqi Luo, Fangjun Huang, and Jiwu Huang. 2010. Edge adaptive image steganography based on LSB matching
revisited. IEEE Transactions on Information Forensics and Security 5, 2 (2010), 201–214. DOI:https://doi.org/10.1109/
TIFS.2010.2041812
[236] Xiangyang Luo, Fenlin Liu, Jianming Chen, and Yining Zhang. 2008. Image universal steganalysis based on wavelet
packet transform. In 2008 IEEE 10th Workshop on Multimedia Signal Processing. 780–784. DOI:https://doi.org/10.1109/
MMSP.2008.4665180
[237] Xiangyang Luo, Fenlin Liu, Chunfang Yang, Shiguo Lian, and Ying Zeng. 2012. Steganalysis of adaptive image
steganography in multiple gray code bit-planes. Multimedia Tools and Applications 57, 3 (2012), 651–667.
[238] Xiang-Yang Luo, Dao-Shun Wang, Ping Wang, and Fen-Lin Liu. 2008. A review on blind detection for image
steganography. Signal Processing 88, 9 (2008), 2138–2157.
[239] Yuanjing Luo, Jiaohua Qin, Xuyu Xiang, and Yun Tan. 2021. Coverless image steganography based on multi-object
recognition. IEEE Transactions on Circuits and Systems for Video Technology 31, 7 (2021), 2779–2791. DOI:https://doi.
org/10.1109/TCSVT.2020.3033945
[240] Siwei Lyu and Hany Farid. 2003. Detecting hidden messages using higher-order statistics and support vector ma-
chines. In International Workshop on Information Hiding, Fabien A. P. Petitcolas (Ed.). Springer, Berlin, 340–354.
[241] Siwei Lyu and Hany Farid. 2004. Steganalysis using color wavelet statistics and one-class support vector machines.
In Security, Steganography, and Watermarking of Multimedia Contents VI, Edward J. Delp III and Ping W. Wong (Eds.),
Vol. 5306. International Society for Optics and Photonics, SPIE, 35–45. DOI:https://doi.org/10.1117/12.526012
[242] S. Lyu and H. Farid. 2006. Steganalysis using higher-order image statistics. IEEE Transactions on Information Forensics
and Security 1, 1 (2006), 111–119. DOI:https://doi.org/10.1109/TIFS.2005.863485

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:31

[243] Yuanyuan Ma, Xinquan Yu, Xiangyang Luo, Dong Liu, and Yi Zhang. 2023. Adaptive feature selection for image
steganalysis based on classification metrics. Information Sciences 644 (2023), 118973. DOI:https://doi.org/10.1016/j.
ins.2023.118973
[244] Pratap Chandra Mandal, Imon Mukherjee, Goutam Paul, and B. N. Chatterji. 2022. Digital image steganography: A
literature survey. Information Sciences 609 (2022), 1451–1488. DOI:https://doi.org/10.1016/j.ins.2022.07.120
[245] C. Manikopoulos, Yun-Qing Shi, Sui Song, Zheng Zhang, Zhicheng Ni, and Dekun Zou. 2002. Detection of block
DCT-based steganography in gray-scale images. In 2002 IEEE Workshop on Multimedia Signal Processing. 355–358.
DOI:https://doi.org/10.1109/MMSP.2002.1203319
[246] Alejandro Martín, Alfonso Hernández, Moutaz Alazab, Jason Jung, and David Camacho. 2023. Evolving genera-
tive adversarial networks to improve image steganography. Expert Systems with Applications 222 (2023), 119841.
DOI:https://doi.org/10.1016/j.eswa.2023.119841
[247] L. M. Marvel, C. G. Boncelet, and C. T. Retter. 1999. Spread spectrum image steganography. IEEE Transactions on
Image Processing 8, 8 (1999), 1075–1083. DOI:https://doi.org/10.1109/83.777088
[248] S. M. Masud Karim, Md. Saifur Rahman, and Md. Ismail Hossain. 2011. A new approach for LSB based image
steganography using secret key. In 14th International Conference on Computer and Information Technology (ICCIT’11).
286–291. DOI:https://doi.org/10.1109/ICCITechn.2011.6164800
[249] Ruohan Meng, Qi Cui, Zhili Zhou, Zhangjie Fu, and Xingming Sun. 2019. A steganography algorithm based on
CycleGAN for covert communication in the Internet of Things. IEEE Access 7 (2019), 90574–90584. DOI:https://doi.
org/10.1109/ACCESS.2019.2920956
[250] Ruohan Meng, Steven G. Rice, Jin Wang, and Xingming Sun. 2018. A fusion steganographic algorithm based on
faster R-CNN. Computers, Materials & Continua 55, 1 (2018), 1–16.
[251] Jarno Mielikainen. 2006. LSB matching revisited. IEEE Signal Processing Letters 13, 5 (2006), 285–287.
[252] Julián D. Miranda and Diego J. Parada. 2022. LSB steganography detection in monochromatic still images using
artificial neural networks. Multimedia Tools and Applications 81, 1 (2022), 785–805.
[253] Aref Miri and Karim Faez. 2017. Adaptive image steganography based on transform domain via genetic algorithm.
Optik 145 (2017), 158–168.
[254] Aref Miri and Karim Faez. 2018. An image steganography method based on integer wavelet transform. Multimedia
Tools and Applications 77, 11 (2018), 13133–13144.
[255] Xianbo Mo, Shunquan Tan, Bin Li, and Jiwu Huang. 2021. MCTSteg: A Monte Carlo tree search-based reinforce-
ment learning framework for universal non-additive steganography. IEEE Transactions on Information Forensics and
Security 16 (2021), 4306–4320. DOI:https://doi.org/10.1109/TIFS.2021.3104140
[256] Xianbo Mo, Shunquan Tan, Weixuan Tang, Bin Li, and Jiwu Huang. 2023. ReLOAD: Using reinforcement learning to
optimize asymmetric distortion for additive steganography. IEEE Transactions on Information Forensics and Security
18 (2023), 1524–1538. DOI:https://doi.org/10.1109/TIFS.2023.3244094
[257] A. H. Mohsin, A. A. Zaidan, B. B. Zaidan, O. S. Albahri, A. S. Albahri, M. A. Alsalem, K. I. Mohammed, Shahad Nidhal,
Nawar. S. Jalood, Ali Najm Jasim, and Ali. H. Shareef. 2019. New method of image steganography based on particle
swarm optimization algorithm in spatial domain for high embedding capacity. IEEE Access 7 (2019), 168994–169010.
DOI:https://doi.org/10.1109/ACCESS.2019.2949622
[258] Tayana Morkel, Jan H. P. Eloff, and Martin S. Olivier. 2005. An overview of image steganography. In Proceedings of
the 5th Annual Information Security South Africa Conference (ISSA’05), Vol. 1.
[259] Khan Muhammad, Jamil Ahmad, Naeem Ur Rehman, Zahoor Jan, and Muhammad Sajjad. 2017. CISSKA-LSB: Color
image steganography using stego key-directed adaptive LSB substitution method. Multimedia Tools and Applications
76, 6 (2017), 8597–8626.
[260] Khan Muhammad, Muhammad Sajjad, Irfan Mehmood, Seungmin Rho, and Sung Wook Baik. 2016. A novel magic
LSB substitution method (M-LSB-SM) using multi-level encryption and achromatic component of an image. Multi-
media Tools and Applications 75, 22 (2016), 14867–14893.
[261] Pranab K. Muhuri, Zubair Ashraf, and Swati Goel. 2020. A novel image steganographic method based on integer
wavelet transformation and particle swarm optimization. Applied Soft Computing 92 (2020), 106257. DOI:https://doi.
org/10.1016/j.asoc.2020.106257
[262] Trivikram Muralidharan, Aviad Cohen, Assaf Cohen, and Nir Nissim. 2022. The infinite race between steganography
and steganalysis in images. Signal Processing 201 (2022), 108711. DOI:https://doi.org/10.1016/j.sigpro.2022.108711
[263] Amitava Nag, Sushanta Biswas, Debasree Sarkar, and Partha Pratim Sarkar. 2011. A novel technique for image
steganography based on DWT and Huffman encoding. International Journal of Computer Science and Security 4,
6 (2011), 497–610.
[264] Mahboubeh Nazari and Iman Dorostkar Ahmadi. 2020. A novel chaotic steganography method with three ap-
proaches for color and grayscale images based on FIS and DCT with flexible capacity. Multimedia Tools and Ap-
plications 79, 19 (2020), 13693–13724.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:32 M. H. Kombrink et al.

