Learning persistent and invisible latent watermarks for medical images under deep AI transformations
Khaldi A., Zermi N., Moad S., Boukhamla A., Kafi R., Sahu A.K.
Article, Optics and Laser Technology, 2026, DOI Link
View abstract ⏷
Protecting medical images in AI‑driven healthcare workflows is challenging because conventional watermarks are destroyed by routine deep learning operations such as classification, segmentation, or denoising. In this paper, a novel framework called AI‑Pipeline Persistent Latent Watermarking (APPL‑WM) is proposed. A binary watermark is embedded into multi‑scale latent features extracted by a hybrid encoder, and an invertible neural network guarantees perfect reversibility of the original image. A differentiable simulator of a realistic medical AI pipeline (CNN classifier, U‑Net segmenter, diffusion denoiser, and JPEG compression) is employed during training to enforce robustness. Experimental results on three medical imaging datasets show that the proposed method achieves high imperceptibility (PSNR of 42.35 dB, SSIM of 0.991) and retains 94.22 %-bit accuracy after the full AI pipeline, significantly outperforming state‑of‑the‑art methods (which drop below 72 %). The framework provides a robust and reversible solution for copyright protection and integrity verification of medical images processed by AI systems.
Volumetric Blind Watermarking for 3D Medical Images: Adaptive DWT–SVD Embedding with Multi-Slice Fusion for Robust Protection
Ikram H., Sayah M.M., Redouane K., Narima Z., Amine K., Boukhamla A., Sahu A.K.
Article, International Journal of Computational Intelligence and Applications, 2026, DOI Link
View abstract ⏷
This paper proposes a volumetric blind watermarking framework for 3D medical images based on adaptive DWT–SVD embedding with multi-slice fusion. The method dynamically adjusts embedding strength per slice using two complementary criteria: entropy (to exploit perceptual masking in textured regions) and anatomical position (to prioritize centrally located slices that are less likely to be cropped). A content-dependent chaotic encryption, initialized from the SHA-256 hash of the entire volume, secures the watermark before embedding into the low-frequency DWT subbands via singular value modification. During extraction, a multi-slice fusion mechanism aggregates watermark estimates from all slices using positional weights, ensuring robust recovery even under localized attacks. Experiments on 20 brain MRI and 15 chest CT volumes (256×256× up to 160 slices) demonstrate high imperceptibility (average PSNR = 44.8dB for MRI, 43.6dB for CT; SSIM >0.997) and strong robustness (average BER = 0.044 across 15 distinct attack types including JPEG compression, Gaussian noise, median filtering, rotation, scaling, cropping, and slice drop-out). Compared to nine state-of-the-art methods, including transform-based, feature-based, and deep learning approaches, the proposed framework achieves the lowest average BER (0.044 versus 0.063 for the closest competitor MCANet) while maintaining competitive imperceptibility and requiring only CPU-based computation (≈1.3 seconds per volume). These results position the proposed method as a practical, secure, and clinically viable solution for protecting patient data in telemedicine and PACS environments.
Securing Federated Learning in Medical Image Analysis: A Systematic Review of Privacy Threats and Defense Mechanisms
Abid M., Benkaddour M.K., Benouis M., Khaldi A., Sahu M., Sahu A.K.
Review, CMES - Computer Modeling in Engineering and Sciences, 2026, DOI Link
View abstract ⏷
Federated Learning (FL) is a cutting-edge method in the medical imaging field that allows hospitals to collaboratively build models without revealing patient data. Nevertheless, FL is still vulnerable to numerous security and privacy issues, including, but not limited to, data poisoning, Byzantine attacks, and inference attacks. The existing literature has only partly dealt with this topic by focusing either on particular threats or on mitigation strategies, thus leaving the overall comprehension of the problems and their solutions in medical imaging as inadequate. The main threats to FL are systematically classified in this systematic review, with two major vulnerable assets, medical data and model parameters, being pointed out. We review existing countermeasures based on cryptographic techniques, secure aggregation, perturbation methods, and security protocols with an emphasis on their efficiency in ensuring patient privacy and model integrity. We also discuss the impact of FL in medical imaging, where it serves as a tool for privacy preservation and has the potential to improve diagnostic accuracy. Data heterogeneity, communication overhead, and lack of standardization are key challenges that are considered, as well as potential future research paths to explore for solving these problems. Overall, the systematic review reveals that, although federated learning provides enormous privacy-preserving benefits in medical imaging, its actual implementation needs to be very cautious regarding the merging of very strong security measures and the use of standard protocols so that the weaknesses are reduced, and the reliability of the diagnostics is increased.
Secure and imperceptible medical image watermarking via multiscale QR embedding and attention-based optimization
Beggari A.S., Wali A., Khaldi A., Kafi M.R., Sahu A.K.
Article, Engineering Science and Technology, an International Journal, 2026, DOI Link
View abstract ⏷
The exponential growth of telemedicine and digital health platforms has introduced serious challenges in maintaining the confidentiality, authenticity, and diagnostic integrity of medical images transmitted over insecure networks. This study specifically addresses these challenges by developing a blind and imperceptible watermarking architecture that ensures both data privacy and image reliability. The proposed method integrates four complementary techniques—Non-Subsampled Shearlet Transform (NSST) for multiscale feature extraction, QR decomposition for numerically stable embedding, Particle Swarm Optimization (PSO) for adaptive block selection, and Grad-CAM attention maps for perceptual guidance. Together, these components solve three long-standing issues in medical image protection: (1) preserving diagnostic quality while embedding sensitive data, (2) achieving robustness against signal and geometric distortions without reference to the original image, and (3) reducing computational complexity for real-time telemedicine integration. The watermark encodes both compressed patient metadata and biometric images using BCH error correction and XOR encryption. Experiments on colorized CT and X-ray datasets show high imperceptibility (PSNR = 45.21 dB, SSIM = 0.9864), strong robustness (NCC ≥ 0.897), and fast runtime (≈ 2 s per image), confirming the method’s suitability for secure and practical clinical deployment.
A Comprehensive Review of Traditional and Machine Learning Based Approaches in Digital Image Watermarking
Sahu A.K., Pradhan A., Sahu M., Swain G.
Conference paper, Lecture Notes in Electrical Engineering, 2026, DOI Link
View abstract ⏷
The popularity of networks and the ongoing advancement of multimedia technology have led to an increase in the transmission of digital images through insecure channels, the need to conserve network bandwidth, and the steady increase in public awareness of copyright protection for multimedia information. Digital watermarking gives a powerful method for identifying model ownership and provides a defense against such dangers, which implies the creation of numerous eminent watermarking strategies by possible researchers. This paper reviews available techniques on image watermarking for implementation and its growth in using machine learning models. This research proposed a taxonomy for categorizing and analyzing several classes of watermarking techniques for machine learning models. Finally, thorough comparisons are made between machine learning-based watermarking techniques that offer robustness, interpretability, and embedding solid capability. Researchers can thus understand an effective machine learning model for applications with the guidance of this review paper.
A Qualitative Analysis of the Internet of Everything (IoE) and Industrial IoT (IIoT) in the Context of Industry 5.0
Sahu A.K., Anitha K., Hemalatha J., Sahu M.
Conference paper, Lecture Notes in Electrical Engineering, 2026, DOI Link
View abstract ⏷
The Internet of Everything (IoE) and the Industrial Internet of Things (IIoT) represent transformative paradigms in the world of connected technology, extending the reach of the Internet to encompass physical devices, data, people, and processes. IoE and IIoT play crucial roles in Industry 5.0, emphasizing the symbiotic relationship between humans and automation. These paradigms offer substantial benefits but also pose challenges that require attention. Recent developments in edge computing, AI, ML, and 5G networks continue to enhance the capabilities of IoE and IIoT, ushering in a more connected, intelligent, and efficient future. This qualitative research study aims to investigate the intricate dynamics of the Internet of Everything (IoE) and Industrial Internet of Things (IIoT) to explore their classification, benefits within the framework of Industry 5.0, challenges faced, diverse applications, and recent developments. Drawing on the insights gathered from meticulous content analysis of pertinent literature, this research offers valuable perspectives on these transformative technologies and their impact on various sectors.
SEF-QIM: Scale-based Embedding Factor Quantization Index Modulation Image Watermarking for Tamper Detection and Localization
Dhar S., Manna R., Amine K., Sahu A.K.
Article, Circuits, Systems, and Signal Processing, 2026, DOI Link
View abstract ⏷
This paper introduces a robust and efficient digital image watermarking technique designed for precise tamper detection and localization. The proposed approach integrates Singular Value Decomposition (SVD) and QR decomposition to generate two distinct authentication bits (Au1 and Au2), ensuring enhanced imperceptibility and security. These authentication bits are embedded into the cover image (CI) using a novel Scale-based Embedding Factor Quantization Index Modulation (SEF-QIM) technique. This embedding strategy guarantees resilience against various challenges, including geometric distortions, compression, and noise. Comprehensive experimental evaluations conducted on grayscale images demonstrate exceptional imperceptibility, achieving peak signal-to-noise ratio (PSNR) values exceeding 54 dB and 52 dB for scaling factors of 2 and 4, respectively. Additionally, the proposed method exhibits strong robustness against common attacks such as Gaussian noise, cropping, and rotation. The computational efficiency of the scheme ensures reduced time complexity for both embedding and extraction processes, making it a reliable and secure solution for image authentication in high-integrity digital media applications.
Hybrid fragile image watermarking for tamper detection, localization and dual self-recovery
Article, Engineering Science and Technology, an International Journal, 2026, DOI Link
View abstract ⏷
This paper presents a novel image watermarking framework that effectively addresses the issue of random block mapping. This phenomenon compromises tampered regions and their corresponding recovery blocks, resulting in irretrievable image data. To mitigate the random block mapping issue, a crisscross block mapping strategy (CrCsBMS) is proposed to enhance the robustness of block mapping by ensuring non-randomised reference allocation. The authentication bit generation leverages Gram-Schmidt Orthonormalization (GSO), extracting pivotal image characteristics, such as mean intensity, variance, and edge strength, thereby fortifying the integrity verification mechanism. The hybrid embedding strategy integrates discrete wavelet transform (DWT), discrete cosine transform (DCT), and singular value decomposition (SVD) to maintain an optimal balance between imperceptibility and embedding capacity, while distortion compensated quantization index modulation (DC-QIM) is employed for recovery bit encoding. A dual self-recovery mechanism incorporating bilinear interpolation-based inpainting and an 8-neighborhood method with a 255-color range scaling (255-CRS) is introduced, significantly augmenting recovery efficiency and ensuring precise restoration of tampered pixels. Experimental analysis demonstrates superior imperceptibility, robustness against image processing attacks, and reduced computational complexity compared to contemporary techniques. The proposed scheme achieves an average PSNR of 52.22 dB, an SSIM of 0.9983, and a payload capacity of 1 bit per pixel, surpassing existing self-recovery watermarking frameworks in both accuracy and resilience.
Securing patient-specific ECG data in telemedicine through adaptive wavelet-based watermarking
Hamami R., Zermi N., Boubchir L., Khaldi A., Kafi M.R., Sahu A.K., Mimoune N.
