• HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
Friday, November 20, 2026
BIOENGINEER.ORG
No Result
View All Result
  • Login
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
  • HOME
  • NEWS
  • EXPLORE
    • CAREER
      • Companies
      • Jobs
        • Lecturer
        • PhD Studentship
        • Postdoc
        • Research Assistant
    • EVENTS
    • iGEM
      • News
      • Team
    • PHOTOS
    • VIDEO
    • WIKI
  • BLOG
  • COMMUNITY
    • FACEBOOK
    • INSTAGRAM
    • TWITTER
No Result
View All Result
Bioengineer.org
No Result
View All Result
Home NEWS Science News Technology

Zero Most Significant Bit Plane Substitution Enables Lossless Data Hiding in Encrypted Images

by
November 20, 2026
in Technology
Reading Time: 6 mins read
0
Zero Most Significant Bit Plane Substitution Enables Lossless Data Hiding in Encrypted Images

Zero Most Significant Bit Plane Substitution Enables Lossless Data Hiding in Encrypted Images

Share on FacebookShare on TwitterShare on LinkedinShare on RedditShare on Telegram

Computer science researchers have introduced a streamlined reversible data hiding technique for encrypted visual records that eliminates traditional block rearrangement sorting and complex adaptive Huffman codebooks while ensuring completely lossless image reconstruction. Published in the journal Discover Computing, the study addresses a persistent operational tension in cloud computing infrastructure where remote storage servers must append administrative metadata, such as write timestamps, device provenance identifiers, custody-chain records, and cryptographic integrity checksums, directly to stored visual assets without holding decryption keys or viewing underlying plaintexts. Traditional vacating-room-after-encryption techniques typically hit an embedding ceiling below half a bit per pixel due to the loss of spatial correlation caused by standard cryptographic ciphers, whereas earlier reserving-room-before-encryption frameworks often incurred heavy auxiliary bookkeeping, super-linear block sorting overheads, or synchronization desynchronization vulnerabilities. By exploiting the natural bit-plane sparsity of predictive residuals partitioned into small spatial tiles, the new scheme achieves high net payload capacity alongside guaranteed reversibility across tens of thousands of benchmark images.

The foundational premise of the methodology rests on the statistical properties of natural visual scenes when processed through spatial prediction filters. Operating under the reserving-room-before-encryption paradigm, an eight-bit grayscale image is transformed into a modulo-256 residual domain using the causal median-edge-detector of the LOCO-I architecture. Because LOCO-I prediction errors over natural photographic scenes follow a sharp, unimodal, two-sided Laplacian distribution centered tightly at zero, the vast majority of residual values exhibit very small absolute magnitudes. Aggregated measurements across the comprehensive 10,000-image BOSSbase benchmark dataset confirmed that prediction residuals with absolute magnitudes of at least 64 account for only approximately 0.1 percent of all measured values. Consequently, when mapped to unsigned integer representations, the highest-order bit positions of neighboring residuals across local neighborhoods consist overwhelmingly of contiguous streams of zero bits.

To convert this distribution into usable embedding room without sign overhead, the design maps signed prediction residuals into non-negative integers via an unsigned ZigZag bijection. Under standard mathematical representation, preserving negative prediction errors normally demands storing an auxiliary sign bit for every pixel, which rapidly consumes embeddable capacity. The researchers proved through a formal lemma that the alternating ZigZag mapping function uniquely maps signed residuals spanning the interval from negative 128 to positive 127 into unsigned values from zero to 255. Because both the sign and magnitude of every residual are intrinsically encoded within the resulting numerical magnitude, the mapping completely eliminates the need for independent sign-bit storage or nine-bit word expansions, ensuring that the subsequently derived capacity calculations fully account for all mandatory reconstruction metadata.

Following ZigZag mapping, the residual image is divided into uniform, non-overlapping spatial tiles of four by four pixels. A systematic top-down bit-plane scan inspects each sixteen-pixel block to identify the count of consecutive, zero-valued most significant bit-planes shared across all sixteen residuals within that tile. To preserve a lean and uniform metadata structure, this count is capped at a maximum of seven, allowing each block descriptor to fit neatly into a fixed three-bit header. In rare instances where an entire tile consists exclusively of zero residuals—which represents fewer than 1.4 percent of blocks in standard photographic corpora—capping the descriptor at seven vacates the top seven bit-planes while retaining the lowest plane literally. This fixed cap avoids variable-length overflow flags while ensuring that the auxiliary bitstream remains compact and immune to bitstream desynchronization.