[265] Bui Cong Nguyen, Sang Moon Yoon, and Heung-Kyu Lee. 2006. Multi bit plane image steganography. In Digital
Watermarking, Yun Qing Shi and Byeungwoo Jeon (Eds.). Springer, Berlin, 61–70.
[266] Tuan Duc Nguyen, Somjit Arch-Int, and Ngamnij Arch-Int. 2016. An adaptive multi bit-plane image steganography
using block data-hiding. Multimedia Tools and Applications 75, 14 (2016), 8319–8345.
[267] Michiharu Niimi, Richard O. Eason, Hideki Noda, and Eiji Kawaguchi. 2001. Intensity histogram steganalysis in
BPCS-steganography. In Security and Watermarking of Multimedia Contents III, Ping Wah Wong and Edward J. Delp
III (Eds.), Vol. 4314. International Society for Optics and Photonics, SPIE, 555–564. DOI:https://doi.org/10.1117/12.
435440
[268] M. Niimi, T. Ei, H. Noda, E. Kawaguchi, and B. Segee. 2004. An attack to BPCS-steganography using complexity
histogram and countermeasure. In 2004 International Conference on Image Processing, 2004 (ICIP’04), Vol. 2. 733–736.
DOI:https://doi.org/10.1109/ICIP.2004.1419402
[269] M. Niimi, H. Noda, and B. Segee. 2005. A robust BPCS-steganography against the visual attack. In 2005 5th Interna-
tional Conference on Information Communications Signal Processing. 1116–1120. DOI:https://doi.org/10.1109/ICICS.
2005.1689227
[270] N. Nikolaidis and I. Pitas. 1998. Robust image watermarking in the spatial domain. Signal Processing 66, 3 (1998),
385–403. https://doi.org/10.1016/S0165- 1684(98)00017- 6
[271] Arooj Nissar and A. H. Mir. 2010. Classification of steganalysis techniques: A study. Digital Signal Processing: A
Review Journal 20, 6 (2010), 1758–1770. DOI:https://doi.org/10.1016/j.dsp.2010.02.003
[272] Hideki Noda, Michiharu Niimi, and Eiji Kawaguchi. 2006. High-performance JPEG steganography using quantiza-
tion index modulation in DCT domain. Pattern Recognition Letters 27, 5 (2006), 455–461. DOI:https://doi.org/10.1016/
j.patrec.2005.09.008
[273] Hideki Noda, Jeremiah Spaulding, Mahdad N. Shirazi, Michiharu Niimi, and Eiji Kawaguchi. 2003. Bit-plane decom-
position steganography combined with JPEG2000 compression. In International Workshop on Information Hiding,
Fabien A. P. Petitcolas (Ed.). Springer, Berlin, 295–309.
[274] T. Ogihara, D. Nakamura, and N. Yokoya. 1996. Data embedding into pictorial images with less distortion using
discrete cosine transform. In Proceedings of 13th International Conference on Pattern Recognition, Vol. 2. 675–679.
DOI:https://doi.org/10.1109/ICPR.1996.546908
[275] Hsiang-Kuang Pan, Yu-Yuan Chen, and Yu-Chee Tseng. 2000. A secure data hiding scheme for two-color images. In
Proceedings of the 5th IEEE Symposium on Computers and Communications (ISCC’00). 750–755. DOI:https://doi.org/
10.1109/ISCC.2000.860731
[276] Hari Mohan Pandey. 2020. Secure medical data transmission using a fusion of bit mask oriented genetic algorithm,
encryption and steganography. Future Generation Computer Systems 111 (2020), 213–225. DOI:https://doi.org/10.
1016/j.future.2020.04.034
[277] Constantinos Patsakis and Nikolaos Aroukatos. 2014. LSB and DCT steganographic detection using compressive
sensing. Journal of Information Hiding and Multimedia Signal Processing 5, 1 (2014), 20–32.
[278] Goutam Paul, Sanjoy Kumar Saha, Debanjan Burman, et al. 2020. A PVD based high capacity steganography
algorithm with embedding in non-sequential position. Multimedia Tools and Applications 79, 19 (2020), 13449–
13479.
[279] Tomáš Pevny, Patrick Bas, and Jessica Fridrich. 2010. Steganalysis by subtractive pixel adjacency matrix. IEEE Trans-
actions on Information Forensics and Security 5, 2 (2010), 215–224. https://doi.org/10.1109/TIFS.2010.2045842
[280] Tomáš Pevný, Tomáš Filler, and Patrick Bas. 2010. Using high-dimensional image models to perform highly un-
detectable steganography. In International Workshop on Information Hiding, Rainer Böhme, Philip W. L. Fong, and
Reihaneh Safavi-Naini (Eds.). Springer, Berlin, 161–177.
[281] Tomáš Pevný and Jessica Fridrich. 2006. Multi-class blind steganalysis for JPEG images. In Security, Steganography,
and Watermarking of Multimedia Contents VIII, Edward J. Delp III and Ping Wah Wong (Eds.), Vol. 6072. International
Society for Optics and Photonics, SPIE, 60720O. DOI:https://doi.org/10.1117/12.640943
[282] Tomas Pevny and Jessica Fridrich. 2007. Merging Markov and DCT features for multi-class JPEG steganalysis. In
Security, Steganography, and Watermarking of Multimedia Contents IX, Edward J. Delp III and Ping Wah Wong (Eds.),
Vol. 6505. International Society for Optics and Photonics, SPIE, 650503. DOI:https://doi.org/10.1117/12.696774
[283] Tomas Pevny and Jessica Fridrich. 2007. Merging Markov and DCT features for multi-class JPEG steganalysis. In
Security, Steganography, and Watermarking of Multimedia Contents IX, Edward J. Delp III and Ping Wah Wong (Eds.),
Vol. 6505. International Society for Optics and Photonics, SPIE, 650503. https://doi.org/10.1117/12.696774
[284] Tomas Pevny and Jessica Fridrich. 2008. Detection of double-compression in JPEG images for applications in
steganography. IEEE Transactions on Information Forensics and Security 3, 2 (2008), 247–258.
[285] Tomas Pevny and Jessica Fridrich. 2008. Multiclass detector of current steganographic methods for JPEG format.
IEEE Transactions on Information Forensics and Security 3, 4 (2008), 635–650. DOI:https://doi.org/10.1109/TIFS.2008.
2002936