Article, Intelligence-Based Medicine, 2026, DOI Link
View abstract ⏷
Watermarking proves to be an effective technique for safeguarding crucial medical information. In this research, we propose a robust and imperceptible watermarking method designed to enhance the security of telemedicine-transmitted medical electrocardiogram (ECG) data. Embedding a mark in medical ECGs enables precise patient identification, reduces the risk of confusion during scans, and helps prevent diagnostic errors that could have adverse consequences. To ensure the security of ECG signals exchanged in telemedicine, our approach involves a frequency-domain watermarking method that conceals electronic patient records within the corresponding ECG signals. In this methodology, the signal undergoes a conversion into a 2D image, followed by a three-layer transform to extract the frequency content of the medical image. The low-frequency subbands undergo Schur decomposition, and the watermark bits are subsequently incorporated into the values of the upper triangular matrix. According to experimental results, these proposed techniques maintain a significant level of watermarked ECG quality while demonstrating high resistance to standard attacks. Experimental results show that the proposed SWT–Schur-based watermarking scheme achieves an average PSNR of 44.56 dB and an NCC higher than 0.95 under most common signal processing attacks. The average embedding capacity is 0.27 bits per pixel (BPP), while preserving the diagnostic quality of the ECG signals.
Spatially-adaptive Gaussian perturbation for reversible privacy-preserving medical image sharing
Zermi N., Moad M.S., Khaldi A., Boukhamla A., Kafi M.R., Sahu A.K.
Article, Results in Optics, 2026, DOI Link
View abstract ⏷
The rapid integration of digital technologies into modern healthcare has led to an unprecedented exchange of medical imaging data across clinical and remote platforms, raising critical concerns about patient privacy and data exposure. While conventional encryption techniques ensure secure storage and transmission, protected images become fully vulnerable once decrypted. Moreover, many existing protection schemes fail to guarantee faithful reversibility of the original diagnostic content. We propose a key-controlled, entropy-guided Gaussian perturbation framework for reversible privacy preservation. The method performs local entropy analysis to identify sensitive regions, then injects spatially adaptive Gaussian noise using cryptographically secure pseudo-random sequences from a 256-bit key. High-entropy pathological structures receive amplified perturbation while anatomical contexts are preserved. Exact reversibility is achieved through deterministic inversion. Experiments on ChestX-ray14, BraTS 2021, and OCTID demonstrate: imperceptibility (PSNR 41.8 dB, SSIM 0.971), protection against ResNet-50/U-Net (ASR 8.2%, SRR 28.5%), near-exact reconstruction (MAE 0.14), and real-time processing (57 ms/image). Cross-dataset generalization and JPEG compression robustness confirm practical viability. This training-free method enables secure telemedicine and research data sharing without compromising diagnostic integrity.
Frequency domain watermarking of medical images based on fractional discrete Cosine, Mellin, and Schur transforms
Saadaoui N., Akram Zine Eddine B., Zermi N., Khaldi A., Kafi M.R., Sahu A.K.
Article, Multimedia Tools and Applications, 2026, DOI Link
View abstract ⏷
Protecting medical images in interconnected healthcare systems requires maintaining both diagnostic integrity and secure verification. This study presents a watermarking framework combining deep feature extraction, chaotic cryptography, adaptive frequency-domain embedding, and AI-assisted extraction. Deep convolutional networks identify perceptually tolerant regions for content-aware embedding while preserving vital diagnostic areas. Hybrid chaos-based encryption secures watermark data against unauthorized recovery. The embedding operates in the fractional discrete cosine transform (FDCT) domain with adaptive coefficient selection and multi-objective optimization balancing imperceptibility, robustness, and capacity. During extraction, a convolutional autoencoder refines recovered watermarks, maintaining fidelity under compression, noise, and geometric distortions. Experimental validation on medical datasets demonstrates high embedding capacity (0.07309 BPP), excellent visual similarity (PSNR = 46.85 dB, SSIM = 0.9995), and strong resilience against attacks (NCC ≥ 0.94), ensuring secure and compliant medical image transmission.
Audio watermarking for medical traceability based on local entropy and perceptual modeling
Euschi S., Zermi N., Moad M.S., Khaldi A., Kafi M.R., Sahu A.K., Mimoune N.
Article, Multimedia Tools and Applications, 2026, DOI Link
View abstract ⏷
Securing and tracing medical audio data is crucial in telemedicine and digital archiving. This paper presents a blind and irreversible audio watermarking scheme designed to satisfy imperceptibility, robustness, and embedding capacity requirements for sensitive medical applications. The method integrates the Fractional Charlier Transform (FrCT) for adaptive time-frequency analysis, local entropy analysis with the Watson perceptual model for intelligent coefficient selection, and adaptive logarithmic quantization index modulation (LQIM) for embedding. It securely incorporates patient and acquisition metadata, ensuring confidentiality and integrity via cryptographic and error-correction techniques. Experiments demonstrate a payload of 67.3 bits per second, high audio transparency (SNR > 36 dB, PESQ > 4.0), and robustness against various signal processing attacks (average BER 4.5%). The approach is computationally efficient and suitable for telemedicine workflows, supporting authentication, integrity verification, and traceability of medical audio records.
Biometric Embedded Non-Blind Color Image Watermarking with Geometric Tamper Resistance via SIFT-ORB Keypoint Matching
Dhar S., Manna R., Amine K., Sahu A.K.
Article, Computers, 2026, DOI Link
View abstract ⏷
This work introduces a non-blind watermarking framework for color images to address tamper detection, particularly under geometric transformations. The proposed scheme fuses two watermarks, a personal signature and a biometric fingerprint, into a unified composite watermark embedded into the chrominance component of the cover image using a multi-level transform domain approach, discrete wavelet transforms (DWTs), discrete cosine transforms (DCTs), and singular value decomposition (SVD). By leveraging the rotation-invariant properties of scale-invariant feature transform (SIFT) and oriented FAST and rotated BRIEF (ORB) descriptors, the framework ensures robust tamper detection without requiring alignment, thus mitigating the limitations of conventional detection techniques vulnerable to transformation-induced tamper obfuscation (TITO). Extensive experimentation demonstrates that the method maintains high perceptual fidelity, achieving PSNR values ranging from 50 to 55 dB for embedding strength factor (Formula presented.) (0.01–0.04) and SSIM indices near 1 across multiple benchmark images. Furthermore, the scheme exhibits notable resilience to a range of image processing attacks and geometric distortion. Comparative evaluation reveals its superiority over existing grayscale, color, SIFT-based and DWT-DCT-SVD-based watermarking techniques, affirming its applicability in scenarios demanding secure, imperceptible, and transformation-invariant image watermarking.
A neural‑guided spread spectrum watermarking framework for diagnostic medical imaging
Hacini I., Moad M.S., Kafi M.R., Zermi N., Khaldi A., Boukhamla A., Sahu A.K.
Article, Journal of Ambient Intelligence and Humanized Computing, 2026, DOI Link
View abstract ⏷
The widespread use of medical imaging in telemedicine and EHRs demands robust watermarking that preserves diagnostic quality. Conventional spread spectrum methods, despite their robustness, are limited by shared secret keys, geometric vulnerabilities, and weak resilience to generative AI attacks. This paper proposes a spread spectrum discrete wavelet transform (DWT) watermarking framework for medical images, which uses a deep perceptual masking network (JNDnet) to place the watermark where it remains invisible, a lightweight CNN to adaptively select resilient sub‑bands, and a neural detector that replaces fixed‑threshold correlation for improved extraction accuracy while remaining blind. Experiments on three medical datasets show imperceptibility (PSNR > 44 dB, SSIM > 0.98) and robust performance against common and generative AI attacks, with bit error rates below 6%. Ablation studies confirm the contribution of each component, and computational efficiency supports real‑time clinical deployment.
Comprehensive Review of Machine Learning Models for Breast Cancer Diagnosis (2018–2023)
Fathima S.H., Palaparthi K.K., Patro P., Rayapoodi H.K., Dash A.K., Sahu A.K.
Article, Big Data and Computing Visions, 2026, DOI Link
View abstract ⏷
Breast Cancer (BC) remains one of the most prevalent and life-threatening diseases among women worldwide. Early and accurate detection is crucial for effective treatment and for reducing mortality rates. Aim: This study aims to provide a comprehensive review of existing Machine Learning (ML) and Deep Learning (DL) algorithms applied to BC diagnosis and classification, emphasizing their performance, strengths, and limitations. A detailed comparative analysis of various ML and DL techniques used in BC prognosis was conducted. The review covers algorithms applied to diverse data types, including mammograms, histopathological images, Fine Needle Aspiration (FNA) samples, clinical records, and genetic markers. Most algorithms demonstrate high Accuracy (ACC) in differentiating benign from malignant tumors. Techniques leveraging hybrid or ensemble learning approaches show improved diagnostic precision. However, model interpretability, dataset imbalance, and generalization across heterogeneous data sources remain key challenges. The findings highlight the growing role of ML and DL in advancing computer-aided BC diagnosis. Future research should focus on explainable models and data harmonization to enhance clinical trust and real-world applicability.
A comprehensive review of DDoS attack prevention, detection, and mitigation in IoT and SDN-IoT networks
Sutradhar S., Chowdhury K., Deb S., Sarkar J.L., Kumar C., Sahu A.K.
Review, Discover Internet of Things, 2026, DOI Link
View abstract ⏷
Distributed denial-of-service (DDoS) attacks pose a significant threat to software-defined networking with the Internet of Things (SDN-IoT) at present. Although SDN improves network agility and control, the restricted resources of IoT devices also expose new security flaws. This review provides a comprehensive analysis of DDoS attack types, taxonomies, and defense mechanisms in IoT and SDN-IoT networks. The study employs the PRISMA approach to analyze research papers published between 2020 and 2025 that focus on prevention, detection, and mitigation techniques. Unlike prior surveys that are mainly concerned with detection methods, this survey presents an in-depth and unified cross-layer comparison of prevention, detection, and mitigation techniques across the layers of IoT and SDN-IoT networks. Benchmark datasets and evaluation metrics are also compared to identify reproducibility and data imbalance issues. The review further provides a discussion on architectural elements influencing resilience, such as centralized and distributed controller architectures and controller placement in SDN-IoT systems. It also acknowledges the fact that full protection from DDoS attacks is unattainable and highlights resilience, risk mitigation, and response adaptability. Finally, key research gaps and future directions are identified to guide the development of scalable, intelligent, and collaborative DDoS defense frameworks for next-generation SDN-IoT systems.
Neural Steganography and Steganalysis: A Comprehensive Review on the Future of Hidden Communication
Review, Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 2026, DOI Link
View abstract ⏷
As the digital landscape continues to expand, the secure concealment and retrieval of information have become equally critical for ensuring confidentiality and integrity. Steganography and steganalysis serve as a cornerstone for securing and retrieving sensitive information and facilitating covert communication, while effectively mitigating potential threats. Despite significant advancements in digital communication technologies, ensuring information security remains an ongoing challenge, particularly in the context of neural network (NN)-based steganography and steganalysis architectures. This study presents a comprehensive analysis of steganography and steganalysis methods, evolving from classical to advanced intelligence frameworks by emphasizing image quality, security of hidden information, payload capacity, and robust detection mechanisms against adversarial perturbations. By leveraging state-of-the-art methods such as convolutional neural network (CNN), generative adversarial network (GAN), autoencoders, and fuzzy logic (FL) for security, imperceptibility, robustness, and the ability to address ongoing challenges related to payload capacity, adaptive information embedding and retrieval, and computational efficiency. This study explores the fields of steganography and steganalysis, focusing on embedding and extraction methods from conventional methods (non-neural network [N-NN]) to NN architectures. The insights presented serve as a foundational reference for transitioning from conventional embedding and decoding approaches to advanced NN-driven techniques, thereby enhancing security, efficiency, and resilience in covert communication systems.
