Pixel-Sensitive and Robust Steganography Based on Polar Codes

Jinsheng Li; Ling Liu; Qi Cao; Tao Dai; Yujun Ji

REVIEW 4 major objections 5 minor 31 references

Pixel-Sensitive and Robust Steganography Based on Polar Codes

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves polar codes can reach the robust embedding capacity for adaptive steganography when each pixel suffers independent Bernoulli noise with known strength.

desk verdict A plausible but unproven polar-code scheme for per-pixel robust steganography, with experiments that sidestep the advertised operating point. read the letter →

arxiv 2506.07404 v1 pith:4ALDCZJW submitted 2025-06-09 cs.CR cs.ITmath.IT

Yujun Ji , Jinsheng Li , Ling Liu , Qi Cao , Tao Dai This is my paper · ORCID

classification cs.CRcs.ITmath.IT MSC 94A1794A2494A2994B35

keywords steganographypolarcodesadaptiverobustrate-distortioncodingchannelpolarizationattackpixel-wisenoise

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that polar codes can hide a secret message in an image twice over: they must respect the per-pixel cost of modifying pixels (so the stego image looks ordinary) and still survive a noise attack whose strength can vary from pixel to pixel. The proposed construction polarizes two channel families at once, one describing embedding and one describing attack, and carries the message only on indices that are safe for both. The main theorem says that if each pixel's attack noise is independent, known to both sender and receiver, and no stronger than the embedding probability, then the scheme achieves the robust embedding capacity in the limit, with distortion and message-error probability both tending to zero exponentially in the block length. This matters because earlier polar-code steganography could match adaptive distortion or resist uniform noise, but not both; the paper's scheme gives a capacity-approaching way to do both when the attack model is known.

What carries the argument

The load-bearing mechanism is a two-sided polarization index partition. For a block of $N$ pixels, the sender polarizes the embedding channels $W_i = \mathrm{BSC}(p_i)$, producing the set $F_1$ of subchannels that are bad (Bhattacharyya parameter near $1$), and polarizes the attack channels $Q_i = \mathrm{BSC}(\theta_i)$, producing the set $F_2$ of subchannels that are good (Bhattacharyya parameter near $0$). Because each embedding channel is degraded with respect to the corresponding attack channel whenever $\theta_i \leq p_i$, degradation is preserved by polarization, so for long codes $F_2 \subseteq F_1$. The secret message is carried on the difference set $I = F_1 \setminus F_2$: these positions are unreliable or frozen for embedding distortion yet reliable for the attack channel, while $F = F_1 \cap F_2$ carries pre-shared frozen bits. Encoding uses successive cancellation with randomized rounding; decoding uses successive-cancellation list decoding whose initial log-likelihood ratios are computed from the known attack parameters $\theta_i$.

What would settle it

Pick a linear distortion profile with $N = 2^{16}$, preset $\theta = 0.01$, actual attack noise $\tilde{\theta} = 0.01$, and embed at $R = 0.5$ bits per pixel; Theorem 2 predicts block error probability decaying as $O(2^{-(NJ)^\beta})$ and secret-message bit error rate below $10^{-5}$. If the measured error rate stays above $10^{-3}$ or fails to decrease as $N$ grows, the capacity-approaching claim is refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that robust adaptive steganography can be solved by a single polar-code construction that tracks two channels at once. Treating embedding as passing the cover $X$ through independent binary symmetric channels with crossover probabilities $p_i$, and treating attack as passing the stego $Y$ through independent binary symmetric channels with crossover probabilities $\theta_i$, the authors define $F_1$ from the embedding polarization and $F_2$ from the attack polarization, place the secret message in the index set $I = F_1 \setminus F_2$, and prove (Theorem 2) that when $0 \leq \theta_i \leq p_i \leq 1/2$ and the attack parameters are known to both parties, the achievable rate approaches $\frac{1}{N}\sum_i h_2(p_i) - \frac{1}{N}\sum_i h_2(\theta_i)$, with expected distortion $D + O(2^{-(NJ)^\beta})$ and block error probability $O(2^{-(NJ)^\beta})$. This is the robust embedding capacity in the asymptotic limit, and the simulations report secret-message bit error rates at the $10^{-5}$ level when the actual attack noise does not exceed the preset value.