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:33

[286] Tomáš Pevný and Jessica Fridrich. 2008. Novelty detection in blind steganalysis. In Proceedings of the 10th ACM
Workshop on Multimedia and Security (MM&Sec’08). Association for Computing Machinery, New York, NY, USA,
167–176. DOI:https://doi.org/10.1145/1411328.1411357
[287] Tomáš Pevny, Jessica Fridrich, and Andrew D. Ker. 2012. From blind to quantitative steganalysis. IEEE Transactions
on Information Forensics and Security 7, 2 (2012), 445–454. DOI:https://doi.org/10.1109/TIFS.2011.2175918
[288] Tomáš Pevný and Andrew D. Ker. 2013. The challenges of rich features in universal steganalysis. In Media Water-
marking, Security, and Forensics 2013, Adnan M. Alattar, Nasir D. Memon, and Chad D. Heitzenrater (Eds.), Vol. 8665.
International Society for Optics and Photonics, SPIE, 86650M. DOI:https://doi.org/10.1117/12.2006790
[289] A. Piva, M. Barni, F. Bartolini, and V. Cappellini. 1997. DCT-based watermark recovering without resorting to the
uncorrupted original image. In Proceedings of International Conference on Image Processing, Vol. 1. 520–523. https:
//doi.org/10.1109/ICIP.1997.647964
[290] Mohammad Rasoul PourArian and Ali Hanani. 2016. Blind steganography in color images by double wavelet trans-
form and improved Arnold transform. Indonesian Journal of Electrical Engineering and Computer Science 3, 3 (2016),
586–600.
[291] Anita Pradhan, K. Raja Sekhar, and Gandharba Swain. 2017. Adaptive PVD steganography using horizontal, vertical,
and diagonal edges in six-pixel blocks. Security and Communication Networks 2017 (2017), 1–13.
[292] Sabyasachi Pramanik. 2023. An adaptive image steganography approach depending on integer wavelet transform
and genetic algorithm. Multimedia Tools and Applications 82, 22 (2023), 34287–34319.
[293] Sabyasachi Pramanik and S. Suresh Raja. 2020. A secured image steganography using genetic algorithm. Advances
in Mathematics: Scientific Journal 9, 7 (2020), 4533–4541.
[294] Sabyasachi Pramanik, R. P. Singh, and Ramkrishna Ghosh. 2020. Application of bi-orthogonal wavelet transform
and genetic algorithm in image steganography. Multimedia Tools and Applications 79 (2020), 17463–17482.
[295] Niels Provos. 2001. Defending against statistical steganalysis. In 10th USENIX Security Symposium (USENIX
Security’01).
[296] Niels Provos and Peter Honeyman. 2001. Detecting Steganographic Content on the Internet. Technical Report. Center
for Information Technology Integration.
[297] N. Provos and P. Honeyman. 2003. Hide and seek: An introduction to steganography. IEEE Security & Privacy 1,
3 (2003), 32–44. DOI:https://doi.org/10.1109/MSECP.2003.1203220
[298] Yinlong Qian, Jing Dong, Wei Wang, and Tieniu Tan. 2015. Deep learning for steganalysis via convolutional neural
networks. In Media Watermarking, Security, and Forensics 2015, Adnan M. Alattar, Nasir D. Memon, and Chad D.
Heitzenrater (Eds.), Vol. 9409. International Society for Optics and Photonics, SPIE, 94090J. DOI:https://doi.org/10.
1117/12.2083479
[299] Yinlong Qian, Jing Dong, Wei Wang, and Tieniu Tan. 2016. Learning and transferring representations for image
steganalysis using convolutional neural network. In 2016 IEEE International Conference on Image Processing (ICIP’16).
2752–2756. DOI:https://doi.org/10.1109/ICIP.2016.7532860
[300] Yinlong Qian, Jing Dong, Wei Wang, and Tieniu Tan. 2018. Feature learning for steganalysis using convolutional
neural networks. Multimedia Tools and Applications 77, 15 (2018), 19633–19657.
[301] Tong Qiao, Xiangyang Luo, Ting Wu, Ming Xu, and Zhenxing Qian. 2021. Adaptive steganalysis based on statistical
model of quantized DCT coefficients for JPEG images. IEEE Transactions on Dependable and Secure Computing 18,
6 (2021), 2736–2751. DOI:https://doi.org/10.1109/TDSC.2019.2962672
[302] Tong Qiao, Florent Retraint, Rémi Cogranne, and Cathel Zitzmann. 2015. Steganalysis of JSteg algorithm using
hypothesis testing theory. EURASIP Journal on Information Security 2015, 1 (2015), 1–16.
[303] Jiaohua Qin, Yuanjing Luo, Xuyu Xiang, Yun Tan, and Huajun Huang. 2019. Coverless image steganography: A
survey. IEEE Access 7 (2019), 171372–171394. DOI:https://doi.org/10.1109/ACCESS.2019.2955452
[304] Tamer Rabie and Ibrahim Kamel. 2017. High-capacity steganography: A global-adaptive-region discrete cosine trans-
form approach. Multimedia Tools and Applications 76, 5 (2017), 6473–6493.
[305] Tamer Rabie and Ibrahim Kamel. 2017. Toward optimal embedding capacity for transform domain steganography:
A quad-tree adaptive-region approach. Multimedia Tools and Applications 76, 6 (2017), 8627–8650.
[306] Eko Hari Rachmawanto, et al. 2019. An improved security and message capacity using AES and Huffman coding on
image steganography. TELKOMNIKA Telecommunication, Computing, Electronics and Control 17, 5 (2019), 2400–2409.
[307] K. B. Raja, C. R. Chowdary, K. R. Venugopal, and L. M. Patnaik. 2005. A secure image steganography using LSB,
DCT and compression techniques on raw images. In 2005 3rd International Conference on Intelligent Sensing and
Information Processing. 170–176. DOI:https://doi.org/10.1109/ICISIP.2005.1619431
[308] Mritha Ramalingam and Nor Ashidi Mat Isa. 2016. A data-hiding technique using scene-change detection for video
steganography. Computers & Electrical Engineering 54 (2016), 423–434. DOI:https://doi.org/10.1016/j.compeleceng.
2015.10.005

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:34 M. H. Kombrink et al.

[309] GMK Ramani, E. V. Prasad, S. Varadarajan, Tirupati SVUCE, and Kakinada JN T UCE. 2007. Steganography using
BPCS to the integer wavelet transformed image. International Journal of Computer Science and Network Security 7,
7 (2007), 293–302.
[310] Mahendra Rathor and Anirban Sengupta. 2020. Design flow of secured N-Point DFT application specific processor
using obfuscation and steganography. IEEE Letters of the Computer Society 3, 1 (2020), 13–16. DOI:https://doi.org/
10.1109/LOCS.2020.2973586
[311] Biswarup Ray, Souradeep Mukhopadhyay, Sabbir Hossain, Sudipta Kr Ghosal, and Ram Sarkar. 2021. Image
steganography using deep learning based edge detection. Multimedia Tools and Applications 80, 24 (2021), 33475–
33503.
[312] H. S. Manjunatha Reddy and K. B. Raja. 2011. Wavelet based non LSB steganography. International Journal of Ad-
vanced Networking and Applications 3, 3 (2011), 1203.
[313] Tabares-Soto Reinel, Arteaga-Arteaga Harold Brayan, Bravo-Ortiz Mario Alejandro, Mora-Rubio Alejandro, Arias-
Garzón Daniel, Alzate-Grisales Jesús Alejandro, Burbano-Jacome Alejandro Buenaventura, Orozco-Arias Simon,
Isaza Gustavo, and Ramos-Pollán Raúl. 2021. GBRAS-Net: A convolutional neural network architecture for spatial
image steganalysis. IEEE Access 9 (2021), 14340–14350. DOI:https://doi.org/10.1109/ACCESS.2021.3052494
[314] Tabares-Soto Reinel, Ramos-Pollán Raúl, and Isaza Gustavo. 2019. Deep learning applied to steganalysis of digital
images: A systematic review. IEEE Access 7 (2019), 68970–68990. DOI:https://doi.org/10.1109/ACCESS.2019.2918086
[315] Supriadi Rustad, De Rosal Ignatius Moses Setiadi, Abdul Syukur, and Pulung Nurtantio Andono. 2022. Inverted LSB
image steganography using adaptive pattern to improve imperceptibility. Journal of King Saud University, Computer
and Information Sciences 34, 6, Part B (2022), 3559–3568. DOI:https://doi.org/10.1016/j.jksuci.2020.12.017
[316] Vajiheh Sabeti, Masomeh Sobhani, and Seyed Mohammad Hossein Hasheminejad. 2022. An adaptive image
steganography method based on integer wavelet transform using genetic algorithm. Computers and Electrical Engi-
neering 99 (2022), 107809. DOI:https://doi.org/10.1016/j.compeleceng.2022.107809
[317] Vasiliy Sachnev, Hyoung Joong Kim, and Rongyue Zhang. 2009. Less detectable JPEG steganography method based
on heuristic optimization and BCH syndrome coding. In Proceedings of the 11th ACM Workshop on Multimedia and
Security (MM&Sec’09). Association for Computing Machinery, New York, NY, USA, 131–140. DOI:https://doi.org/10.
1145/1597817.1597841
[318] Aditya Kumar Sahu and Monalisa Sahu. 2020. Digital image steganography and steganalysis: A journey of the past
three decades. Open Computer Science 10, 1 (2020), 296–342. DOI:https://doi.org/doi:10.1515/comp- 2020- 0136
[319] Aditya Kumar Sahu and Monalisa Sahu. 2020. Digital image steganography and steganalysis: A journey of the past
three decades. Open Computer Science 10, 1 (2020), 296–342.
[320] Aditya Kumar Sahu and Gandharba Swain. 2019. An optimal information hiding approach based on pixel value
differencing and modulus function. Wireless Personal Communications 108, 1 (2019), 159–174.
[321] Aditya Kumar Sahu and Gandharba Swain. 2020. Reversible image steganography using dual-layer LSB matching.
Sensing and Imaging 21, 1 (2020), 1–21.
[322] Aditya Kumar Sahu, Gandharba Swain, Monalisa Sahu, and J. Hemalatha. 2021. Multi-directional block based PVD
and modulus function image steganography to avoid FOBP and IEP. Journal of Information Security and Applications
58 (2021), 102808. DOI:https://doi.org/10.1016/j.jisa.2021.102808
[323] Marwa Saidi, Houcemeddine Hermassi, Rhouma Rhouma, and Safya Belghith. 2017. A new adaptive image steganog-
raphy scheme based on DCT and chaotic map. Multimedia Tools and Applications 76, 11 (2017), 13493–13510.
[324] Hedieh Sajedi. 2016. Steganalysis based on steganography pattern discovery. Journal of Information Security and
Applications 30 (2016), 3–14. DOI:https://doi.org/10.1016/j.jisa.2016.04.001
[325] Phil Sallee. 2004. Model-based steganography. In Digital Watermarking, Ton Kalker, Ingemar Cox, and Yong Man
Ro (Eds.). Springer, Berlin, 154–167.
[326] Phil Sallee. 2005. Model-based methods for steganography and steganalysis. International Journal of Image and
Graphics 5, 1 (2005), 167–189.
[327] Wellia Shinta Sari, Eko Hari Rachmawanto, D. R. I. M. Setiadi, and Christy Atika Sari. 2017. A good performance OTP
encryption image based on DCT-DWT steganography. TELKOMNIKA Telecommunication, Computing, Electronics
and Control 15, 4 (2017), 1987–1995.
[328] Vahid Sedighi, Rémi Cogranne, and Jessica Fridrich. 2015. Content-adaptive steganography by minimizing statistical
detectability. IEEE Transactions on Information Forensics and Security 11, 2 (2015), 221–234.
[329] De Rosal Igantius Moses Setiadi. 2021. PSNR vs SSIM: Imperceptibility quality assessment for image steganography.
Multimedia Tools and Applications 80, 6 (2021), 8423–8444.
[330] De Rosal Ignatius Moses Setiadi, Supriadi Rustad, Pulung Nurtantio Andono, and Guruh Fajar Shidik. 2023. Dig-
ital image steganography survey and investigation (goal, assessment, method, development, and dataset). Signal
Processing 206 (2023), 108908. DOI:https://doi.org/10.1016/j.sigpro.2022.108908