Self-recovery based dual layer fragile image watermarking
Dhar S., Manna R., Pandit A., Nedungadi P., Sahu A.K.
Article, Intelligence-Based Medicine, 2026, DOI Link
View abstract ⏷
Ensuring the validity and recoverability of digital images is vital in contexts where visual data integrity is paramount. In order to address this need, this work proposed a dual-layer fragile watermarking system for safe image authentication, precise tamper localization, and self-recovery. The proposed framework produces watermark information utilizing a 4D-hypercomplex number transformation (4D-HCNT) grounded on quaternionic algebra, in conjunction with the Fast Fourier Transform (FFT) and logarithmic differencing mapping (LDM). In level-1, authentication bits are encoded utilizing a hybrid singular value decomposition (SVD) method combined with Rivest-Shamir-Adleman (RSA) cryptographic protocol to facilitate dependable integrity verification. In the level-2, recovery bits are integrated by the combination of discrete wavelet transform (DWT) and principal component analysis (PCA), enhancing imperceptibility while preserving robustness against distortions. Tamper detection is accomplished by contrasting retrieved watermark bits with regenerated watermark bits, facilitating accurate identification of altered areas. Following detection of tampering, the compromised areas are restored via a hybrid inpainting technique couples with 255-color range scaling (255-CRS) for adaptive restoration. Experimental findings indicate exceptional visual quality, evidenced by an average peak-signal-to-noise-ration (PSNR) of 52.21 dB, and structural similarity index matrix (SSIM) of 0.9990. Moreover, this framework has also achieved an average MSE, UIQI, AD, NCC, LMSE, and NAE of 0.2961, 0.9931, 0.3586, 0.9843, 0.7698, and 0.0062 respectively. The framework sustains a payload capacity of 1 bit per pixel (bpp) and attains a high-fidelity recovery for images with distortion levels of up to 50%. Overall, the proposed framework delivers efficient image authentication, precise tamper localization, and dependable self-recovery, surpassing current methods in computing efficiency and perceptual quality.
A robust latent watermarking framework for diffusion-generated images with multi-objective optimization
Bekkari F., Moad S., Amine K., Boukhamla A., Redouane K., Sahu A.K.
Article, Intelligent Systems with Applications, 2026, DOI Link
View abstract ⏷
Diffusion models enable high-quality synthetic image generation but raise serious concerns about copyright infringement and semantic manipulation. Existing watermarking methods either rely on manual hyperparameter tuning or fail under semantic AIGC edits such as inpainting and instruction-based image modification. To address these limitations, we propose LatentShield, a robust watermarking framework that embeds messages directly into the latent space of a diffusion model and optimises the trade-off between imperceptibility, extraction accuracy, and robustness using a hybrid variant of the Fireworks Optimisation Algorithm. Unlike prior work that fine-tunes the denoising U-Net or VAE decoder, LatentShield leaves the generative network unchanged, embedding watermarks via a learnable projection into the VAE latent space. The extraction network is trained adversarially against both conventional distortions and simulated AIGC semantic edits. The Hybrid Deep Fireworks Algorithm (H-DFWA) dynamically calibrates the loss weights, automatically exploring the Pareto frontier without manual tuning. Experiments on COCO, ImageNet and MedPix demonstrate that LatentShield achieves 44.27 dB PSNR, maintains bit accuracy above 96% under conventional distortions (JPEG, rotation, cropping), and reaches 97.8% and 88.4% under challenging AIGC edits (inpainting and global instruction-based editing), significantly outperforming state-of-the-art methods. An extensive ablation study confirms the contribution of each component, while statistical testing establishes the significance of the reported improvements.
Explainable and proactive fragile watermarking for medical Deepfake detection
Khaldi A., Zermi N., Moad M.S., Boukhamla A., Kafi M.R., Sahu A.K.
Article, Intelligent Systems with Applications, 2026, DOI Link
View abstract ⏷
Medical deepfakes and adversarial manipulations threaten AI-based diagnosis and telemedicine security. Existing watermarking methods for image authentication do not reliably distinguish clinically neutral transformations from semantic tampering, offer no interpretability of detected alterations, and often exceed latency constraints of clinical workflows. We propose Proactive Forensic Fragile Watermarking (PFF-WM), a framework that embeds two complementary watermarks: a fragile watermark in wavelet detail coefficients (pixel-level sensitivity) and a semi-robust watermark in the DCT domain whose embedding strength is modulated by a multi-scale attention map that prioritises diagnostically relevant regions. At the receiver side, a stacked autoencoder trained exclusively on authentic images detects manipulations via reconstruction error, a lightweight refinement network produces a tamper localisation mask, and a gradient-based explainability layer estimates the clinical impact of any alteration. Experiments on CheXpert (resampled to 512 × 512), LiTS, and ISIC 2019 show that PFF-WM achieves 97.6% detection accuracy and an AUC of 0.989, with tamper localisation IoU of 83.2%, with a false positive rate below 0.5% under each tested non‑geometric benign transformation (JPEG, resizing, contrast, blur) and below 0.7% under chained non‑geometric transformations. Geometric transformations (rotation, translation) are a recognised limitation, reaching 18.7% FPR at 15° rotation.The method shows competitive or superior performance against existing watermarking‑based forensic methods under the tested conditions, although direct comparability is limited for methods designed for different resolutions or modalities. Inference time is 44 ms per 512 × 512 image, making it computationally feasible for real‑time verification under the tested conditions.
AI-assisted simulation-based quantum image watermarking using NEQR and reversible quantum logic
Sahu A.K., Dhar S., Deb S., Khaldi A., Sahu M., Setiadi D.R.I.M.
Article, Array, 2026, DOI Link
View abstract ⏷
The Flexible Representation of Quantum Images (FRQI) is widely used for encoding images into normalized quantum states for secure processing. However, it suffers from high computational and qubit complexity, limiting its scalability. In addition, quantum comparator models, often employed for qubit comparison, remain vulnerable to geometric attacks such as scaling and rotation and require a slow embedding and extraction rate. To address these challenges, this paper proposes a novel quantum watermarking technique that integrates Quantum Image Representation (QIR) with the Novel Enhanced Quantum Representation (NEQR), offering improved efficiency, robustness, and security for quantum image processing (QIP). The proposed technique initiates watermark qubit (QB) generation by integrating pixel position bits (PPB) with an arithmetic average (AA)-based logic, implemented through a quantum circuit comprising reverse parallel adder (RPA), dividing-4 (D4) models, and controlled-NOT (CNOT) gates. For watermark embedding, the proposed circuit employs RPA, quantum equal (QE) models, CNOT gates, and Pauli-X gates, embedding the watermark via a least significant bit (LSB) strategy within the NEQR-encoded image. The extraction process mirrors the embedding architecture, utilizing reverse parallel subtractor (RPS), QE models, CNOT, and Pauli-X gates to ensure accurate watermark retrieval. Experimental results demonstrate high robustness by achieving the highest peak-signal-to-noise-ratio(PSNR) values of 59.97 dB (dB) along with a structural similarity index matrix (SSIM) of 0.9999. The proposed framework is validated using IBM Qiskit-based quantum circuit simulation on classical computing hardware and is not intended to demonstrate fault-tolerant quantum hardware execution or provable quantum computational advantage. Rather, the work focuses on simulation-level reversible quantum-circuit modelling for secure image watermarking using NEQR-based image representation and quantum-compatible reversible logic structures. Furthermore, the proposed technique is subjected to a range of image processing attacks, quantum noise and decoherence to rigorously evaluate its resilience and performance. In addition, an AI-assisted attack classification module is incorporated to automatically identify the type of distortion applied to the attacked watermarked images. A lightweight Random Forest classifier is trained to classify different attack categories, including no attack, Gaussian noise, salt-and-pepper noise, median filtering, histogram equalization, JPEG compression, cropping, rotation, and text addition. The proposed AI-assisted module achieves an overall classification accuracy of 96.56%, demonstrating its effectiveness in automatic attack identification and further strengthening the applicability of the proposed quantum watermarking framework for secure image authentication. Resource analysis shows that the watermark generation circuit requires 354 QBs (including 320 for RPA, 10 for D4, and 24 for CNOT), while both embedding and extraction consume 400 QBs each. Theoretical evaluation and empirical validation confirm the technique's effectiveness in preserving image integrity and confidentiality, establishing it as a viable approach for secure quantum image watermarking in real time applications.
A learned perceptual and neural detection framework for spread spectrum watermarking in medical images
Hacini I., Sayah M.M., Kafi M.R., Zermi N., Khaldi A., Boukhamla A., Sahu A.K.
Article, Array, 2026, DOI Link
View abstract ⏷
Medical image watermarking must ensure data authenticity and integrity without compromising diagnostic quality. Conventional spread spectrum methods suffer from key dependency, geometric vulnerability, hand-crafted perceptual models, and generative AI susceptibility. This paper presents a hybrid DWT-based spread spectrum framework for medical imaging with three innovations: a learned perceptual masking network (JND_net) that replaces analytical JND models, a lightweight CNN (SubBandSelector) that dynamically selects resilient wavelet sub-bands, and a neural detector (DetectorNet) that supplants fixed-threshold detection, all integrated with dual-key security (2128 key space) and MAC authentication. Experiments on 1200 medical images with radiologist validation (51.3% detection accuracy, chance level; diagnostic confidence unchanged, p = 0.34) show bit error rates below 0.03 under JPEG, noise, and filtering, outperforming seven baselines (p < 0.001). Under generative AI attacks, BER reaches 0.112 (diffusion) and 0.087 (inpainting). The framework operates in real time (3.7 ms on T4 GPU, 25 ms on CPU) under blind extraction.
Quantum-inspired post-quantum secure DCT watermarking for tamper detection and localization in medical images
Ikram H., Sayah M.M., Redouane K., Narima Z., Amine K., Boukhamla A., Sahu A.K.
Article, Array, 2026, DOI Link
View abstract ⏷
As medical imaging becomes increasingly central to telemedicine and electronic health records; robust integrity verification is required to preserve diagnostic quality while enabling precise tamper detection. Existing quantum image watermarking methods lack post-quantum key security and deliberate fragility. This paper proposes a quantum-inspired watermarking framework for medical images with post-quantum key security and deliberate fragility based on entangled DCT coefficients. Watermarks are carried by collective coefficient sums, making any alteration detectable. Post-quantum security is ensured through SHA3-256 key derivation. Experiments on three medical datasets (ChestX-ray14, BraTS, OCTID) show excellent imperceptibility (PSNR 44.2 dB, SSIM >0.98), high tamper sensitivity (BER >30%), and precise localization (accuracy >80%). Unlike prior methods, the framework integrates post-quantum security, deliberate fragility, and tamper localization. Furthermore, a lightweight Random Forest classifier is trained on global and block-wise BER statistics to automatically distinguish intact images from those subjected to benign processing or malicious tampering, enhancing the system's practical utility as an AI-assisted integrity verification tool for medical images in future quantum-threat environments.