Load-bearing premise

The load-bearing premise is that the attack flips each pixel independently with a probability $\theta_i$ that both sender and receiver know in advance and that is no larger than the pixel's embedding-modification probability $p_i$, with both probabilities below $1/2$; if the real attack is correlated, non-Bernoulli, or stronger than declared, the robustness guarantee collapses.

Editorial extensions

If this is right

For any rate $R < \frac{1}{N}\sum_i h_2(p_i) - \frac{1}{N}\sum_i h_2(\theta_i)$, there exists a sequence of polar codes achieving expected distortion within $O(2^{-(NJ)^\beta})$ of the design distortion and block error probability $O(2^{-(NJ)^\beta})$.
The scheme matches pixel-wise adaptive distortion profiles (constant, linear, and square) while tolerating attacks of different strengths per pixel, a combination prior polar steganography did not offer.
When the actual attack noise is no stronger than the preset value, the experiments report secret-message error rates at or below $10^{-5}$ for both uniform-noise and ratio-proportional attack models.
Robustness comes at a predictable cost: the payload is reduced by the attack entropy $\frac{1}{N}\sum_i h_2(\theta_i)$.
Encoding and decoding run in $O(NJ \log(NJ))$ time, so the construction remains practical at the simulated block lengths.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The authors leave implicit that the same two-set partition could apply when the attack channel is any binary discrete memoryless channel, not just a binary symmetric channel, as long as the per-pixel embedding channel is degraded with respect to the attack channel; the degradation-preservation lemma is channel-agnostic.
Because the rate loss equals the attack entropy, the scheme could be adapted to content-dependent security by letting the sender choose $\theta_i$ from a perceptual noise model, trading payload for guaranteed extractability under a declared attack profile.
A testable extension is to run the scheme against a learned steganalyzer: the paper measures distortion and robustness, not statistical detectability, so it remains open whether the robustness layer changes how detectable the stego image is.
The paper's remark that $|F_2 \setminus F_1|/N$ is typically small suggests a practical heuristic for mixed cases where some pixels have $\theta_i > p_i$: reserve a separate error-correcting layer on the extracted bits rather than relying on universal polar codes.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this paper targets a real gap—robust adaptive steganography under per-pixel Bernoulli attack noise—and offers a plausible polar-code construction with an achievability theorem. The problem is well chosen, but the main theorem has a proof gap and the experiments never test the operating point the abstract advertises. It deserves a serious referee, but not acceptance in this form.

The new piece is the rate expression (1/N) * sum h2(p_i) - (1/N) * sum h2(theta_i) for per-pixel attack intensities, with a nested polar construction (F1 for the embedding channel, F2 for the attack channel). That is a genuine generalization of [20] (constant noise) and [21] (adaptive but not per-pixel attack). The two-phase AM2 solution is a reasonable workaround, and the distortion experiments in Fig. 4 do approach the theoretical bound, so the embedding half of the story is credible. The paper is also honest about the nesting limitation in Remark 2, which is to its credit.

The soft spots are concentrated in the robustness half.

First, Theorem 2's key containment F2 subset of F1 is asserted from channel degradation alone. That does not follow for finite N; the thresholds are far apart, and the proof just says 'When J is sufficiently large, F2 subset of F1' without argument. You would need a polarization-based argument that the bad set of Q eventually sits inside the bad set of W. Without that, the theorem is not established.

Second, the experimental tables dodge the exact equal-noise point. In Tables I and II the columns stop below the preset theta (or below R=1). The closest point, e_theta=0.004 against theta=0.005, gives BER around 0.12—not near 10^-5. The abstract's claim about 'known to both' has no direct support in the shown data; all near-zero rows are at noise strictly weaker than the design value.

Third, AM1—the model used for most experiments—explicitly violates the theorem's premise, since p_i >= theta is not guaranteed. The paper admits this, then relies on experiments that avoid the equal-noise point, which is exactly where the violation bites.

The capacity label is also loose: there is no converse, so 'achieves the embedding capacity' is really an achievability result for a defined rate. That is minor compared to the above.

For a reader: anyone working on polar steganographic codes gets value from the construction and the problem statement, but should not take the robustness claim at face value. I would send it to peer review with a request for major revision: fix the F2 subset of F1 argument, add the e_theta=theta and R=1 experiments, and compare against [21]. If those land, the paper could be solid.