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:35

[331] Pratik D. Shah and Rajankumar S. Bichkar. 2021. Secret data modification based image steganography technique
using genetic algorithm having a flexible chromosome structure. International Journal of Engineering Science and
Technology 24, 3 (2021), 782–794.
[332] Deepa D. Shankar and Vinod Kumar Shukla. 2018. Result analysis of cross-validation on low embedding feature-
based blind steganalysis of 25 percent on JPEG images using SVM. In 2018 International Conference on Circuits and
Systems in Digital Enterprise Technology (ICCSDET’18). 1–5. DOI:https://doi.org/10.1109/ICCSDET.2018.8821059
[333] Mehdi Sharifzadeh, Mohammed Aloraini, and Dan Schonfeld. 2020. Adaptive batch size image merging steganogra-
phy and quantized gaussian image steganography. IEEE Transactions on Information Forensics and Security 15 (2020),
867–879. DOI:https://doi.org/10.1109/TIFS.2019.2929441
[334] Kartik Sharma, Ashutosh Aggarwal, Tanay Singhania, Deepak Gupta, and Ashish Khanna. 2019. Hiding data in
images using cryptography and deep neural network. Journal of Artificial Intelligence and Systems 1, 1 (2019), 143–
162. DOI:https://doi.org/10.33969/ais.2019.11009
[335] Toby Sharp. 2001. An implementation of key-based digital signal steganography. In International Workshop on In-
formation Hiding, Ira S. Moskowitz (Ed.). Springer, Berlin, 13–26.
[336] Mansour Sheikhan, Mansoureh Pezhmanpour, and M. Shahram Moin. 2012. Improved contourlet-based steganalysis
using binary particle swarm optimization and radial basis neural networks. Neural Computing and Applications 21,
7 (2012), 1717–1728.
[337] Haichao Shi, Jing Dong, Wei Wang, Yinlong Qian, and Xiaoyu Zhang. 2017. SSGAN: Secure steganography based
on generative adversarial networks. In Pacific Rim Conference on Multimedia. Springer, 534–544.
[338] Y. Q. Shi, Guorong Xuan, D. Zou, Jianjiong Gao, Chengyun Yang, Zhenping Zhang, Peiqi Chai, W. Chen, and C. Chen.
2005. Image steganalysis based on moments of characteristic functions using wavelet decomposition, prediction-
error image, and neural network. In 2005 IEEE International Conference on Multimedia and Expo. 4 pp. DOI:https:
//doi.org/10.1109/ICME.2005.1521412
[339] Yun Q. Shi, Chunhua Chen, and Wen Chen. 2007. A Markov process based approach to effective attacking JPEG
steganography. In International Workshop on Information Hiding, Jan L. Camenisch, Christian S. Collberg, Neil F.
Johnson, and Phil Sallee (Eds.). Springer, Berlin, 249–264.
[340] Saman Shojae Chaeikar and Ali Ahmadi. 2019. Ensemble SW image steganalysis: A low dimension method for LSBR
detection. Signal Processing: Image Communication 70 (2019), 233–245. DOI:https://doi.org/10.1016/j.image.2018.10.
004
[341] Saman Shojae Chaeikar, Mazdak Zamani, Azizah Bt Abdul Manaf, and Akram M. Zeki. 2018. PSW statistical LSB
image steganalysis. Multimedia Tools and Applications 77, 1 (2018), 805–835.
[342] Awdhesh K. Shukla, Akanksha Singh, Balvinder Singh, and Amod Kumar. 2018. A secure and high-capacity data-
hiding method using compression, encryption and optimized pixel value differencing. IEEE Access 6 (2018), 51130–
51139. DOI:https://doi.org/10.1109/ACCESS.2018.2868192
[343] Amritpal Singh and Harpal Singh. 2015. An improved LSB based image steganography technique for RGB images.
In 2015 IEEE International Conference on Electrical, Computer and Communication Technologies (ICECCT’15). 1–4.
DOI:https://doi.org/10.1109/ICECCT.2015.7226122
[344] Anuradha Singhal and Punam Bedi. 2021. Multi-class blind steganalysis using deep residual networks. Multimedia
Tools and Applications 80, 9 (2021), 13931–13956.
[345] Deepak Singla and Rupali Syal. 2012. Data security using LSB & DCT steganography in images. International Journal
of Computational Engineering Research 2, 2 (2012), 359–364.
[346] Kaushal Solanki, Anindya Sarkar, and B. S. Manjunath. 2007. YASS: Yet another steganographic scheme that resists
blind steganalysis. In International Workshop on Information Hiding, Teddy Furon, François Cayre, Gwenaël Doërr,
and Patrick Bas (Eds.). Springer, Berlin, 16–31.
[347] Xiaofeng Song, Fenlin Liu, Chunfang Yang, Xiangyang Luo, and Yi Zhang. 2015. Steganalysis of adaptive JPEG
steganography using 2D gabor filters. In Proceedings of the 3rd ACM Workshop on Information Hiding and Multimedia
Security (IH&MMSec’15). Association for Computing Machinery, New York, NY, USA, 15–23. DOI:https://doi.org/
10.1145/2756601.2756608
[348] Xianhua Song, Shen Wang, and Xiamu Niu. 2012. An integer DCT and affine transformation based image steganogra-
phy method. In 2012 8th International Conference on Intelligent Information Hiding and Multimedia Signal Processing.
102–105. DOI:https://doi.org/10.1109/IIH-MSP.2012.30
[349] Neetika Soni, Indu Saini, and Butta Singh. 2021. An integer wavelet transform and pixel value differencing based
feature specific hybrid technique for 2D ECG steganography with high payload capacity. Multimedia Tools and
Applications 80, 6 (2021), 8505–8540.
[350] D. Soukal, J. Fridrich, and M. Goljan. 2005. Maximum likelihood estimation of secret message length embedded
using±k steganography in spatial domain. In Proceedings SPIE, Electronic Imaging, Security, Steganography, and Wa-
termarking of Multimedia Contents VII, Vol. 5681. 595–606.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:36 M. H. Kombrink et al.