Secure and imperceptible medical image watermarking based on QR factorization in DT-CWT domain
Hamami R., Zermi N., Boubchir L., Khaldi A., Kafi M.R., Sahu A.K., Mimoune N.
Article, Systems and Soft Computing, 2026, DOI Link
View abstract ⏷
With the growing reliance on telemedicine, ensuring the authenticity and confidentiality of medical images has become a critical challenge. This study presents a blind watermarking framework for medical images that balances imperceptibility, robustness, and computational efficiency. The proposed method combines Dual-Tree Complex Wavelet Transform (DT-CWT) for multi-directional decomposition, QR decomposition for numerically stable embedding, and Glowworm Swarm Optimization (GSO) for adaptive coefficient selection. To maintain visual fidelity, watermark insertion is restricted to the luminance channel and guided by a Human Visual System (HVS)-based quantization strategy. The embedded payload includes both compressed patient metadata and a biometric facial image, protected by BCH error correction and lightweight encryption. Experimental evaluations on a public dataset of RGB CT and X-ray images demonstrate strong imperceptibility (PSNR 44.67 dB, SSIM 0.9853), which is competitive with published values for comparable watermarking methods. The scheme achieves high robustness under noise, filtering, compression, and moderate geometric distortions (NCC ≥ 0.879), with competitive or improved robustness metrics (up to 15.3% higher NCC under certain attacks compared to published results), while maintaining efficient processing (1.32 s embedding, 1.08 s extraction per image). These results confirm the method's suitability for real-world telemedicine applications requiring secure and visually intact medical image transmission.
Blind audio watermarking for medical data authentication using fractional Charlier transform and adaptive dithered quantization index modulation
Euschi S., Zermi N., Med Moad S., Khaldi A., Kafi M.R., Sahu A.K., Mimoune N.
Article, Systems and Soft Computing, 2026, DOI Link
View abstract ⏷
Secure authentication and traceability of medical audio data remain critical challenges in modern telemedicine systems and digital health record management. This paper proposes a novel blind and robust audio watermarking scheme for medical applications. The method combines the Fractional Charlier Transform (FrCT) for optimized time–frequency decomposition, local entropy analysis with critical-band masking for intelligent coefficient selection, and adaptive dithered quantization index modulation (ADQIM) for imperceptible watermark embedding. The proposed scheme provides comprehensive encryption of metadata including patient information and acquisition context through AES-based cryptographic mechanisms, while maintaining imperceptibility and embedding robustness. Comprehensive experimental validation on a diverse medical audio corpus demonstrates that the method achieves a practical payload capacity of 71.8 bits per second, high audio transparency with an SNR of 38.2 dB and a PESQ score of 4.15, and strong resilience against various signal processing attacks with an average BER of 3.2 %. The approach provides a computationally efficient solution suitable for integration into operational telemedicine platforms and large-scale medical archiving systems, offering reliable authentication and integrity verification of medical audio records.
Robust medical image watermarking based on Ridgelet transform and Ant Colony Optimization for telemedicine security
Beggari A.S., Wali A., Khaldi A., Kafi M.R., Aditya Kumar S.
Article, Systems and Soft Computing, 2025, DOI Link
View abstract ⏷
The development of telemedicine needs strong solutions to prevent manipulation of medical data while maintaining accurate diagnosis. Current watermarking methods aim to establish an effective compromise between robustness, imperceptibility, and payload capacity, particularly against geometric and compression attacks. This work presents a blind medical image watermarking approach that combines adaptive Quantization Index Modulation (QIM), Ant Colony Optimization (ACO), Ridgelet transform, and QR decomposition. While QR decomposition stabilizes coefficients against noise and compression, the Ridgelet transform isolates diagnostically important linear features. Adaptive QIM dynamically modifies quantization steps according to local texture complexity, while ACO optimizes embedding locations by reducing perceptual distortion. The method achieves a high payload capacity of 73,728 bits/image when tested on a brain tumor MRI dataset, enabling the smooth integration of authentication hashes and patient details. Imperceptibility is confirmed with PSNR = 48.63 dB and SSIM = 0.9917, ensuring minimal visual distortion, while robustness evaluations show strong resistance to common attacks such as noise, filtering, and JPEG compression, with Normalized Cross-Correlation (NCC) values above 0.99 for these scenarios. Furthermore, the method maintains practical computational efficiency (0.89 s embedding, 0.52 s extraction), highlighting its applicability for securing sensitive medical information while preserving interoperability and diagnostic reliability in telemedicine workflows. However, limitations include vulnerability to severe geometric distortions (rotation >75°, cropping >25 %), and future work will focus on integrating geometric-invariant features and optimizing computational efficiency for real-time applications.
Secured textual medical information using a modified LSB image steganography technique
Oluwaseun Ogundokun R., Christiana Abikoye O., Adebayo Ogundepo E., Nathaniel Babatunde A., Tosho Abdulahi A.R., Kumar Sahu A.
Book chapter, Securing the Digital World: a Comprehensive Guide to Multimedia Security, 2025, DOI Link
View abstract ⏷
Health professionals are increasingly concerned with the welfare of their patients and the security of their medical records. With the shift toward electronic methods for obtaining and recording patient information, these records have become more vulnerable to cyberattacks. Ensuring that unauthorized individuals do not gain access to sensitive medical information is paramount. This chapter aims to enhance the security of textual medical records using an improved least significant bit (LSB) steganography technique. The objective is to develop a robust medical information system that secures patient data against potential cyber threats by implementing a modified LSB procedure called circular shift LSB steganography. The proposed system was developed and programmed in the MATLAB 2018a environment. The enhancement involved rational bit shift operations to improve the traditional LSB steganography method. The performance of the modified LSB technique was assessed using key metrics such as peak signal-to-noise ratio (PSNR), mean squared error (MSE), and the number of shifts. The modified LSB method demonstrated superior performance compared to traditional LSB methods. Quantitative analysis revealed PSNR values ranging from 74.3458 to 80.364, indicating higher image quality and reduced distortion. MSE values ranged from 0.002391 to 0.000598, showing minimal error and high fidelity of the stego images. Additionally, the number of shifts used in the embedding process ranged from 32,640 to 88,410, enhancing the security and robustness of the stego images. The enhanced LSB steganography technique, employing rational bit shift operations, outperformed traditional LSB methods regarding robustness, capacity, and imperceptibility. The introduction of the number of shifts as a new output measure further validated the improved security and effectiveness of the proposed method. The results confirm that the updated LSB technique provides a more secure solution for protecting textual medical records against unauthorized access and cyber threats. Future research should explore further optimization of the bit shift operations to enhance the security and efficiency of the LSB steganography technique. Additionally, expanding the application of this method to other types of sensitive data and evaluating its performance in different environments and scenarios will help establish its broader utility and effectiveness.
Cloud-based analysis with quantum cryptography-based cloud security model (QC-CSM) for enhanced data security in storage and access
Chandanan A.K., Sarathe V.K., Dwivedi A., Chandrasekaran R., Roy V., Sahu A.K.
Book chapter, Fortressing Pixels: Information Security for Images, Videos, Audio And Beyond, 2025, DOI Link
View abstract ⏷
Data security challenges in cloud computing have grown as a fundamental issue because of rising cyber threats during this period. The proposed model QC-CSM relies on the quantum key distribution process (QKDP) and ABE to use quantum cryptography in developing a cloud security framework that enhances security measures. The model provides key distribution together with authentication and encryption security through the application of the quantum no-cloning theorem alongside quantum mechanics principles. The experimental results conducted through CloudSim and iQuantum generate data that confirms the model exhibits efficient performance across storage, encryption, and processing speed. The key generation process takes 35% less time than conventional multi-authority attribute-based encryption (MAABE) and pairing-based provable multi-copy data possession (PB-PMDP) encryption methods. Additionally, the encryption and decryption operations operate at optimized speeds. The reduction of storage overhead (SO) amounts to 20% which leads to improved memory performance. The model demonstrates effective detection abilities to identify quantum key distribution (QKD) network eavesdropping which prevents secure network breaches. The experimental data demonstrate that QC-CSM delivers both a secure platform and a scalable and efficient cloud security infrastructure. The chapter shows that quantum cryptography enables cloud system protection when it incorporates QKD components with contemporary encryption technologies.
Privacy protection of medical data using NTRU-based post-quantum cryptography
Praneeth B.V.S.S., Ch R., Manikanta C.N., Kumar D.P., Sahu M., Sahu A.K.
Book chapter, Fortressing Pixels: Information Security for Images, Videos, Audio And Beyond, 2025, DOI Link
View abstract ⏷
In the medical field, images play a crucial role in patient monitoring, treatment planning, and diagnosis. Since the digital healthcare revolution, a large number of medical images are being transmitted and stored, making security essential. Any breach in data confidentiality can lead to misdiagnosis of patients. The traditional encryption techniques, like McEliece, have been used to shield medical data. However, the rise of quantum computers poses a threat to the prevailing encryption techniques, as they can easily break and expose medical data to quantum attacks. To overcome these attacks, post-quantum cryptography (PQC) has been introduced. Nth degree truncated polynomial ring unit (NTRU) is a lattice-based PQC algorithm that uses complex mathematical concepts that cannot be solved by quantum computers. Harnessing NTRU's lattice-based security ensures the robustness and resilience of medical data. The proposed system implements the PQC-based NTRU algorithm to protect medical data against quantum attacks.
Secure and Imperceptible Frequency-Based Watermarking for Medical Images
Naima S., Boukhamla A.Z.E., Narima Z., Amine K., Redouane K.M., Sahu A.K.
Article, Circuits, Systems, and Signal Processing, 2025, DOI Link
View abstract ⏷
Medical image security is a critical concern in the healthcare domain, and various watermarking techniques have been explored to embed imperceptible and secure data within medical images. This paper introduces an innovative frequency-based watermarking technique for medical images, utilizing the Fractional Discrete Cosine Transform (FDCT) and Schur decomposition to ensure robust and secure watermark embedding. The watermark bits are integrated by modulating the obtained Schur coefficients, thereby ensuring robust and secure watermarking without significantly altering the visual quality of the medical images. The experiments conducted on the ocular database demonstrate the capacity, imperceptibility, and robustness of the proposed method. This approach achieved a favorable trade-off between imperceptibility and information embedding capacity for ensuring the authenticity and integrity of medical images during transmission.
ECG signal protection using redundant discrete wavelet transform-based data hiding
Sayah M.M., Narima Z., Amine K., Redouane K.M., Sahu A.K.