Referee Report

4 major / 5 minor

Summary. The paper studies adaptive steganography with polar codes under per-pixel binary symmetric embedding channels and per-pixel binary symmetric attack channels. It proposes a three-part index partition F = F1 ∩ F2, I = F1 \ F2, and P = F1^c, with a randomized-rounding SC/SCL encoder and an SCL decoder that treats F as frozen bits. Theorem 2 states that when 0 ≤ θ_i ≤ p_i ≤ 1/2 for every pixel, the scheme achieves expected distortion D + O(2^{-(NJ)^β}) and block error probability O(2^{-(NJ)^β}) at any embedding rate below (1/N)∑ h2(p_i) - (1/N)∑ h2(θ_i). The authors give two attack models (constant θ and constant ratio θ_i/p_i), report simulations on linear and square distortion profiles, and claim a 10^-5 secret-message error rate when the attack noise is known to both sender and receiver.

Significance. If Theorem 2 is correct, this is a genuine advance: it provides an explicit polar-code construction for pixel-dependent robust steganography with a rate approaching the robust embedding limit, polynomial-time encoding and decoding, and finite-length distortion and error exponents. The paper correctly separates the source-coding (embedding) and channel-coding (robustness) roles of the polarization indices and exploits two-stage polarization. The experimental section gives a concrete implementation using SCL decoders and shows that the scheme can be very robust when the actual attack is much weaker than the design value. The main weaknesses are that the theorem's key inclusion F2 ⊆ F1 is underproved, and the experimental validation does not actually hit the theorem's assumptions or the claimed operating point.

major comments (4)

[IV, Appendix B, Eqs. (23)-(25)] The proof of Theorem 2 asserts 'When J is sufficiently large, F2 ⊆ F1' after invoking W_NJ^(i) ≼ Q_NJ^(i). Degradation together with the definitions (14)-(15) gives only Z(W_NJ^(i)) ≥ Z(Q_NJ^(i)) ≥ 2^{-(NJ)^β}, which does not imply the required Z(W_NJ^(i)) ≥ 1 - 2^{-(NJ)^β}. This inclusion is load-bearing because the partition F = F1 ∩ F2, I = F1 \ F2 and the rate expression (25) both depend on it. The authors should either supply a proof of the inclusion, for example using the vanishing fraction of unpolarized subchannels, or replace the exact inclusion by a quantitative bound on |F2 \ F1|/(NJ) that is compatible with the stated O(2^{-(NJ)^β}) exponents.
[IV, Attack Model 1 (AM1), and Table I] The text after equation (11) explicitly states 'we cannot guarantee that p_i ≥ θ for all i∈[N]', so the main hypothesis 0 ≤ θ_i ≤ p_i of Theorem 2 is not met by AM1. Nevertheless Figure 4 and Table I validate the scheme on AM1 with θ = 0.05 and θ = 0.005/0.01. Since Theorem 2 is the only analytical support for the scheme, the paper must either modify AM1 to enforce p_i ≥ θ, for example through a constrained PLS formulation, or provide a separate error and distortion analysis for the case p_i < θ.
[V-B, Tables I and II, and Abstract] The claimed 10^-5 error rate at known attack noise is not demonstrated at the relevant operating point. Table I has no column with eθ = θ, and Table II has no row with R_{θ=0.005} = 1. In fact, for θ = 0.005 the BER at eθ = 0.004, which is below the preset value, is already 0.122-0.131 across profiles; the reported near-zero errors occur only when the actual noise is strictly below the preset. The authors should report the equal-noise columns and rows, and if the performance at equality is poor, the abstract's headline claim must be revised.
[V-A, Figure 4 and AM1] The distortion experiments use AM1 with θ = 0.05 and rates from 0.1 to 0.5 bpp, so the same hypothesis violation applies: some high-distortion pixels will have p_i < θ. Consequently the conclusion that the scheme 'approaches the theoretical bound' under AM1 is not covered by Theorem 2. Either enforce the theorem's conditions in the experiment design or state explicitly that these curves are empirical demonstrations outside the theorem's hypotheses.

minor comments (5)

[Abstract] There is a typo: 'Futher' should be 'Further'.
[IV, Eqs. (14)-(15) and Theorem 2] The index sets F1 and F2 are first defined on [N] and then used on [NJ] in Theorem 2 without restating the definitions; the notation should be adjusted to avoid ambiguity.
[IV, Lemma 2 and Corollary 1] Because the F2 ⊆ F1 step relies on the degraded polarization ordering, the statement of Lemma 2 should carry a precise citation or proof sketch rather than being dismissed as 'easily adapted'.
[V] The robustness section does not compare against the existing nested-polar robust adaptive scheme of [21] or any other robust steganographic baseline; adding such a comparison would make the claimed advantages concrete.
[Table I] The caption should state explicitly that '0' means an average bit error rate below 10^-5, since the value is used as a display threshold in the abstract.