[351] Jeremiah Spaulding, Hideki Noda, Mahdad N. Shirazi, and Eiji Kawaguchi. 2002. BPCS steganography using EZW
lossy compressed images. Pattern Recognition Letters 23, 13 (2002), 1579–1587. DOI:https://doi.org/10.1016/S0167-
8655(02)00122-8
[352] Mansi S. Subhedar and Vijay H. Mankar. 2014. Current status and key issues in image steganography: A survey.
Computer Science Review 13–14 (2014), 95–113. DOI:https://doi.org/10.1016/j.cosrev.2014.09.001
[353] Nandhini Subramanian, Ismahane Cheheb, Omar Elharrouss, Somaya Al-Maadeed, and Ahmed Bouridane. 2021.
End-to-end image steganography using deep convolutional autoencoders. IEEE Access 9 (2021), 135585–135593.
DOI:https://doi.org/10.1109/ACCESS.2021.3113953
[354] Nandhini Subramanian, Omar Elharrouss, Somaya Al-Maadeed, and Ahmed Bouridane. 2021. Image steganogra-
phy: A review of the recent advances. IEEE Access 9 (2021), 23409–23423. DOI:https://doi.org/10.1109/ACCESS.2021.
3053998
[355] K. Sullivan, U. Madhow, S. Chandrasekaran, and B. S. Manjunath. 2006. Steganalysis for Markov cover data with
applications to images. IEEE Transactions on Information Forensics and Security 1, 2 (2006), 275–287. DOI:https://doi.
org/10.1109/TIFS.2006.873595
[356] Kenneth Sullivan, Upamanyu Madhow, Shivkumar Chandrasekaran, and Bangalore S. Manjunath. 2005. Steganalysis
of spread spectrum data hiding exploiting cover memory. In Security, Steganography, and Watermarking of Multi-
media Contents VII, Edward J. Delp III and Ping W. Wong (Eds.), Vol. 5681. International Society for Optics and
Photonics, SPIE, 38–46. https://doi.org/10.1117/12.588121
[357] Shuliang Sun. 2015. A new information hiding method based on improved BPCS steganography. Advances in Mul-
tiMedia 2015, Article 5, (2015), 1 page. DOI:https://doi.org/10.1155/2015/698492
[358] Shuliang Sun. 2015. A new information hiding method based on improved BPCS steganography. Advances in Mul-
tiMedia 2015, Article 5 (Jan. 2015), 1 page. https://doi.org/10.1155/2015/698492
[359] Gandharba Swain. 2016. Adaptive pixel value differencing steganography using both vertical and horizontal edges.
Multimedia Tools and Applications 75, 21 (2016), 13541–13556.
[360] Théo Taburet, Louis Filstroff, Patrick Bas, and Wadih Sawaya. 2019. An empirical study of steganography and ste-
ganalysis of color images in the JPEG domain. In Digital Forensics and Watermarking, Chang D. Yoo, Yun-Qing Shi,
Hyoung Joong Kim, Alessandro Piva, and Gwangsu Kim (Eds.). Springer International Publishing, Cham, 290–303.
[361] Jingxuan Tan, Xin Liao, Jiate Liu, Yun Cao, and Hongbo Jiang. 2022. Channel attention image steganography
with generative adversarial networks. IEEE Transactions on Network Science and Engineering 9, 2 (2022), 888–903.
DOI:https://doi.org/10.1109/TNSE.2021.3139671
[362] Shunquan Tan, Jiwu Huang, and Yun Q. Shi. 2008. Steganalysis of enhanced BPCS steganography using the Hilbert-
Huang transform based sequential analysis. In Digital Watermarking, Yun Q. Shi, Hyoung-Joong Kim, and Stefan
Katzenbeisser (Eds.). Springer, Berlin, 112–126.
[363] Shunquan Tan and Bin Li. 2012. Targeted steganalysis of edge adaptive image steganography based on LSB matching
revisited using B-Spline fitting. IEEE Signal Processing Letters 19, 6 (2012), 336–339. DOI:https://doi.org/10.1109/LSP.
2012.2194702
[364] Matthew Tancik, Ben Mildenhall, and Ren Ng. 2020. StegaStamp: Invisible hyperlinks in physical photographs. In
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR’20).
[365] Weixuan Tang, Bin Li, Mauro Barni, Jin Li, and Jiwu Huang. 2021. An automatic cost learning framework for image
steganography using deep reinforcement learning. IEEE Transactions on Information Forensics and Security 16 (2021),
952–967. DOI:https://doi.org/10.1109/TIFS.2020.3025438
[366] Weixuan Tang, Bin Li, Weiqi Luo, and Jiwu Huang. 2016. Clustering steganographic modification directions for color
components. IEEE Signal Processing Letters 23, 2 (2016), 197–201. DOI:https://doi.org/10.1109/LSP.2015.2504583
[367] Weixuan Tang, Bin Li, Shunquan Tan, Mauro Barni, and Jiwu Huang. 2019. CNN-based adversarial embedding for
image steganography. IEEE Transactions on Information Forensics and Security 14, 8 (2019), 2074–2087. DOI:https:
//doi.org/10.1109/TIFS.2019.2891237
[368] Weixuan Tang, Haodong Li, Weiqi Luo, and Jiwu Huang. 2014. Adaptive steganalysis against WOW embedding
algorithm. In Proceedings of the 2nd ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’14).
Association for Computing Machinery, New York, NY, USA, 91–96. DOI:https://doi.org/10.1145/2600918.
2600935
[369] Weixuan Tang, Haodong Li, Weiqi Luo, and Jiwu Huang. 2016. Adaptive steganalysis based on embedding proba-
bilities of pixels. IEEE Transactions on Information Forensics and Security 11, 4 (2016), 734–745. DOI:https://doi.org/
10.1109/TIFS.2015.2507159
[370] Weixuan Tang, Shunquan Tan, Bin Li, and Jiwu Huang. 2017. Automatic steganographic distortion learning using a
generative adversarial network. IEEE Signal Processing Letters 24, 10 (2017), 1547–1551. DOI:https://doi.org/10.1109/
LSP.2017.2745572