Book chapter, Fortressing Pixels: Information Security for Images, Videos, Audio And Beyond, 2025, DOI Link
View abstract ⏷
Telemedicine provides a variety of products and services aimed at expediting the flow of digitized patient records. For efficient consultations, it is essential to centralize patient data, enabling easy access to the patient's medical history during consultations. Consequently, it is crucial for patients to utilize a secure tool with comprehensive security solutions to safeguard their information. In our effort to enhance the security of electrocardiogram (ECG) signals exchanged in telemedicine, we propose a frequency-domain watermarking approach in this chapter. This method involves concealing electronic patient records within corresponding ECG signals. The signal is initially transformed into a two-dimensional (2D) image, and the frequency content is extracted using a redundant discrete wavelet transform (RDWT). The resulting coefficients undergo Schur decomposition, and the watermark bits are incorporated by adjusting the least significant bit of the eigenvalues. Imperceptibility tests demonstrate that this approach generates a watermarked signal closely resembling the original, thereby preserving the diagnostic content. Robustness tests further indicate that the watermark can withstand commonly employed attacks in watermarking.
DWT-DCT Image Watermarking with Quantum-inspired Optimization
Rijati N., Ghosal S.K., Sahu A.K., Sambas A., Setiadi D.R.I.M.
Article, International Journal of Intelligent Engineering and Systems, 2025, DOI Link
View abstract ⏷
This study presents a robust and imperceptible image watermarking method combining Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT), and quantum-inspired optimization. The approach applies DWT to decompose the image into subbands and embeds the watermark in the low-frequency subband. DCT is used to identify Alternating Current (AC) coefficients, with Quantum-Inspired Annealing (QIA) optimizing their selection and Quantum Variational Circuits (QVC) dynamically adjusting the embedding intensity (α) for each block. The novelty of this study lies in integrating QIA and QVC to optimize both the embedding position and intensity, enabling a more adaptive and robust watermarking mechanism compared to existing quantum-inspired methods. Experimental results in standard images show high imperceptibility, with average Peak signal-to-noise ratio (PSNR) and Structural Similarity Index Measurement (SSIM) values of 46.93 dB and 0.9979, respectively. The method demonstrates strong robustness, achieving an average Normalized Correlation (NC) of 0.9752 across various attacks, including JPEG compression, noise addition, and cropping. Compared to existing methods, the proposed approach performs better in maintaining watermark quality and robustness. This study highlights the potential of quantum-inspired techniques in watermarking, offering a promising direction for further research and real-world applications.
FDCT-based watermarking for robust and imperceptible medical image protection
Said B.A., Ali W., Amine K., Redouane K.M., Sahu A.K.
Article, Intelligence-Based Medicine, 2025, DOI Link
View abstract ⏷
Security in medical imaging is a pivotal concern within the healthcare domain, prompting exploration into various watermarking techniques designed to embed imperceptible and secure data within medical images. In this study, we introduce a frequency-based medical image watermarking approach that leverages the Fractional Discrete Cosine Transform (FDCT), Mellin Transform, and Schur decomposition to extract the frequency content of the image. This process is followed by the selection of low-frequency coefficients for further transformation using Schur decomposition. The integration of watermark bits occurs through modulation of the obtained Schur coefficients, ensuring robust and secure watermarking without significantly altering the visual quality of the medical images. The experiments conducted on the ocular database illustrate the capacity, imperceptibility, and robustness of the proposed method. The proposed approach achieves a PSNR of 39.38 dB and SSIM of 0.9998, demonstrating excellent imperceptibility with a capacity of 0.07031 bits per pixel (BPP). The method is robust against various attacks, including JPEG compression, noise addition, and geometric transformations, with NCC values consistently above 0.85 for most common image processing operations. This approach successfully achieves a favorable trade-off between imperceptibility and information embedding capacity, ensuring the authenticity and integrity of medical images during transmission.
Implementation and analysis of digital watermarking techniques for multimedia authentication
Das S., Biswas P., Kar N., Kumar Sahu A.
Book chapter, Securing the Digital World: a Comprehensive Guide to Multimedia Security, 2025, DOI Link
View abstract ⏷
The world we live in has evolved to a state where information and knowledge have become power. Media consumption has increased drastically with the introduction of high-speed and affordable internet services. However, the internet is not the paradise that we dream of, it is a double-edged sword that contains information as well as misinformation. Hence, the question of authenticity arises. Determining the authenticity of multimedia content has attracted the attention of several researchers. Out of these authentication algorithms, one popular area is digital watermarking, which uses watermarks to attain robustness against malicious, manipulative attacks that might destroy the authenticity of multimedia content. This chapter presents a comprehensive analysis of such recent techniques involving watermarking that help in establishing the authenticity of images, audio, and videos. A wide variety of methods for watermarking have been chosen, namely, discrete wavelet transform, singular value decomposition, quantum index modulation, LSB substitution, and some others. The performance of the techniques has been evaluated using several metrics when subjected to common attacks.
A novel pixel pair shuffling based image watermarking for tamper detection and self-recovery
Murapaka R.R., Pavan Kumar A.V.S., Sahu A.K.
Article, Intelligence-Based Medicine, 2025, DOI Link
View abstract ⏷
This work has introduced a novel image watermarking scheme leveraging a pixel pair-based shuffling (PPSh) technique for tamper detection and self-recovery. The proposed technique consists of five steps, initiating from secret bits generation, collectively known as watermark bits. Then, the next step is watermark embedding, after that, watermark extraction, tamper detection, and finally, dual self-recovery approaches have been implemented. For watermark bit generation, two prominent interpolation techniques, such as bipolar and bilinear, are applied to the cover image (CI) to obtain the compressed image. Later, Advanced Encryption Standard (AES) and Camellia with Cipher Block Chaining (CBC) mode of operation is utilized on the compressed image to generate watermark bits. Afterwards, a PPSh-based watermark embedding strategy has been utilized to achieve the watermarked image (WI) while maintaining a standard payload capacity. Further, a variety of image processing attacks is performed on the WI to check the imperceptibility and similarity of the proposed scheme. Consequently, tamper region detection is followed by the watermark extraction procedure. Therefore, to reconstruct the tampered pixels, inpainting based dual recovery approaches are presented, named as TELEA and Naiver-Stokes (NS). The robustness and imperceptibility of the proposed scheme is measured through peak-signal-to-noise ratio (PSNR), structural similarity index matrix (SSIM), and mean square error (MSE). The proposed technique has achieved an average PSNR and SSIM of 54.24 dB and 0.9983, respectively. With an increment of more than 2 dB in terms of PSNR the proposed technique outperforms the existing watermarking techniques. Additionally, the proposed technique obtains a recovery increment up to 5 dB in terms of PSNR for 10 %–50 % tampering rates against the existing methods.
Pixel recurrence based image watermarking for block based integrity verification
Murapaka R.R., Kumar A.V.S.P., Sahu A.K.
Article, International Journal of Electronic Security and Digital Forensics, 2025, DOI Link
View abstract ⏷
This paper proposes a pixel recurrence-based digital image watermarking (PRDIW) scheme to identify the tampered blocks from an image. The proposed scheme obtains two mirrored image (MI) blocks consisting of 2×2 pixels from each carrier image (CI) pixel block. Next, a one-digit integrity value (ivone) is computed from each block and encoded inside the block to identify the tampered blocks successfully. Additionally, the proposed scheme is reversible. Therefore, it can successfully recover the CI and the encoded watermark bits at the receiving end. The results of the proposed scheme suggest that the quality of the obtained watermarked images (WIs) is superior, with an average peak signal-to-noise ratio (PSNR) of 56.11 dB, 54.68 dB and 53.05 dB, 51.68 dB while watermarking 65,536 and 131,072 bits, respectively. At the same time, the structural similarity index (SSIM) for the entire obtained watermarked image is superior to that of the existing works.
Securing the Digital World: A Comprehensive Guide to Multimedia Security
Book, Securing the Digital World: a Comprehensive Guide to Multimedia Security, 2025, DOI Link
View abstract ⏷
Securing the Digital World: A Comprehensive Guide to Multimedia Security is indispensable reading in today's digital age. With the outbreak of digital range and ever-evolving cyber threats, the demand to protect multimedia data has never been more imperative. This book provides comprehensive research on multimedia information security and bridges the gap between theoretical bases and practical applications. Authored by leading experts in the area, the book focusses on cryptography, watermarking, steganography and its advanced security solution while keeping a clear and engaging description and sets this book apart in its capability to make complex concepts accessible and practical, making it an incalculable resource for beginners and seasoned professionals alike. Key Features: Detailed study of encryption techniques, including encryption and decryption methods adapted to multimedia data A comprehensive discussion of techniques for embedding and detecting hidden information in digital media A survey of the latest advances in multimedia security, including quantum cryptography and blockchain applications Real-world case studies and illustrations that demonstrate the application of multimedia information security techniques in various initiatives Contributions from computer science and information technology experts offer a comprehensive perspective on multimedia security This book is an invaluable help for cybersecurity professionals, IT professionals, and computer and information technology students. Securing the Digital World equips readers with the information and tools required to safeguard multimedia content in a cyber-spatiality full of security challenges.
A novel image compression method using wavelet coefficients and Huffman coding
Thomas S., Krishna A., Govind S., Sahu A.K.
Article, Journal of Engineering Research (Kuwait), 2025, DOI Link
View abstract ⏷
Compressing medical images to reduce their size while maintaining their clinical and diagnostic information is crucial. Because medical images can be large and demand a lot of storage and transmission capacity, effective compression methods aid medical institutions in better storing and transmitting medical images, reducing costs, speeding up data transfer, and simplifying managing image databases. However, it is essential to note that image compression in medical imaging can also introduce drawbacks, such as loss of information and poor output image quality. Therefore, a suitable compression algorithm and parameter must be chosen to balance file size and visual fidelity. This paper suggests an effective image compression method employing the Discrete Wavelet Transform (DWT), followed by a reduction operation and Huffman coding to produce a mere lossless encoding to transmit the images over a channel. The extracted DWT coefficients are mapped to the nearest integral value. All four sub-bands of DWT are joined, and then a window of 3 × 3 is selected for reduction operation by choosing the origin as the pivot element. The Huffman coding algorithm is used to compress the processed image. The pivot origin element is used in the reversible reduction while uncompressing the image. When sending compressed data across an unreliable route, the window size and pivot element selection keep the compressed data secure. Standard measures such as bits per pixel (BPP) and compression ratio (CR) are used to assess the suggested approach. The efficiency of the suggested course of action is supported by the research's findings, which use a peak signal-to-noise ratio (PSNR) of 54.66 dB.
Opposing agents evolve the research: a decade of digital forensics
Raman R., Sahu A.K., Nair V.K., Nedungadi P.