Circularity Check

2 steps flagged · score 2.0 of 10

No fitted-input circularity; the robust-capacity label is a self-defined target and one prior work by the same authors is cited for a constant-profile experiment, but the central polar-code achievability theorem is an independent derivation.

self definitional [Section IV, Equations (10)-(13) and Remark 1]
"Restate the PLS and DLS problem as follows... subject to \sum_{i=1}^{N} h_2(p_i) - \sum_{i=1}^{N} h_2(\theta_i) = q... Remark 1. ... can achieve the per-bit average embedding capacity in Equation (12)"

The quantity called 'embedding capacity' is introduced by the authors as the objective of their reformulated robust DLS problem (Eq. 12), and Theorem 2's rate expression is exactly the same expression: R < (1/N)\sum h_2(p_r) - (1/N)\sum h_2(\theta_r). Thus the claim 'achieves the embedding capacity in Equation (12)' is, for the capacity label, a restatement of the authors' own definition rather than a result derived from an independent converse or external capacity theorem. This is a definitional overreach in labeling, not a fitted-parameter circularity; the polar-code achievability proof itself is nonzero and independent.
self citation load bearing [Section V-A, first paragraph]
"The results for the constant distortion profile are provided in [20]. Here, we consider the linear and square distortion profiles."

The constant-profile experimental evidence is delegated to [20], a paper by two of the present authors. This is a minor self-citation and is not load-bearing for the main theorem: the linear and square profile experiments are reported in the present paper, and Theorem 2's achievability argument does not rely on [20]. It is flagged only because the manuscript treats prior self-authored results as a substitute for presenting one of the three claimed distortion-profile validations.

full rationale

Theorem 2's proof constructs polar codes from the polarizations of the embedding channels {W_i} and attack channels {Q_i}, and the resulting rate, distortion, and block error bounds follow from known polar-code polarization theorems (Arikan, Korada-Urbanke, Tal-Vardy) applied to the independently specified BSC model. The proof is an achievability argument with explicit epsilon-delta steps (Equations (20), (23)-(28)); it is not a fitting exercise where parameters are tuned to the experimental data, and no experimental value is fed into the theorem. The main circularity is definitional: the authors label Eq. (12) as the 'embedding capacity' and then prove achievability of that same expression, so the claim 'achieves capacity' is true by their own definition and lacks an independent converse. The only self-citation ([20]) appears in an experimental section where the constant distortion profile is delegated to the authors' previous work; this citation is not used to establish the main theorem or to forbid alternative schemes, and the claim would still stand with independent linear/square profile experiments. No other pattern applies: the F1/F2 index sets are not fitted to achieve a target rate, the randomized rounding and SCL decoding are standard polar-code tools, and the AM1/AM2 mismatch is a hypothesis-experiment gap rather than a circular step. The schemes' headline robustness numbers (e.g., Table I) are reported only for e_theta < preset theta and never at e_theta = theta, but the central derivation itself remains self-contained.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No constants are fitted to data to make the central theorem work; experimental settings (θ, R, L, profiles) are user-selected operating points. The p_i are optimized via the Gibbs distribution, not fitted ad hoc. No new physical or mathematical entities are postulated; index sets F1, F2, I, P are construction tools.