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:37

[371] Thanh Hai Thai, Remi Cogranne, and Florent Retraint. 2014. Statistical model of quantized DCT coefficients: Ap-
plication in the steganalysis of Jsteg algorithm. IEEE Transactions on Image Processing 23, 5 (2014), 1980–1993.
DOI:https://doi.org/10.1109/TIP.2014.2310126
[372] K. Thangadurai and G. Sudha Devi. 2014. An analysis of LSB based image steganography techniques. In 2014 In-
ternational Conference on Computer Communication and Informatics. 1–4. DOI:https://doi.org/10.1109/ICCCI.2014.
6921751
[373] Rohit Thanki and Surekha Borra. 2018. A color image steganography in hybrid FRT–DWT domain. Journal of In-
formation Security and Applications 40 (2018), 92–102. DOI:https://doi.org/10.1016/j.jisa.2018.03.004
[374] Clement Fuji Tsang and Jessica Fridrich. 2018. Steganalyzing images of arbitrary size with CNNs. Electronic Imaging
30 (2018), 1–8.
[375] Yu-Chee Tseng, Yu-Yuan Chen, and Hsiang-Kuang Pan. 2002. A secure data hiding scheme for binary images. IEEE
Transactions on Communications 50, 8 (2002), 1227–1231. DOI:https://doi.org/10.1109/TCOMM.2002.801488
[376] Yu-Chee Tseng and Hsiang-Kuang Pan. 2001. Secure and invisible data hiding in 2-color images. In Proceedings of the
2001 IEEE Conference on Computer Communications (INFOCOM’01) (Cat. No.01CH37213), Vol. 2. 887–896. DOI:https:
//doi.org/10.1109/INFCOM.2001.916280
[377] Milad Yousefi Valandar, Milad Jafari Barani, Peyman Ayubi, and Maryam Aghazadeh. 2019. An integer wavelet
transform image steganography method based on 3D sine chaotic map. Multimedia Tools and Applications 78,
8 (2019), 9971–9989.
[378] S. T. Veena and S. Arivazhagan. 2018. Quantitative steganalysis of spatial LSB based stego images using reduced
instances and features. Pattern Recognition Letters 105 (2018), 39–49. DOI:https://doi.org/10.1016/j.patrec.2017.08.
016. Machine Learning and Applications in Artificial Intelligence.
[379] Denis Volkhonskiy, Ivan Nazarov, and Evgeny Burnaev. 2020. Steganographic generative adversarial networks. In
12th International Conference on Machine Vision (ICMV’19), Wolfgang Osten and Dmitry P. Nikolaev (Eds.), Vol.
11433. International Society for Optics and Photonics, SPIE, 991–1005. DOI:https://doi.org/10.1117/12.2559429
[380] Dr. Ekta Walia, Payal Jain, and Navdeep Navdeep. 2010. An analysis of LSB & DCT based steganography. Global Jour-
nal of Computer Science and Technology 10 (2010), 4–8. https://computerresearch.org/index.php/computer/article/
view/912
[381] Chang Wang and Jiangqun Ni. 2012. An efficient JPEG steganographic scheme based on the block entropy of DCT
coefficients. In 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP’12). 1785–1788.
DOI:https://doi.org/10.1109/ICASSP.2012.6288246
[382] Chung-Ming Wang, Nan-I. Wu, Chwei-Shyong Tsai, and Min-Shiang Hwang. 2008. A high quality steganographic
method with pixel-value differencing and modulus function. Journal of Systems and Software 81, 1 (2008), 150–158.
DOI:https://doi.org/10.1016/j.jss.2007.01.049
[383] Shen Wang, Bian Yang, and Xiamu Niu. 2010. A secure steganography method based on genetic algorithm. Journal
of Information Hiding and Multimedia Signal Processing 1, 1 (2010), 28–35.
[384] Ying Wang and Pierre Moulin. 2007. Optimized feature extraction for learning-based image steganalysis. IEEE Trans-
actions on Information Forensics and Security 2, 1 (2007), 31–45. DOI:https://doi.org/10.1109/TIFS.2006.890517
[385] Yanting Wang, Mingwei Tang, and Zhen Wang. 2020. High-capacity adaptive steganography based on LSB and
Hamming code. Optik 213 (2020), 164685. DOI:https://doi.org/10.1016/j.ijleo.2020.164685
[386] Yaofei Wang, Weiming Zhang, Weixiang Li, Xinzhi Yu, and Nenghai Yu. 2020. Non-additive cost functions for color
image steganography based on inter-channel correlations and differences. IEEE Transactions on Information Forensics
and Security 15 (2020), 2081–2095. DOI:https://doi.org/10.1109/TIFS.2019.2956590
[387] Mohd Arif Wani and Bisma Sultan. 2023. Deep learning based image steganography: A review. Wiley Interdisci-
plinary Reviews: Data Mining and Knowledge Discovery 13, 3 (2023), e1481.
[388] Ranyiah Wazirali, Waleed Alasmary, Mohamed M. E. A. Mahmoud, and Ahmad Alhindi. 2019. An optimized
steganography hiding capacity and imperceptibly using genetic algorithms. IEEE Access 7 (2019), 133496–133508.
DOI:https://doi.org/10.1109/ACCESS.2019.2941440
[389] Kangkang Wei, Weiqi Luo, Shunquan Tan, and Jiwu Huang. 2022. Universal deep network for steganalysis of color
image based on channel representation. IEEE Transactions on Information Forensics and Security 17 (2022), 3022–3036.
DOI:https://doi.org/10.1109/TIFS.2022.3196265
[390] Ping Wei, Sheng Li, Xinpeng Zhang, Ge Luo, Zhenxing Qian, and Qing Zhou. 2022. Generative steganography
network. In Proceedings of the 30th ACM International Conference on Multimedia (MM’22). Association for Computing
Machinery, New York, NY, USA, 1621–1629. DOI:https://doi.org/10.1145/3503161.3548217
[391] Shaowei Weng, Mengfei Chen, Lifang Yu, and Shiyao Sun. 2022. Lightweight and effective deep image steganalysis
network. IEEE Signal Processing Letters 29 (2022), 1888–1892. DOI:https://doi.org/10.1109/LSP.2022.3201727
[392] Andreas Westfeld. 2001. F5—A steganographic algorithm. In International Workshop on Information Hiding, Ira S.
Moskowitz (Ed.). Springer, Berlin, 289–302.

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:38 M. H. Kombrink et al.

[393] Andreas Westfeld. 2003. Detecting low embedding rates. In International Workshop on Information Hiding, Fabien A.
P. Petitcolas (Ed.). Springer, Berlin, 324–339.
[394] Andreas Westfeld. 2008. Generic adoption of spatial steganalysis to transformed domain. In International Workshop
on Information Hiding, Kaushal Solanki, Kenneth Sullivan, and Upamanyu Madhow (Eds.). Springer, Berlin, 161–177.
[395] Andreas Westfeld and Andreas Pfitzmann. 2000. Attacks on steganographic systems. In International Workshop on
Information Hiding, Andreas Pfitzmann (Ed.). Springer, Berlin, 61–76.
[396] Da-Chun Wu and Wen-Hsiang Tsai. 2003. A steganographic method for images by pixel-value differencing. Pattern
Recognition Letters 24, 9 (2003), 1613–1626. DOI:https://doi.org/10.1016/S0167- 8655(02)00402- 6
[397] H.-C. Wu, N.-I. Wu, C.-S. Tsai, and M.-S. Hwang. 2005. Image steganographic scheme based on pixel-value differenc-
ing and LSB replacement methods. IEEE Proceedings—Vision, Image and Signal Processing 152, 5 (Oct. 2005), 611–615.
https://digital- library.theiet.org/content/journals/10.1049/ip- vis_20059022
[398] Min Wu, E. Tang, and B. Lin. 2000. Data hiding in digital binary image. In Proceedings of the 2000 IEEE International
Conference on Multimedia and Expo (ICME’00). Latest Advances in the Fast Changing World of Multimedia (Cat. No.
00TH8532), Vol. 1. 393–396. DOI:https://doi.org/10.1109/ICME.2000.869623
[399] Pin Wu, Yang Yang, and Xiaoqiang Li. 2018. StegNet: Mega image steganography capacity with deep convolutional
network. Future Internet 10, 6 (2018), 1–15. DOI:https://doi.org/10.3390/fi10060054
[400] Songtao Wu, Shenghua Zhong, and Yan Liu. 2018. Deep residual learning for image steganalysis. Multimedia Tools
and Applications 77, 9 (2018), 10437–10453.
[401] Songtao Wu, Sheng-hua Zhong, and Yan Liu. 2020. A novel convolutional neural network for image steganalysis
with shared normalization. IEEE Transactions on Multimedia 22, 1 (2020), 256–270. DOI:https://doi.org/10.1109/TMM.
2019.2920605
[402] Zhihua Xia, Xinhui Wang, Xingming Sun, Quansheng Liu, and Naixue Xiong. 2016. Steganalysis of LSB matching
using differences between nonadjacent pixels. Multimedia Tools and Applications 75, 4 (2016), 1947–1962.
[403] Zhihua Xia, Xinhui Wang, Xingming Sun, and Baowei Wang. 2014. Steganalysis of least significant bit matching
using multi-order differences. Security and Communication Networks 7, 8 (2014), 1283–1291. DOI:https://doi.org/10.
1002/sec.864 arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/sec.864
[404] Guoliang Xie, Jinchang Ren, Stephen Marshall, Huimin Zhao, and Huihui Li. 2021. A new cost function for spatial
image steganography based on 2D-SSA and WMF. IEEE Access 9 (2021), 30604–30614. DOI:https://doi.org/10.1109/
ACCESS.2021.3059690
[405] Guoliang Xie, Jinchang Ren, Stephen Marshall, Huimin Zhao, Rui Li, and Rongjun Chen. 2023. Self-attention en-
hanced deep residual network for spatial image steganalysis. Digital Signal Processing 139 (2023), 104063. DOI:https:
//doi.org/10.1016/j.dsp.2023.104063
[406] Guanshuo Xu. 2017. Deep convolutional neural network to detect J-UNIWARD. In Proceedings of the 5th ACM Work-
shop on Information Hiding and Multimedia Security (IH&MMSec’17). Association for Computing Machinery, New
York, NY, USA, 67–73. DOI:https://doi.org/10.1145/3082031.3083236
[407] Guanshuo Xu, Han-Zhou Wu, and Yun Q. Shi. 2016. Ensemble of CNNs for steganalysis: An empirical study. In
Proceedings of the 4th ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’16). Association
for Computing Machinery, New York, NY, USA, 103–107. DOI:https://doi.org/10.1145/2909827.2930798
[408] Guanshuo Xu, Han-Zhou Wu, and Yun-Qing Shi. 2016. Structural design of convolutional neural networks for ste-
ganalysis. IEEE Signal Processing Letters 23, 5 (2016), 708–712. DOI:https://doi.org/10.1109/LSP.2016.2548421
[409] Youmin Xu, Chong Mou, Yujie Hu, Jingfen Xie, and Jian Zhang. 2022. Robust invertible image steganography. In
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR’22). 7875–7884.
[410] Guorong Xuan, Yun Q. Shi, Jianjiong Gao, Dekun Zou, Chengyun Yang, Zhenping Zhang, Peiqi Chai, Chunhua Chen,
and Wen Chen. 2005. Steganalysis based on multiple features formed by statistical moments of wavelet character-
istic functions. In International Workshop on Information Hiding, Mauro Barni, Jordi Herrera-Joancomartí, Stefan
Katzenbeisser, and Fernando Pérez-González (Eds.). Springer, Berlin, 262–277.
[411] Guorong Xuan, Yun Q. Shi, Cong Huang, Dongdong Fu, Xiuming Zhu, Peiqi Chai, and Jianjiong Gao. 2006. Steganal-
ysis using high-dimensional features derived from co-occurrence matrix and class-wise non-principal components
analysis (CNPCA). In Digital Watermarking, Yun Qing Shi and Byeungwoo Jeon (Eds.). Springer, Berlin, 49–60.
[412] Cheng-Hsing Yang, Chi-Yao Weng, Hao-Kuan Tso, and Shiuh-Jeng Wang. 2011. A data hiding scheme using the
varieties of pixel-value differencing in multimedia images. Journal of Systems and Software 84, 4 (2011), 669–678.
DOI:https://doi.org/10.1016/j.jss.2010.11.889. The 9th International Conference on Quality Software.
[413] Cheng-Hsing Yang, Chi-Yao Weng, Shiuh-Jeng Wang, and Hung-Min Sun. 2008. Adaptive data hiding in edge areas
of images with spatial LSB domain systems. IEEE Transactions on Information Forensics and Security 3, 3 (2008),
488–497. DOI:https://doi.org/10.1109/TIFS.2008.926097
[414] Hongwei Yang, Hui He, Weizhe Zhang, and Xiaochun Cao. 2021. FedSteg: A federated transfer learning frame-
work for secure image steganalysis. IEEE Transactions on Network Science and Engineering 8, 2 (2021), 1084–1094.
DOI:https://doi.org/10.1109/TNSE.2020.2996612