Article, Multimedia Tools and Applications, 2025, DOI Link
View abstract ⏷
This study aims to address the growing cyber threats and the spread of digital misinformation, particularly intensified by digital transformation during the COVID-19 pandemic. It seeks to explore the evolving role of digital forensics (DF) in ensuring cybersecurity and misinformation control, with implications for societal stability and progress. By utilizing bibliometric and content analysis, we examined the landscape of 3608 DF publications from 2013 to 2023 drawn from the Scopus database. The analysis included bibliometric indicators, open access trends, author collaboration patterns, influential publications, and the geographical distribution of contributions. The decade has seen significant growth in DF research, with open access publications increasing from 25% in 2013 to 34.5% in 2023. Notably, a gender gap exists, with male researchers predominantly leading (82%). International collaborations have surged in areas such as forensic engineering and the Internet of Things (IoT). Three primary thematic clusters were identified: computer forensics, multimedia forensics, and cloud network forensics. Remarkably, DF research contributes to Sustainable Development Goals (SDGs), especially SDG 16 (Peace, Justice, and Strong Institutions), followed by SDG 9 (Industry, Innovation, and Infrastructure) and SDG 7 (Affordable and Clean Energy). By highlighting the strategic integration of DF with SDGs, this study underscores DF’s critical role in promoting just and secure societies, sustainable economic growth, and the resilience of industrial and energy infrastructures against cyber threats. DF is advocated for its proactive use in policy-making, urban development, and infrastructure protection to enhance the impact of the SDGs. This holistic approach positions DF not only as a tool for crime investigation but also as a fundamental component in advancing cybersecurity and energy policy, contributing to the broader goals of sustainable development and societal resilience.
Robust and imperceptible medical image watermarking for telemedicine applications based on transform-domain and neural clustering techniques
Beggari A.S., Wali A., Khaldi A., Kafi M.R., Aditya Kumar S.
Article, Journal of the Franklin Institute, 2025, DOI Link
View abstract ⏷
In telemedicine, the protection of exchanged medical data, particularly images, is essential to guarantee the confidentiality and integrity of sensitive information. The aim of this work is to develop a watermarking method that ensures the security of these data while preserving their visual quality. We propose a two-stage approach, in the first step, a text watermark containing sensitive information is embedded in a grayscale image using a combination of the Discrete Cosine Transform (DCT) and the K-means algorithm. In the second step, this watermarked image is embedded in a color medical image using the Discrete Wavelet Transform (DWT) and the Self-Organizing Map (SOM) algorithm, with Reed-Solomon (RS) coding to enhance security. Experimental results show that our method achieves high imperceptibility, with a peak signal-to-noise ratio (PSNR) of 45 dB. In terms of robustness, our approach effectively resists various attacks, displaying a normalized cross correlation coefficient (NCC) close to 1 for several types of attack. Statistically, our method achieves a 5-10 % higher PSNR compared to state-of-the-art techniques and maintains an NCC above 0.99 under noise, compression, and filtering attacks, outperforming existing methods by 8–15 % in robustness metrics.
A novel fiestal structured chromatic series-based data security approach
Ch R., Shaik J., Srikavya R., Sahu M., Sahu A.K.
Article, Discover Internet of Things, 2025, DOI Link
View abstract ⏷
In an era of growing cybersecurity threats, as evidenced by the growing reliance on medical data, it is necessary to develop strong encryption algorithms to protect sensitive data against intruders. Though existing encryption algorithms offer some security, they often fail to address the unique challenges posed by safeguarding multimedia medical data against sophisticated attacks. To enhance security, the proposed system employs a distinctive encryption algorithm that integrates a color-based approach with mathematical transformations. It begins with a randomly generated color grid to map plain text characters to hexadecimal values, which are then converted to ASCII and processed bit wise. The algorithm strengthens security further through block-level encryption, utilizing multiple rounds of encryption logic, including XOR operations and circular shifts. Experimental validation demonstrates the method’s effectiveness, with average encryption times of 0.18 s for text data, 111.03 s for image data, and 1520.47 s for video data. The system achieves a decryption error rate of 2–5%, underscoring its high reliability in reconstructing the original data. Furthermore, the proposed algorithm demonstrates resistance against known plaintext, chosen ciphertext, brute force, and differential cryptanalysis attacks, ensuring robust protection of sensitive medical data. These results position the proposed method as a robust and promising solution for addressing encryption challenges in securing sensitive medical data. Currently, the system uses a randomly generated 10×10 fixed-length secret key; however, future work will focus on implementing a variable-length randomly generated key for increased flexibility and security.
Fortressing Pixels: Information security for images, videos, audio and beyond
Deb S., Gutub A.A.-A., Sahu A.K.
Book, Fortressing Pixels: Information Security for Images, Videos, Audio And Beyond, 2025, DOI Link
View abstract ⏷
Emerging and pivotal technologies such as artificial intelligence and deep learning, cloud computing, internet of things, blockchain, encryption, and quantum cryptography are required to meet multimedia data's evolving needs and security requirements. In digital imaging, a pixel is the smallest element of an image that can be manipulated through software. Pixels within a digital image or graphical user interface must be secured to control unauthorized access or manipulation. In this edited book, the contributors investigate the design and implementation of effective security solutions and their integration in multimedia applications to ensure authenticity and integrity in the digital age. Coverage includes how to secure pixels within a digital image or graphical user interface to control unauthorized access or manipulation. Fortressing Pixels: Information security for images, videos, audio and beyond is intended to equip readers with the knowledge and mechanisms required to safeguard assets and protect their innovative creations, business interests, and private sensitive data. Academic researchers and engineers, will find a useful resource in this book, as will industry technology professionals working in cybersecurity, forensics, computing, networking, image processing and multimedia data science.
Fake news research trends, linkages to generative artificial intelligence and sustainable development goals
Raman R., Kumar Nair V., Nedungadi P., Kumar Sahu A., Kowalski R., Ramanathan S., Achuthan K.
Article, Heliyon, 2024, DOI Link
View abstract ⏷
In the digital age, where information is a cornerstone for decision-making, social media's not-so-regulated environment has intensified the prevalence of fake news, with significant implications for both individuals and societies. This study employs a bibliometric analysis of a large corpus of 9678 publications spanning 2013–2022 to scrutinize the evolution of fake news research, identifying leading authors, institutions, and nations. Three thematic clusters emerge: Disinformation in social media, COVID-19-induced infodemics, and techno-scientific advancements in auto-detection. This work introduces three novel contributions: 1) a pioneering mapping of fake news research to Sustainable Development Goals (SDGs), indicating its influence on areas like health (SDG 3), peace (SDG 16), and industry (SDG 9); 2) the utilization of Prominence percentile metrics to discern critical and economically prioritized research areas, such as misinformation and object detection in deep learning; and 3) an evaluation of generative AI's role in the propagation and realism of fake news, raising pressing ethical concerns. These contributions collectively provide a comprehensive overview of the current state and future trajectories of fake news research, offering valuable insights for academia, policymakers, and industry.
Preface
Editorial, Multimedia Watermarking: Latest Developments and Trends, 2024,
Multimedia Watermarking: Latest Developments and Trends
Book, Multimedia Watermarking: Latest Developments and Trends, 2024, DOI Link
View abstract ⏷
Multimedia watermarking is a key ingredient for integrity verification, transaction tracking, copyright protection, authentication, copy control, and forgery detection. This book provides an extensive survey from the fundamentals to cutting-edge digital watermarking techniques. One of the crucial aspects of multimedia security is the ability to detect forged/tampered regions from the multimedia object. In this book, we emphasized how tampering detection, localization, and recovery of manipulated information not only limits but also eliminates the scope of unauthorized usage. Finally, this book provides the groundwork for understanding the role of intelligent machines and blockchain in achieving better security in multimedia watermarking. Readers will find it easy to comprehend the wide variety of applications, theoretical principles, and effective solutions for protecting intellectual rights soon after reading this book.
A novel medical steganography technique based on Adversarial Neural Cryptography and digital signature using least significant bit replacement
Hameed M.A., Hassaballah M., Abdelazim R., Sahu A.K.
Article, International Journal of Cognitive Computing in Engineering, 2024, DOI Link
View abstract ⏷
With recent advances in technology protecting sensitive healthcare data is challenging. Particularly, one of the most serious issues with medical information security is protecting of medical content, such as the privacy of patients. As medical information becomes more widely available, security measures must be established to protect confidentiality, integrity, and availability. Image steganography was recently proposed as an extra data protection mechanism for medical records. This paper describes a data-hiding approach for DICOM medical pictures. To ensure secrecy, we use Adversarial Neural Cryptography with SHA-256 (ANC-SHA-256) to encrypt and conceal the RGB patient picture within the medical image's Region of Non-Interest (RONI). To ensure anonymity, we use ANC-SHA-256 to encrypt the RGB patient image before embedding. We employ a secure hash method with 256bit (SHA-256) to produce a digital signature from the information linked to the DICOM file to validate the authenticity and integrity of medical pictures. Many tests were conducted to assess visual quality using diverse medical datasets, including MRI, CT, X-ray, and ultrasound cover pictures. The LFW dataset was chosen as a patient hidden picture. The proposed method performs well in visual quality measures including the PSNR average of 67.55, the NCC average of 0.9959, the SSIM average of 0.9887, the UQI average of 0.9859, and the APE average of 3.83. It outperforms the most current techniques in these visual quality measures (PSNR, MSE, and SSIM) across six medical assessment categories. Furthermore, the proposed method offers great visual quality while being resilient to physical adjustments, histogram analysis, and other geometrical threats such as cropping, rotation, and scaling. Finally, it is particularly efficient in telemedicine applications with high achieving security with a ratio of 99% during remote transmission of Electronic Patient Records (EPR) over the Internet, which safeguards the patient's privacy and data integrity.
A Watermark Challenge: Synthetic Speech Detection
Narla V.L., Suresh G., Sahu A.K., Kollati M.
Book chapter, Multimedia Watermarking: Latest Developments and Trends, 2024, DOI Link
View abstract ⏷
Synthetic audio signal generation is an easier task with the help of open-source software and deep learning tools. These freely available tools are maliciously used and it negatively impacts society. To overcome this problem, we made an attempt to develop a synthetic audio signal detector. This work measures statistical and entropy features on discrete wavelength transform (DWT) transformed input audio signal. These features are trained and tested using supervised classification techniques. The proposed work is validated on a publicly available synthetic audio database. The accuracy of the proposed work is 99.0% and is compared with the state-of-the-art works validating superiority over other existing methods.
Multimodal Imputation-Based Multimodal Autoencoder Framework for AQI Classification and Prediction of Indian Cities
Srinivasa Rao R., Rao Kalabarige L., Holla M.R., Kumar Sahu A.
Article, IEEE Access, 2024, DOI Link
View abstract ⏷
Rising urbanization necessitates robust air quality monitoring and prediction systems, particularly in developing nations like India, to mitigate adverse health impacts. Previous research primarily focused on machine learning algorithms for Air Quality Index (AQI) prediction and classification. We propose a novel MI-MMA-XGB which coupled features of multimodal imputer(MI) with the features of multi-modal autoencoder (MMA) and fed to an XGBoost(XGB) algorithm for AQI prediction and classification. Moreover, imputation approaches namely, KNN, MICE, and SVD were employed to address problems with null values and outliers. Furthermore, SMOTE is employed to balance the imputed data and then the model was trained on both balanced and unbalanced imputed data to extract predictive features. In this process, our model MI-MMA-XGB achieves significant accuracy, reaching 97.14% and 93.87% with and without SMOTE, respectively. Additionally, it attains an R2 score of 0.9578 and an RMSE of 0.203 for AQI prediction in Indian cities. The proposed model outperforms baseline models in both classification and regression tasks across various evaluation metrics.
Data Privacy Protection Using Lucas Series Based Hybrid Reversible Watermarking Approach
Rupa C., Malleswari R.P., Sultana S.A., Abbas M., Sahu A.K.