assumptions (5)

domain assumption The cover X is a binary symmetric source (i.i.d. Bernoulli(1/2)).
Used throughout Sections III and IV to ensure Y is also BSS and to apply polar rate-distortion theorems; real image LSB planes are not i.i.d. Bernoulli.
domain assumption Attack noise is independent Bernoulli with known crossover probability θ_i for each pixel.
Section IV defines Q_i = BSC(θ_i) and the decoder LLR uses θ_i; if the noise is not BSC or θ_i is unknown, extraction fails.
domain assumption Degradedness condition 0 ≤ θ_i ≤ p_i ≤ 1/2.
Theorem 2 and the containment F2⊆F1 rely on W_i ≼ Q_i; experiments show failure when actual noise exceeds the preset value.
standard math Known polar-code achievability results [14], [15], [28], [31] are correct.
Theorems 1 and 2 are built on these cited results without re-proving them; I did not independently verify the cited proofs.
domain assumption Additive distortion model with positive weights ρ_i.
Equation (1), standard in adaptive steganography but only an approximation to real perceptual distortion.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Pixel-Sensitive and Robust Steganography Based on Polar Codes." pith.science (2026). https://pith.science/paper/4ALDCZJW

@misc{pith2026250607404,
  author       = {Pith},
  title        = {Pith review of: Pixel-Sensitive and Robust Steganography Based on Polar Codes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4ALDCZJW}},
  note         = {Machine review of arXiv:2506.07404}
}

abstract

Steganography is an information hiding technique for covert communication. The core issue in steganography design is the rate-distortion coding problem. Polar codes, which have been proven to achieve the rate-distortion bound for any binary symmetric source, are utilized to design a steganographic scheme that can reach the embedding capacity for the Distortion-Limited Sender problem in certain cases. In adaptive steganography, for attack scenarios where each noise element can have different intensities, existing steganographic coding methods fail to resist such attacks. In this paper, we propose a pixel-sensitive and robust steganographic scheme based on polar codes. Our steganographic scheme not only matches the adaptive distortion well but is also robust against sophisticated noise attacks. Futher, it is proven that our scheme achieves the embedding capacity in certain cases. Experimentally, a steganographic scheme can be designed and implemented with a secret message error rate at the $10^{-5}$ level when the attack noise is known to both the sender and the receiver. This demonstrates its significant robustness.

Figures

Figures reproduced from arXiv: 2506.07404 by the authors.

**Figure 2.** Illustration of the pixel-sensitive and robust steganography. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Eb variation with N = 2n under linear distortion profile. B. Robustness Recall that θ is the preset attack noise parameter for designing polar codes, not the actual attack noise parameter. For AM1, the actual attack noise Zei ∼ Ber(θe), and for AM2, Zei ∼ Ber(θei), i ∈ [N]. We define the constant Rθ=b = θei θi , where θi is the value determined under AM2 for θ = b. The relationship between the actual noise θei and t… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Per-bit average distortion. TABLE I AVERAGE SECRET MESSAGE BIT ERROR RATE (AM1) Profiles θe 0.001 0.002 0.004 0.006 Constant Adaptive 0.19023 0.48943 0.49883 0.49962 Ours(θ = 0.005) 0.00000 0.00062 0.12212 0.49782 Ours(θ = 0.01) 0.00000 0.00000 0.00011 0.00123 Linear A…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 30 canonical work pages

[20]

Robust steganography based on polar codes,

J. Li and L. Liu, “Robust steganography based on polar codes,” in 2021 13th Int. Conf. Wireless Communications and Signal Processing (WCSP). IEEE, 2021, pp. 1–5

work page 2021
[21]

Reliable robust adaptive steganographic coding based on nested polar codes,

Q. Yao, K. Zeng, W. Zhang, and K. Chen, “Reliable robust adaptive steganographic coding based on nested polar codes,” IEEE Trans. Signal Process., 2024

work page 2024
[1]

Bloch and J

M. Bloch and J. Barros, Physical-layer security: from information theory to security engineering . Cambridge University Press, 2011

work page 2011
[2]

Data-hiding codes,

P. Moulin and R. Koetter, “Data-hiding codes,” Proceedings of the IEEE, vol. 93, no. 12, pp. 2083–2126, 2005

work page 2005
[3]

Fridrich, Steganography in digital media: principles, algorithms, and applications

J. Fridrich, Steganography in digital media: principles, algorithms, and applications. Cambridge university press, 2010

work page 2010
[4]

Steganography in modern smart- phones and mitigation techniques,

W. Mazurczyk and L. Caviglione, “Steganography in modern smart- phones and mitigation techniques,” IEEE Communications Surveys & Tutorials, vol. 17, no. 1, pp. 334–357, 2014

work page 2014
[5]