The encoding framework processes each tile according to an optimal classification threshold. Blocks exhibiting at least one shared zero-valued most significant bit-plane are encoded in plane mode, wherein the shared zero planes are entirely vacated and the remaining lower-order magnitude bit-planes are preserved. Each vacated plane within a four by four block instantly yields sixteen bits of embeddable space without discarding any underlying visual information. Conversely, heavily textured blocks that fail to exhibit a shared zero plane are routed into flat residual mode, where the sixteen residuals are stored literally as standard eight-bit words without vacating space. Mathematical analysis established that setting the threshold at one is strictly optimal because any plane-mode block frees at least sixteen bits while consuming only a three-bit flat header, guaranteeing a positive net gain of at least thirteen bits over flat residual encoding across every eligible tile.

Once embedding space has been reserved across all eligible tiles, the content owner encrypts the modified image structure using a cryptographically secure stream cipher, such as ChaCha20 seeded with a 256-bit secret key, or an AES counter-mode fallback. The auxiliary bitstream, containing the fixed three-bit headers and necessary block configurations, is written into the least significant bits of a contiguous pixel prefix traversed in standard raster scan order. The encrypted visual file is then transferred to an intermediate cloud server or data hider. Possessing an independent data-hiding key, the data hider embeds the secret administrative payload into the remaining available pixels along a pseudo-random permutation path. Because the encryption key and data-hiding key remain entirely separate, the system enables strictly separable three-party operations where a storage custodian can inspect, verify, or extract administrative payloads without possessing the authority or cryptographic capability to reconstruct the original visual content.

At the receiving terminal, payload extraction and original image recovery operate independently depending on the keys held by the recipient. A receiver holding only the data-hiding key regenerates the pseudo-random permutation sequence, determines the appropriate bit depth based on payload volume, and extracts the embedded bitstream directly from the ciphertext without decrypting the visual data. Conversely, an authorized image recipient possessing the encryption key applies keystream XOR decryption, extracts the auxiliary bitstream from the designated prefix region, decodes the individual block zero-plane counts, restores the vacated bit-planes to zero, and reverses the LOCO-I prediction. Because the vacated bit-planes were mathematically zero prior to substitution, restoring them introduces zero approximation error, achieving exact pixel-level recovery with an infinite peak signal-to-noise ratio and a structural similarity index of exactly one.

The empirical performance of the framework was validated across the complete 10,000-image BOSSbase and 10,000-image BOWS-2 benchmark databases, encompassing 20,000 test images in total. Under the default configuration with four by four tiles and a threshold of one, the baseline architecture attained a mean net payload capacity of 3.771 bits per pixel on BOSSbase and 3.287 bits per pixel on BOWS-2. The baseline pipeline achieved a 100.00 percent extraction and recovery success rate with zero image exclusions across the entire 20,000-image corpus. The numerical difference in capacity between the two benchmark sets directly reflects underlying visual textures: BOSSbase contains predominantly smooth outdoor scenes and natural portraits yielding an average of 4.857 zero planes per tile, whereas BOWS-2 consists of complex, highly textured steganalysis test scenes averaging 4.095 zero planes per tile, which routes 9.2 percent of its blocks into flat residual mode compared to only 0.1 percent in BOSSbase.

A critical advantage demonstrated in the study is the linear computational complexity of the pipeline relative to image dimensions. Earlier block-level reserving-room-before-encryption schemes, such as the prominent benchmark established by Yao and colleagues, achieved high capacities by performing global block sorting to cluster smooth tiles and transmitting adaptive Huffman codebooks alongside complex map bitstrings. However, global sorting imposes super-linear complexity of order B log B over the block count, while codebook transmission introduces approximately 0.47 to 0.76 bits per pixel of auxiliary overhead. In contrast, the zero-plane substitution scheme processes pixels in a single raster pass with strict O(MN) linear time complexity, requiring zero codebook metadata and reducing control metadata overhead by approximately 0.28 to 0.57 bits per pixel. Stratified runtime benchmarks conducted across different texture terciles revealed that per-image stage-one analysis times remained flat between 797 and 931 milliseconds on 512 by 512 images, demonstrating stable linear scaling suitable for high-throughput batch cloud processing.

Extensive cryptographic and statistical evaluations confirmed that the data-hiding operations preserve ciphertext randomness. Across both test corpora and standard benchmark test images including Baboon, Jetplane, Lena, and Tiffany, encrypted and payload-marked images exhibited adjacent-pixel correlation coefficients near zero, Shannon entropy exceeding 7.997 bits per pixel—reaching within 0.001 bit of the theoretical maximum of eight bits—and a 95.0 percent compliance rate with critical chi-square distribution thresholds at the five percent significance level. Furthermore, number-of-pixels-change-rate values closely bracketed the ideal boundary of 99.6094 percent, while unified-average-changing-intensity measurements spanned expected theoretical ranges. Comparing marked ciphertext against unmarked ciphertext yielded correlation coefficients below 0.05, verifying that embedding varying payload densities from one to six bits per pixel does not induce structural patterns or provide cryptographic leverage to unauthorized adversaries.