Article, IEEE Access, 2024, DOI Link
View abstract ⏷
In today's digital landscape, maintaining the integrity and ownership of digital content is crucial across various fields, including the critical domain of medical applications. However, existing watermarking techniques face significant challenges, such as vulnerability to attacks and limitations in capacity and robustness. To address these challenges, this work presents a novel hybrid reversible watermarking approach utilizing the Lucas number series (LNS), normalization, and squint ternary least significant bit (STLSB). Additionally, the method incorporates a lightweight encryption technique based on homomorphic binomial coefficients (LHBC) to encrypt watermark images. The encrypted image is then embedded into the cover image at the STLSB position, specifically three bits from the left-most significant bit, resulting in a watermarked image. The Present issues in watermarking medical data include ensuring the watermark's robustness against attacks while maintaining data integrity and confidentiality, and balancing the need for traceability with the risk of compromising sensitive patient information. Experimental results show that the proposed method significantly improves embedding capacity by approximately 10% compared to existing techniques, achieving a PSNR of 52.0 dB, NCC of 1, SSIM of 0.9967, and consistently low BER, demonstrating enhanced robustness against various attacks, including JPEG compression, sharpening, resizing, and rotation. These findings highlight the method's effectiveness for secure medical image watermarking applications, ensuring integrity and confidentiality, and extending to practical uses in digital asset management, copyright protection, and authentication systems across diverse industries.
A Systematic Survey on TPE Schemes for the Cloud: Classification, Challenges, and Future Scopes
Chowdhury K., Deb S., Roy K.S., Podder D., Sahu A.K.
Article, IEEE Access, 2024, DOI Link
View abstract ⏷
It is now essential to ensure security because of the enormous volumes of data being exchanged. Particularly, images are very vulnerable to encryption flaws. Therefore, it is imperative to provide robust security mechanisms for image data. Conventional image encryption methods secure the privacy of necessary information by converting it into noise-like, encrypted ciphertext images that completely obscure visual content. However, the main drawback of this method is that it almost eliminates visual usability in the cloud. Thumbnail-preserving encryption (TPE) was created to solve this problem by retaining the image's thumbnail appearance even after encryption, providing a compromise between privacy and usability. Aspiring researchers should use this review to understand TPE schemes thoroughly. In this survey, the objectives, benefits, and drawbacks of several methods are examined with an emphasis on image privacy and usability. Notable TPE applications and evaluation metrics are also highlighted in the review. These metrics are grouped into categories such as quality, encryption key, security, and resilience to different kinds of attacks. In addition, a characterization of TPE schemes is provided, along with a comparison and summary of each scheme's contributions. In addition, we list the open research questions and difficulties that must be resolved to guarantee secure TPE for real-time applications.
Enhancing Security and Ownership Protection of Neural Networks Using Watermarking Techniques: A Systematic Literature Review Using PRISMA
Ogundokun R.O., Abikoye C.O., Sahu A.K., Akinrotimi A.O., Babatunde A.N., Sadiku P.O., Olabode O.J.
Book chapter, Multimedia Watermarking: Latest Developments and Trends, 2024, DOI Link
View abstract ⏷
The rise of artificial intelligence (AI) and machine learning (ML) has prompted concerns regarding the intellectual property (IP) protection of neural networks (NNs). A proposed solution is watermarking, which incorporates a unique identifier into a NN. However, the effectiveness of watermarking methods in enhancing privacy and ownership secrecy remains questionable. This study intended to evaluate the efficacy of watermarking techniques for enhancing the security and ownership protection (SOP) of NNs. An exhaustive search of scholarly databases for peer-reviewed journal articles and conference proceedings was conducted in accordance with PRISMA standards. Eligible papers evaluated the efficacy of watermarking techniques used to protect NNs. Twenty research articles using various watermarking techniques, including digital watermarking (DW), reversible watermarking (REW), and robust watermarking (ROW), were analyzed. Various performance indicators, such as detection rate (DR), robustness, and distortion, were employed to evaluate the applicability of each method. The results demonstrated that watermarking techniques effectively protected the intellectual property of NNs with minimal impact on performance. However, the need for specialized apparatus and the difficulty of incorporating watermarks into deep neural networks (DNN) hampered their implementation. To improve the practicability and effectiveness of watermarking techniques, additional research is required. Researchers, professionals, and policymakers should consider watermarking to safeguard the intellectual property of NNs in a variety of domains, including finance, healthcare, and national security.
Applications of artificial neural networks in the E-Commerce industry: A qualitative exploration
Book chapter, The Future of Artificial Neural Networks, 2024,
View abstract ⏷
This qualitative study explored the multifaceted applications of Artificial Neural Networks (ANNs) within the rapidly evolving landscape of the electronic commerce (E-Commerce) industry. As the digital marketplace continues to expand, businesses seek innovative solutions to enhance the user experience, optimize operations, and personalize customer interaction. ANNs, a subset of machine learning, offer a promising avenue for addressing these challenges by leveraging their capacity to mimic human brain function and extract intricate patterns from complex datasets. Employing a qualitative research approach, this study aims to uncover a diverse range of applications in which ANNs are integrated within the E-Commerce ecosystem. This study employs thematic analysis to explore the nuances and challenges associated with the integration of ANNs in the E-Commerce industry. By uncovering realworld examples and insights from experts in the field, this study provides a comprehensive understanding of how ANNs reshape the landscape of the industry. The findings shed light on the transformative potential of ANNs, offering valuable implications for businesses seeking to harness these technologies to remain competitive in the dynamic E-Commerce sector.
Advancements in artificial intelligence for biometrics: A deep dive into model-based gait recognition techniques
Parashar A., Parashar A., Shabaz M., Gupta D., Sahu A.K., Khan M.A.
Article, Engineering Applications of Artificial Intelligence, 2024, DOI Link
View abstract ⏷
Over the past decade, Deep Learning (DL) pipelines have undergone significant evolution and demonstrated effectiveness in addressing complex challenges within artificial intelligence domains. The construction of tailored DL pipelines for specific applications necessitates a solid grasp of deep learning principles and the range of intermediary layers at one's disposal. Crafting a DL pipeline involves leveraging appropriate datasets for the intended application and iteratively refining the pipeline by navigating through intermediary layers. The process of selecting and validating configurations demands substantial time and meticulous consideration, making it intricate to identify an optimal and resilient DL pipeline that excels across pertinent datasets. This article seeks to support researchers in comprehending diverse gait sensing technologies while establishing a foundational understanding of deep learning concepts to expedite problem-solving. A comprehensive overview of gait biometrics tailored for surveillance applications is presented herein. The fundamental aspects of deep learning pipelines are expounded upon, encompassing their selection criteria and implications for specific problems. Recent pivotal research on deep learning models is surveyed, encompassing their performance across varying application datasets. By elucidating the merits and limitations of these approaches, this work guides the derivation of an optimized pipeline achieved through a fusion of existing alternatives. The ultimate objective is to attain swifter yet precise outcomes for a given problem.
Multimodal imputation-based stacked ensemble for prediction and classification of air quality index in Indian cities
Rao R.S., Kalabarige L.R., Alankar B., Sahu A.K.
Article, Computers and Electrical Engineering, 2024, DOI Link
View abstract ⏷
Nowadays, monitoring and predicting the air quality is very much needed to identify and control the adverse health effects due to the low air quality, especially in developing countries like India. Recently, it has been an interesting research topic to predict the air quality index (AQI) values and levels using machine learning algorithms. In this paper, we proposed a multimodal imputation based stacked ensemble (MISE) model to classify and predict the quality of air. The multimodel imputation is constructed using various imputation techniques such as KNN Impute, MICE and SVD Impute. We experimented the proposed model with various tree based algorithms such as Random Forest, XGBoost and Extra Tree to identify the best classification and regression model for the dataset. The Stacked ensemble is developed using above algorithms for classifying the AQI bucket. Based on the experimentational study, it is observed that stacked ensemble performed better in classifying AQI with an accuracy of 96.45% using SMOTE method. The proposed stacking model also performed better than other classifier with an accuracy of 91.13% on the imbalanced data. The proposed method MISE is also applied on the dataset for identifying the AQI score using tree based regression algorithms. The stacked ensemble performed better with an R2 score of 0.9687, MAE of 0.1052 and MSE of 0.0272 compared to existing models.
Breast cancer image classification by using HCNN and LeNet5
Patro P., Fathima S.H., Harikishore R., Sahu A.K.
Article, Discover Sustainability, 2024, DOI Link
View abstract ⏷
Medical data from many sectors has greatly increased during the last 10 years. One of the main causes of death and illness among women worldwide is breast cancer. Breast cancer (BC) is one of the most common cancers in women; the death rate is high and holds the second position next to lung cancer. Breast cancer develops when cells in the mammary glands and the ducts that transfer milk to the nipple grow out of control. This is the first step in the progression of the disease. However, the time complexity of the available techniques is immense due to the many processes. Additionally, the current research attempts to improve the computation time involved in the detection process. Therefore, an effective hybrid deep learning model is introduced to improve the prediction performance and reduce the time consumption compared to the machine learning model. The breast cancer dataset, obtained from Kaggle, is used as the input data. A Wiener filter preprocessing technique is applied to enhance the image quality, with an active Wiener filter employed for this purpose. The segmentation step is achieved using a Modified Watershed Algorithm, which isolates the region of interest within the images. Finally, classification is performed using a hybrid deep learning model. This model combines a Convolutional Neural Network (CNN) with an Enhanced Recurrent Neural Network (ERNN), leveraging the strengths of both architectures. According to experimental results, the proposed Hybrid Convolutional Neural Network (HCNN) model achieves an accuracy of 96.12%, a precision of 96.99%, a recall of 97.52%, and an F-measure of 97.25%, outperforming other existing models.
Digital to quantum watermarking: A journey from past to present and into the future
Review, Computer Science Review, 2024, DOI Link
View abstract ⏷
With the amplification of digitization, the surge in multimedia content, such as text, video, audio, and images, is incredible. Concomitantly, the incidence of multimedia tampering is also apparently increasing. Digital watermarking (DW) is the means of achieving privacy and authentication of the received content while preserving integrity and copyright. Literature has produced a plethora of state-of-the-art DW techniques to achieve the right balance between its performance measuring parameters, including high imperceptibility, increased watermarking ability, and tamper-free recovery. Meanwhile, during the vertex of DW, scientific advances in quantum computing led to the emergence of quantum-based watermarking. Though quantum watermarking (QW) is in its nascent stage, it has become captivating among researchers to dive deep inside it. This study not only investigates the performance of existing DW techniques but also extensively assesses the recently devised QW techniques. It further presents how the principles of quantum entanglement and superposition can be decisive in achieving superior immunity against several watermarking attacks. To the best of our knowledge, this study is the unique one to present a comprehensive review of both DW as well as QW techniques. Therefore, the facts presented in this study could be a baseline for the researchers to devise a novel DW or QW technique.
Improved multiview biometric object detection for anti spoofing frauds
Asmitha P., Rupa C., Nikitha S., Hemalatha J., Sahu A.K.