A coding problem in steganography,

W. Zhang and S. Li, “A coding problem in steganography,” Designs, Codes and Cryptography , vol. 46, pp. 67–81, 2008

work page 2008
[6]

An efficient embedder for bch coding for steganography,

R. Zhang, V . Sachnev, M. B. Botnan, H. J. Kim, and J. Heo, “An efficient embedder for bch coding for steganography,” IEEE Trans. Inf. Theory , vol. 58, no. 12, pp. 7272–7279, 2012

work page 2012

Show all 31 references

[7]

Adaptive steganography scheme based on ldpc codes,

I. Diop, S. Farss, K. Tall, P. Fall, M. Diouf, and A. Diop, “Adaptive steganography scheme based on ldpc codes,” in 16th Int. Conf. Advanced Communication Technology. IEEE, 2014, pp. 162–166

2014
[8]

Steganography embedding cost learning with generative multi-adversarial network,

D. Huang, W. Luo, M. Liu, W. Tang, and J. Huang, “Steganography embedding cost learning with generative multi-adversarial network,” IEEE Trans. Inf. F orens. Security , 2023

2023
[9]

Automatic asymmetric embedding cost learning via generative adversarial networks,

D. Huang, W. Luo, P. Zheng, and J. Huang, “Automatic asymmetric embedding cost learning via generative adversarial networks,” in Proc. 31st ACM Int. Conf. Multimedia , 2023, pp. 8316–8326

2023
[10]

Reload: Using reinforcement learning to optimize asymmetric distortion for additive steganography,

X. Mo, S. Tan, W. Tang, B. Li, and J. Huang, “Reload: Using reinforcement learning to optimize asymmetric distortion for additive steganography,” IEEE Trans. Inf. F orens. Security , vol. 18, pp. 1524– 1538, 2023

2023
[11]

Minimizing additive distortion in steganography using syndrome-trellis codes,

T. Filler, J. Judas, and J. Fridrich, “Minimizing additive distortion in steganography using syndrome-trellis codes,” IEEE Trans. Inf. F orens. Security, vol. 6, no. 3, pp. 920–935, 2011

2011
[12]

Channel polarization: A method for constructing capacity- achieving codes for symmetric binary-input memoryless channels,

E. Arıkan, “Channel polarization: A method for constructing capacity- achieving codes for symmetric binary-input memoryless channels,”IEEE Trans. Inf. Theory , vol. 55, no. 7, pp. 3051–3073, 2009

2009
[13]

Source polarization,

——, “Source polarization,” in 2010 IEEE International Symposium on Information Theory . IEEE, 2010, pp. 899–903

2010
[14]

Polar codes for channel and source coding,

S. B. Korada, “Polar codes for channel and source coding,” EPFL, Tech. Rep., 2009

2009
[15]

Polar codes are optimal for lossy source coding,

S. B. Korada and R. L. Urbanke, “Polar codes are optimal for lossy source coding,” IEEE Trans. Inf. Theory , vol. 56, no. 4, pp. 1751–1768, 2010

2010
[16]

Polar coding steganographic embedding using successive cancellation,

B. Diouf, I. Diop, K. W. Keita, M. Diouf, S. M. Farsi, K. Tall, and O. Khouma, “Polar coding steganographic embedding using successive cancellation,” in Innovation and Interdisciplinary Solutions for Under- served Areas: 1st Int. Conf., InterSol 2017 and 6th Collogue Nat. sur ...

2017
[17]

Designing near-optimal steganographic codes in practice based on polar codes,

W. Li, W. Zhang, L. Li, H. Zhou, and N. Yu, “Designing near-optimal steganographic codes in practice based on polar codes,” IEEE Trans. Commun., vol. 68, no. 7, pp. 3948–3962, 2020

2020
[18]

High-performance steganographic coding based on sub-polarized channel,

H. Fu, X. Zhao, and X. He, “High-performance steganographic coding based on sub-polarized channel,” in Int. Workshop on Digital Water- marking. Springer, 2022, pp. 3–19

2022
[19]

On the performance bounds of stegano- graphic polar codes,

H. Fu, Y . Cao, and X. He, “On the performance bounds of stegano- graphic polar codes,” in 2024 16th Int. Conf. Wireless Communications and Signal Processing (WCSP) . IEEE, 2024, pp. 115–120

2024
[22]