To push embedding boundaries further, the authors evaluated a stacked protocol incorporating three zero-transmission enhancements on representative 500-image subsamples. Replacing the flat three-bit block descriptor with a static, pre-published canonical prefix code reduced header entropy to roughly 2.62 bits per tile, providing a universal gain of 0.024 bits per pixel. Applying static canonical prefix coding to the skewed, retained lower magnitude bit-plane contributed an additional 0.163 bits per pixel. Finally, adopting dataset-adaptive two by two tile partitioning for dense textures isolated high-energy residual spikes and boosted smooth-block qualification on BOWS-2 from 90.8 percent to 96.7 percent. When combined, this stacked configuration elevated mean net embedding capacity to 4.056 bits per pixel on BOSSbase and 3.572 bits per pixel on BOWS-2, successfully narrowing the capacity gap on dense textures while preserving exact lossless reconstruction and linear-time execution.

Subject of Research: Reversible data hiding in encrypted images using zero bit-plane substitution and unsigned ZigZag mapping

Article Title: Zero MSB-plane substitution for reversible data hiding in encrypted images without block rearrangement

Article References: Mayoor, P. S., Gujjunoori, S., Gopalachari, M. V., Harika, A., Srivardhan, P., Selvam, A. K., & Sivakumar, N. R. (2026). Zero MSB-plane substitution for reversible data hiding in encrypted images without block rearrangement. Discover Computing, 29(1), Article 913. https://doi.org/10.1007/s10791-026-10855-x

Image Credits: AI Generated

DOI: 10.1007/s10791-026-10855-x

Keywords: reversible data hiding, encrypted images, reserving room before encryption, LOCO-I prediction, bit-plane substitution, ZigZag mapping, lossless image recovery, stream cipher, ChaCha20, BOSSbase, BOWS-2, ciphertext security, separable extraction, linear time complexity, cloud data management, Shannon entropy, block-level encoding

News Source: Denise Maddox. (November 20, 2026). Zero Most Significant Bit Plane Substitution Enables Lossless Data Hiding in Encrypted Images. Scienmag.

Tags: bit-plane substitutionblock-level encodingBOSSbaseBOWS-2ChaCha20ciphertext securitycloud data managementencrypted imageslinear time complexityLOCO-I predictionlossless image recoveryreserving room before encryptionreversible data hidingseparable extractionShannon entropystream cipherZigZag mapping
Share12Tweet7Share2ShareShareShare1

Related Posts

Swarm Intelligence Framework Clusters Event Logs to Improve Discovered Process Models

Swarm Intelligence Framework Clusters Event Logs to Improve Discovered Process Models

November 20, 2026
Mathematicians Establish Fractal Generation in M-Metric Spaces Using Generalized F-Contraction Mappings

Mathematicians Establish Fractal Generation in M-Metric Spaces Using Generalized F-Contraction Mappings

November 20, 2026

Automated Aerodynamic Cleaner Separates Cotton Impurities and Cuts Conveying Line Risks

November 20, 2026

Threshold-Aware Selective Prediction Framework Restores Balance Across Clinical Patient Trajectories

November 20, 2026

POPULAR NEWS

  • Gene Knockouts and Buffer Systems Boost Bacillus subtilis Spore Display Yields

    Gene Knockouts and Buffer Systems Boost Bacillus subtilis Spore Display Yields

    29 shares
    Share 12 Tweet 7
  • Narrative Review Integrates Imaging, Biomechanics, and Movement in Lumbar Spondylolysis

    29 shares
    Share 12 Tweet 7
  • Bifrontal Brain Stimulation Improves Working Memory in Schizophrenia Without Sex Differences

    29 shares
    Share 12 Tweet 7
  • Robotic and Conventional Knee Replacements Show Comparable One-Year Joint Infection Rates

    29 shares
    Share 12 Tweet 7

About

We bring you the latest biotechnology news from best research centers and universities around the world. Check our website.

Follow us

Recent News

Swarm Intelligence Framework Clusters Event Logs to Improve Discovered Process Models

Study Finds Many Recurrent Cystitis Diagnoses Inaccurate While Bladder Retraining Shows Promise

Modified Survey Formats Cut Response Errors in Experimental EQ-HWB Long Form

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 89 other subscribers
  • Contact Us

Bioengineer.org © Copyright 2023 All Rights Reserved.

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • Homepages
    • Home Page 1
    • Home Page 2
  • News
  • National
  • Business
  • Health
  • Lifestyle
  • Science

Bioengineer.org © Copyright 2023 All Rights Reserved.