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
Image Steganography Approaches and Their Detection Strategies: A Survey 33:39

[415] Jianhua Yang, Danyang Ruan, Jiwu Huang, Xiangui Kang, and Yun-Qing Shi. 2020. An embedding cost learning
framework using GAN. IEEE Transactions on Information Forensics and Security 15 (2020), 839–851. DOI:https://doi.
org/10.1109/TIFS.2019.2922229
[416] Yang Yang, Genyan Huang, Tianrui Zou, and Weiming Zhang. 2021. A novel reversible data hiding based on adaptive
block-partition and payload-allocation method. IET Image Processing 15 (2021), 57–585.
[417] Jian Ye, Jiangqun Ni, and Yang Yi. 2017. Deep learning hierarchical representations for image steganalysis. IEEE
Transactions on Information Forensics and Security 12, 11 (2017), 2545–2557. DOI:https://doi.org/10.1109/TIFS.2017.
2710946
[418] Mehdi Yedroudj, Frédéric Comby, and Marc Chaumont. 2018. Yedroudj-Net: An efficient CNN for spatial ste-
ganalysis. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP’18). 2092–2096.
DOI:https://doi.org/10.1109/ICASSP.2018.8461438
[419] Weike You, Hong Zhang, and Xianfeng Zhao. 2021. A Siamese CNN for image steganalysis. IEEE Transactions on
Information Forensics and Security 16 (2021), 291–306. DOI:https://doi.org/10.1109/TIFS.2020.3013204
[420] Yassine Yousfi, Jan Butora, Jessica Fridrich, and Clément Fuji Tsang. 2021. Improving EfficientNet for JPEG ste-
ganalysis. In Proceedings of the 2021 ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec’21).
Association for Computing Machinery, New York, NY, USA, 149–157. DOI:https://doi.org/10.1145/3437880.3460397
[421] Yassine Yousfi, Jan Butora, Jessica Fridrich, and Quentin Giboulot. 2019. Breaking ALASKA: Color separation for
steganalysis in JPEG domain. In Proceedings of the ACM Workshop on Information Hiding and Multimedia Security
(IH&MMSec’19). Association for Computing Machinery, New York, NY, USA, 138–149. DOI:https://doi.org/10.1145/
3335203.3335727
[422] Yassine Yousfi, Jan Butora, Eugene Khvedchenya, and Jessica Fridrich. 2020. ImageNet Pre-trained CNNs for JPEG
steganalysis. In 2020 IEEE International Workshop on Information Forensics and Security (WIFS’20). 1–6. DOI:https:
//doi.org/10.1109/WIFS49906.2020.9360897
[423] Yassine Yousfi and Jessica Fridrich. 2020. An intriguing struggle of CNNs in JPEG steganalysis and the OneHot
solution. IEEE Signal Processing Letters 27 (2020), 830–834. DOI:https://doi.org/10.1109/LSP.2020.2993959
[424] Jiang Yu, Fengyong Li, Hang Cheng, and Xinpeng Zhang. 2016. Spatial steganalysis using contrast of residuals. IEEE
Signal Processing Letters 23, 7 (2016), 989–992. DOI:https://doi.org/10.1109/LSP.2016.2575100
[425] Jiwen Yu, Xuanyu Zhang, Youmin Xu, and Jian Zhang. 2023. CRoSS: Diffusion model makes controllable, ro-
bust and secure image steganography. In Advances in Neural Information Processing Systems, A. Oh, T. Neumann,
A. Globerson, K. Saenko, M. Hardt, and S. Levine (Eds.), Vol. 36. Curran Associates, Inc., 80730–80743. https:
//proceedings.neurips.cc/paper_files/paper/2023/file/ff99390b6e942f b1dd7023f787f b0a27- Paper- Conference.pdf
[426] Lifang Yu, Yao Zhao, Rongrong Ni, and Ting Li. 2010. Improved adaptive LSB steganography based on chaos and
genetic algorithm. EURASIP Journal on Advances in Signal Processing 2010 (2010), 1–6.
[427] Xiangyu Yu, Huabin Tan, Hui Liang, Chang-Tsun Li, and Guangjun Liao. 2018. A multi-task learning CNN for image
steganalysis. In 2018 IEEE International Workshop on Information Forensics and Security (WIFS’18). 1–7. DOI:https:
//doi.org/10.1109/WIFS.2018.8630766
[428] Xiaoyi Yu, Tieniu Tan, and Yunhong Wang. 2004. Reliable detection of BPCS-steganography in natural images. In
Third International Conference on Image and Graphics (ICIG’04). 333–336. https://doi.org/10.1109/ICIG.2004.123
[429] Xiaopi Yu, Tieniu Tan, and Yunhong Wang. 2005. Extended optimization method of LSB steganalysis. In IEEE Inter-
national Conference on Image Processing 2005, Vol. 2. II–1102. DOI:https://doi.org/10.1109/ICIP.2005.1530252
[430] Xiaoyi Yu, Yunhong Wang, and Tieniu Tan. 2004. On estimation of secret message length in JSteg-like steganog-
raphy. In Proceedings of the 17th International Conference on Pattern Recognition, 2004 (ICPR’04)., Vol. 4. 673–676.
DOI:https://doi.org/10.1109/ICPR.2004.1333862
[431] Nazanin Zaker and Ali Hamzeh. 2012. A novel steganalysis for TPVD steganographic method based on differences
of pixel difference histogram. Multimedia Tools and Applications 58, 1 (2012), 147–166.
[432] Jishen Zeng, Shunquan Tan, Bin Li, and Jiwu Huang. 2018. Large-scale JPEG image steganalysis using hybrid deep-
learning framework. IEEE Transactions on Information Forensics and Security 13, 5 (2018), 1200–1214. DOI:https:
//doi.org/10.1109/TIFS.2017.2779446
[433] Jishen Zeng, Shunquan Tan, Guangqing Liu, Bin Li, and Jiwu Huang. 2019. WISERNet: Wider separate-then-reunion
network for steganalysis of color images. IEEE Transactions on Information Forensics and Security 14, 10 (2019), 2735–
2748. https://doi.org/10.1109/TIFS.2019.2904413
[434] Kai Zeng, Kejiang Chen, Weiming Zhang, Yaofei Wang, and Nenghai Yu. 2022. Improving robust adaptive steganog-
raphy via minimizing channel errors. Signal Processing 195 (2022), 108498. DOI:https://doi.org/10.1016/j.sigpro.2022.
108498
[435] Chaoning Zhang, Philipp Benz, Adil Karjauv, Geng Sun, and In So Kweon. 2020. UDH: Universal deep hiding for
steganography, watermarking, and light field messaging. In Advances in Neural Information Processing Systems,
H. Larochelle, M. Ranzato, R. Hadsell, M. F. Balcan, and H. Lin (Eds.), Vol. 33. Curran Associates, Inc., 10223–10234.
https://proceedings.neurips.cc/paper/2020/file/73d02e4344f71a0b0d51a925246990e7-Paper.pdf