Using non-binary embedding operation to minimize additive distortion functions in steganography,

T. Filler and J. Fridrich, “Using non-binary embedding operation to minimize additive distortion functions in steganography,” in 2nd IEEE Int. Workshop on Information F orensics and Security , 2010

2010
[23]

Gibbs construction in steganography,

T. Filler and J. Fridrich, “Gibbs construction in steganography,” IEEE Trans. Inf. F orens. Security, vol. 5, no. 4, pp. 705–720, 2010

2010
[24]

A performance comparison of polar codes and reed-muller codes,

E. Arıkan, “A performance comparison of polar codes and reed-muller codes,” IEEE Commun. Lett. , vol. 12, no. 6, pp. 447–449, 2008

2008
[25]

How to construct polar codes,

I. Tal and A. Vardy, “How to construct polar codes,” IEEE Trans. Inf. Theory, vol. 59, no. 10, pp. 6562–6582, 2013

2013
[26]

Llr-based suc- cessive cancellation list decoding of polar codes,

A. Balatsoukas-Stimming, M. B. Parizi, and A. Burg, “Llr-based suc- cessive cancellation list decoding of polar codes,” IEEE Trans. Signal Process., vol. 63, no. 19, pp. 5165–5179, 2015

2015
[27]

List successive cancellation decoding of polar codes,

K. Chen, K. Niu, and J. Lin, “List successive cancellation decoding of polar codes,” Electronics letters, vol. 48, no. 9, pp. 500–501, 2012

2012
[28]

Practical polar code construction over parallel channels,

K. Chen, K. Niu, and J.-R. Lin, “Practical polar code construction over parallel channels,” Iet Communications, vol. 7, no. 7, pp. 620–627, 2013

2013
[29]

Universal polar codes,

S. H. Hassani and R. Urbanke, “Universal polar codes,” in 2014 IEEE Int. Symp. Information Theory . Ieee, 2014, pp. 1451–1455

2014
[30]

Universal polarization,

E. S ¸as ¸o˘glu and L. Wang, “Universal polarization,” IEEE Trans. Inf. Theory, vol. 62, no. 6, pp. 2937–2946, 2016

2016
[31]

On the rate of channel polarization,

E. Arıkan and E. Telatar, “On the rate of channel polarization,” in 2009 IEEE Int. Symp. Information Theory . IEEE, 2009, pp. 1493–1495

2009

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [20]

Robust steganography based on polar codes,

J. Li and L. Liu, “Robust steganography based on polar codes,” in 2021 13th Int. Conf. Wireless Communications and Signal Processing (WCSP). IEEE, 2021, pp. 1–5

work page 2021

[2] [21]

Reliable robust adaptive steganographic coding based on nested polar codes,

Q. Yao, K. Zeng, W. Zhang, and K. Chen, “Reliable robust adaptive steganographic coding based on nested polar codes,” IEEE Trans. Signal Process., 2024

work page 2024

[3] [1]

Bloch and J

M. Bloch and J. Barros, Physical-layer security: from information theory to security engineering . Cambridge University Press, 2011

work page 2011

[4] [2]

Data-hiding codes,

P. Moulin and R. Koetter, “Data-hiding codes,” Proceedings of the IEEE, vol. 93, no. 12, pp. 2083–2126, 2005

work page 2005

[5] [3]

Fridrich, Steganography in digital media: principles, algorithms, and applications

J. Fridrich, Steganography in digital media: principles, algorithms, and applications. Cambridge university press, 2010

work page 2010

[6] [4]

Steganography in modern smart- phones and mitigation techniques,

W. Mazurczyk and L. Caviglione, “Steganography in modern smart- phones and mitigation techniques,” IEEE Communications Surveys & Tutorials, vol. 17, no. 1, pp. 334–357, 2014

work page 2014

[7] [5]

A coding problem in steganography,

W. Zhang and S. Li, “A coding problem in steganography,” Designs, Codes and Cryptography , vol. 46, pp. 67–81, 2008

work page 2008

[8] [6]

An efficient embedder for bch coding for steganography,

R. Zhang, V . Sachnev, M. B. Botnan, H. J. Kim, and J. Heo, “An efficient embedder for bch coding for steganography,” IEEE Trans. Inf. Theory , vol. 58, no. 12, pp. 7272–7279, 2012

work page 2012