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.
33:40 M. H. Kombrink et al.

[436] Hua Zhang, Shihuan Sun, Liting Hu, and Fanli Meng. 2022. A novel data hiding scheme based on improved diamond
encoding in IWT domain. Multimedia Tools and Applications 82 (Nov. 2022), 18727–18745.
[437] Hao Zhang, Tao Zhang, and Huajin Chen. 2019. Revisiting weighted Stego-image steganalysis for PVD steganogra-
phy. Multimedia Tools and Applications 78, 6 (2019), 7479–7497.
[438] Jun Zhang, Ingemar J. Cox, and Gwenael Doerr. 2007. Steganalysis for LSB matching in images with high-frequency
noise. In 2007 IEEE 9th Workshop on Multimedia Signal Processing. 385–388. DOI:https://doi.org/10.1109/MMSP.2007.
4412897
[439] Kevin Alex Zhang, Alfredo Cuesta-Infante, Lei Xu, and Kalyan Veeramachaneni. 2019. SteganoGAN: High Capacity
Image Steganography with GANs. (2019). arxiv:cs.CV/1901.03892
[440] Ru Zhang, Shiqi Dong, and Jianyi Liu. 2019. Invisible steganography via generative adversarial networks. Multimedia
Tools and Applications 78, 7 (2019), 8559–8575.
[441] Ru Zhang, Feng Zhu, Jianyi Liu, and Gongshen Liu. 2020. Depth-wise separable convolutions and multi-level pooling
for an efficient spatial CNN-based steganalysis. IEEE Transactions on Information Forensics and Security 15 (2020),
1138–1150. DOI:https://doi.org/10.1109/TIFS.2019.2936913
[442] Tao Zhang, Wenxiang Li, Yan Zhang, Ergong Zheng, and Xijian Ping. 2010. Steganalysis of LSB matching based
on statistical modeling of pixel difference distributions. Information Sciences 180, 23 (2010), 4685–4694. DOI:https:
//doi.org/10.1016/j.ins.2010.07.037
[443] Tao Zhang and Xijian Ping. 2003. A fast and effective steganalytic technique against JSteg-like algorithms. In Pro-
ceedings of the 2003 ACM Symposium on Applied Computing (SAC’03). Association for Computing Machinery, New
York, NY, USA, 307–311. DOI:https://doi.org/10.1145/952532.952595
[444] Tao Zhang and Xijian Ping. 2003. A new approach to reliable detection of LSB steganography in natural images.
Signal Processing 83, 10 (2003), 2085–2093. DOI:https://doi.org/10.1016/S0165- 1684(03)00169- 5
[445] Tao Zhang and Xijian Ping. 2003. Reliable detection of LSB steganography based on the difference image histogram.
In Proceedings of the 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003 (ICASSP’03).,
Vol. 3. III–545. DOI:https://doi.org/10.1109/ICASSP.2003.1199532
[446] Xinpeng Zhang and Shuozhong Wang. 2004. Vulnerability of pixel-value differencing steganography to histogram
analysis and modification for enhanced security. Pattern Recognition Letters 25, 3 (2004), 331–339. DOI:https://doi.
org/10.1016/j.patrec.2003.10.014
[447] Xinpeng Zhang and Shuozhong Wang. 2005. Steganography using multiple-base notational system and human
vision sensitivity. IEEE Signal Processing Letters 12, 1 (2005), 67–70. DOI:https://doi.org/10.1109/LSP.2004.838214
[448] Yi Zhang, Xiangyang Luo, Chunfang Yang, Dengpan Ye, and Fenlin Liu. 2015. A JPEG-compression resistant adaptive
steganography based on relative relationship between DCT coefficients. In 2015 10th International Conference on
Availability, Reliability and Security. 461–466. DOI:https://doi.org/10.1109/ARES.2015.53
[449] Yi Zhang, Xiangyang Luo, Chunfang Yang, Dengpan Ye, and Fenlin Liu. 2016. A framework of adaptive steganog-
raphy resisting JPEG compression and detection. Security and Communication Networks 9, 15 (2016), 2957–
2971.
[450] Li Zhi, Sui Ai Fen, and Yang Yi Xian. 2003. A LSB steganography detection algorithm. In 14th IEEE Proceedings on
Personal, Indoor and Mobile Radio Communications, 2003 (PIMRC’03). Vol. 3. 2780–2783. DOI:https://doi.org/10.1109/
PIMRC.2003.1259249
[451] Zhili Zhou, Xiaohua Dong, Ruohan Meng, Meimin Wang, Hongyang Yan, Keping Yu, and Kim-Kwang Raymond
Choo. 2023. Generative steganography via auto-generation of semantic object contours. IEEE Transactions on Infor-
mation Forensics and Security 18 (2023), 2751–2765. DOI:https://doi.org/10.1109/TIFS.2023.3268843
[452] Zhili Zhou, Yuecheng Su, Jin Li, Keping Yu, Q. M. Jonathan Wu, Zhangjie Fu, and Yunqing Shi. 2023. Secret-to-image
reversible transformation for generative steganography. IEEE Transactions on Dependable and Secure Computing 20,
5 (2023), 4118–4134. DOI:https://doi.org/10.1109/TDSC.2022.3217661
[453] Jiren Zhu, Russell Kaplan, Justin Johnson, and Li Fei-Fei. 2018. HiDDeN: Hiding data with deep networks. In Pro-
ceedings of the European Conference on Computer Vision (ECCV’18).
[454] Zhiqiang Zhu, Ning Zheng, Tong Qiao, and Ming Xu. 2019. Robust steganography by modifying sign of DCT coef-
ficients. IEEE Access 7 (2019), 168613–168628. DOI:https://doi.org/10.1109/ACCESS.2019.2953504
[455] Cathel Zitzmann, Rémi Cogranne, Lionel Fillatre, Igor Nikiforov, Florent Retraint, and Philippe Cornu. 2012. Hidden
information detection based on quantized Laplacian distribution. In 2012 IEEE International Conference on Acoustics,
Speech and Signal Processing (ICASSP’12). 1793–1796. DOI:https://doi.org/10.1109/ICASSP.2012.6288248
[456] Han Zong, Fen lin Liu, and Xiang yang Luo. 2012. Blind image steganalysis based on wavelet coefficient correlation.
Digital Investigation 9, 1 (2012), 58–68. DOI:https://doi.org/10.1016/j.diin.2012.02.003

Received 6 March 2023; revised 29 April 2024; accepted 1 August 2024

ACM Comput. Surv., Vol. 57, No. 2, Article 33. Publication date: October 2024.