Article, Multimedia Tools and Applications, 2024, DOI Link
View abstract ⏷
Computer vision and deep learning are essential in human authentication. It provides answers to numerous issues faced in the real world. Moreover, it has a great potential in detecting and recognizing the biometrics. It plays a significant role in reducing frauds such as spoofing, identification (ID) theft, and masking types of issues, which are difficult to perform manually, and many case studies used deep learning algorithms (DLA) like viola-jones, AlexNet, and Tiny YOLO3 but the main limitations of these studies are that they are not capable of giving high accuracy and robustness in the multi-face scenarios. So, in this article, an enhanced ArcFace (Additive Angular Margin loss) referred to as Improved ArcFace (I-AF) utilizes Convolution Neural Network (CNN) as its base architecture for feature extraction and RetinaFace are combined to overcome the above limitation, whereas RetinFace is for detecting and I-AF is for recognizing and authenticating human faces. It gives robust and accurate results while dealing with multi-faces. To evaluate the performance of the human monitoring system, it is implemented on real-time student data in a classroom to track the attendance of individuals. The faces of individuals in a classroom are detected and each detected face will be recognized and finally the result will be mapped for the attendance of individuals. To improve accuracy, the data set termed as labeled data, which contains images of students, is trained using I-AF. The system is 97% accurate, which is better than other methods.
Robust data hiding method based on frequency coefficient variance in repetitive compression
Solak S., Abdirashid A.M., Adjevi A., Sahu A.K.
Article, Engineering Science and Technology, an International Journal, 2024, DOI Link
View abstract ⏷
Sharing accurate and lossless images with higher quality through digital mediums is challenging, particularly, images shared on social media platforms can serve as good carriers for sending hidden data. However, social media platforms apply severe compression when transferring images end-to-end to serve efficient network transporting bandwidth and provide enough storage space to the users. Within the scope of this research, the proposed method introduces a novel approach by combining cryptographic and steganographic techniques, providing a robust solution to protect hidden data even when subjected to repeatedly compression. The method first encrypts the secret data to be hidden using the Advanced Encryption Standard Cipher Block Chaining (AES-CBC) technique. Then, data hiding is performed on the coefficients obtained by applying Discrete Cosine Transform (DCT) to repeatedly compressed JPEG images, coefficients that are minimally affected by compression or remain unaffected are specifically selected for data hiding. Therefore, secret data is extracted with high accuracy. Experimental results show that the proposed method outperforms the state-of-the-art in achieving robust and effective data hiding techniques based on bit error rate.
Robust medical and color image cryptosystem using array index and chaotic S-box
Podder D., Deb S., Banik D., Kar N., Sahu A.K.
Article, Cluster Computing, 2024, DOI Link
View abstract ⏷
Providing robust security within an image cryptosystem during network communication is more essential than ever, highlighting the fundamental aspects of confusion and diffusion. This study presents a novel approach for securely transmitting images through insecure channels, using array index manipulation for scrambling and simple pixel-level confusion and diffusion through XOR operations, effectively involving row and column permutation to achieve robust scrambling. This method operates two layers of confusion by rearranging array indexes using a Tent map and constructing an S-box algorithm derived from the Henon map, while the diffusion process uses a pseudo-random sequence generator based on the Henon map, with their chaotic dynamics analyzed through investigations of Lyapunov exponents and bifurcation diagrams. The expanded chaotic range and improved effects of chaotic maps produce new S-boxes with satisfactory cryptographic performance, including balancedness and non-linearity, with S-box3 achieving the highest non-linearity at 108. Following a performance and security assessment of the proposed technique, it has been found that the entropy value for the cipher image is nearly 7.998; also, the NPCR and UACI values for the cipher images are 99.6 and 33.4, respectively, and it demonstrates closeness to the ideal value. Finally, the proposed method demonstrates high randomness, undergoes evaluation using the NIST test suite, has good operation efficiency, and exhibits resilience against various attack forms such as statistical, differential, data-loss, and noise attacks, affirming its security and relevance for real-time cryptosystems.
Introduction to the special section on recent advances in multimedia forensics for cyber security and data tampering (VSI-forens)
Editorial, Computers and Electrical Engineering, 2024, DOI Link
Dual image-based reversible fragile watermarking scheme for tamper detection and localization
Sahu A.K., Sahu M., Patro P., Sahu G., Nayak S.R.
Article, Pattern Analysis and Applications, 2023, DOI Link
View abstract ⏷
This study proposes an efficient dual image-based reversible fragile watermarking scheme (DI-RFWS) that can accurately detect and locate the tampering regions from an image. The proposed scheme embeds two secret bits in each host image (HI) pixel using a pixel readjustment strategy to obtain dual watermarked images (WIs). The pixel readjustment strategy performs a maximum modification of ± 1 to the non-boundary pixels of an image based on the watermark information. The results of the study suggest that in addition to reversibility, the proposed scheme offers triple objective of high capacity, better perceptual transparency, and robustness. Experimental results also show that the proposed scheme achieves a superior peak signal-to-noise ratio (PSNR) of above 52 dB for both the WIs. Further, the proposed scheme can efficiently detect and locate the tampering regions from an image with a high true positive rate, low false positive and negative rate for various tampering rates. Additionally, the proposed scheme shows superior resistance against various intentional and unintentional attacks.
Hazardous Asteroid Prediction using Majority Voting Technique
Reddy C.V.R., Sai T.N., Sushanth V., Muvva S., Rani D.R., Sahu A.K.
Conference paper, Proceedings of the 7th International Conference on Intelligent Computing and Control Systems, ICICCS 2023, 2023, DOI Link
View abstract ⏷
The existence of life on the earth not only depends upon trees, water, food, minerals, and other resources of the earth, but also upon the asteroids and objects that travel from the outer atmosphere of the earth. When an asteroid from space, with very high speed, hits the earth's surface [Meteorite], it deals a lot of damage to the human beings and this planet. The goal of this research study is to build one such machine learning model which predicts the Hazardous asteroid effectively. In this work, six ML models namely Naive Bayes, Logistic Regression, Decision Tree, Random Forest, K-NN and SVM were used for hazardous asteroid prediction. Based on accuracy, Random Forest, Logistic Regression and Decision Tree were combined using Majority Voting Technique to predict whether the asteroid is hazardous or not. Majority Voting Technique (MVT) of machine learning models has shown higher significance when compared to the individual machine learning model in terms of accuracy. The accuracy of different individual models is ranging from 90% to 99.86% and MVT produced 100% accuracy.
Chaotic-Map Based Encryption for 3D Point and 3D Mesh Fog Data in Edge Computing
Raghunandan K.R., Dodmane R., Bhavya K., Rao N.S.K., Sahu A.K.
Article, IEEE Access, 2023, DOI Link
View abstract ⏷
Recent decades have seen dramatic development and adoption of digital technology. This technological advancement generates a large amount of critical data that must be safeguarded. The security of confidential data is one of the primary concerns in fog computing. As a result, achieving a reliable level of security in the fog computing environment is crucial. In this context, 3D point and mesh fog data are becoming increasingly popular among the various types of data stored in the fog. Data encryption using chaotic behavior is one of the preferred research areas due to its unique properties, such as randomness, determinism, sensitivity to initial conditions, and ergodicity. In this paper, we have taken advantage of this chaotic behavior to achieve higher security. This study presents a novel approach for protecting the privacy of 3D point and mesh fog data. Initially, the fog data coordinates are transformed using the sequence generated by the chaotic behavior. Then, bifurcation analysis is used to depict the enhanced scope of the proposed map. The quality of the proposed chaotic system is assessed using metrics such as the Lyapunov exponent and approximate entropy. Results show that the proposed encryption framework performs superior when subjected to brute-force and statistical attacks. Further, the designed framework produces better results than the prior literature.
Latest Trends in Deep Learning Techniques for Image Steganography
Kumar V., Sharma S., Kumar C., Sahu A.K.
Article, International Journal of Digital Crime and Forensics, 2023, DOI Link
View abstract ⏷
The development of deep convolutional neural networks has been largely responsible for the significant strides forward made in steganography over the past decade. In the field of image steganography, generative adversarial networks (GAN) are becoming increasingly popular. This study describes current development in image steganographic systems based on deep learning. The authors’ goal is to lay out the various works that have been done in image steganography using deep learning techniques and provide some notes on the various methods. This study proposed a result that could open up some new avenues for future research in deep learning based on image steganographic methods. These new avenues could be explored in the future. Moreover, the pros and cons of current methods are laid out with several promising directions to define problems that researchers can work on in future research avenues.
Local-Ternary-Pattern-Based Associated Histogram Equalization Technique for Cervical Cancer Detection
Srinivasan S., Raju A.B.K., Mathivanan S.K., Jayagopal P., Babu J.C., Sahu A.K.
Article, Diagnostics, 2023, DOI Link
View abstract ⏷
Every year, cervical cancer is a leading cause of mortality in women all over the world. This cancer can be cured if it is detected early and patients are treated promptly. This study proposes a new strategy for the detection of cervical cancer using cervigram pictures. The associated histogram equalization (AHE) technique is used to improve the edges of the cervical image, and then the finite ridgelet transform is used to generate a multi-resolution picture. Then, from this converted multi-resolution cervical picture, features such as ridgelets, gray-level run-length matrices, moment invariant, and enhanced local ternary pattern are retrieved. A feed-forward backward propagation neural network is used to train and test these extracted features in order to classify the cervical images as normal or abnormal. To detect and segment cancer regions, morphological procedures are applied to the abnormal cervical images. The cervical cancer detection system’s performance metrics include 98.11% sensitivity, 98.97% specificity, 99.19% accuracy, a PPV of 98.88%, an NPV of 91.91%, an LPR of 141.02%, an LNR of 0.0836, 98.13% precision, 97.15% FPs, and 90.89% FNs. The simulation outcomes show that the proposed method is better at detecting and segmenting cervical cancer than the traditional methods.
Secure Reversible Data Hiding Using Block-Wise Histogram Shifting
Kamil S., Sahu M., Raghunandan K.R., Sahu A.K.
Article, Electronics (Switzerland), 2023, DOI Link
View abstract ⏷
Reversible data hiding (RDH) techniques recover the original cover image after data extraction. Thus, they have gained popularity in e-healthcare, law forensics, and military applications. However, histogram shifting using a reversible data embedding technique suffers from low embedding capacity and high variability. This work proposes a technique in which the distribution obtained from the cover image determines the pixels that attain a peak or zero distribution. Afterward, adjacent histogram bins of the peak point are shifted, and data embedding is performed using the least significant bit (LSB) technique in the peak pixels. Furthermore, the robustness and embedding capacity are improved using the proposed dynamic block-wise reversible embedding strategy. Besides, the secret data are encrypted before embedding to further strengthen security. The experimental evaluation suggests that the proposed work attains superior stego images with a peak signal-to-noise ratio (PSNR) of more than 58 dB for 0.9 bits per pixel (BPP). Additionally, the results of the two-sample t-test and the Kolmogorov–Smirnov test reveal that the proposed work is resistant to attacks.
A Study on Content Tampering in Multimedia Watermarking
Sahu A.K., Umachandran K., Biradar V.D., Comfort O., Sri Vigna Hema V., Odimegwu F., Saifullah M. A
Article, SN Computer Science, 2023, DOI Link
View abstract ⏷
Technological progresses offer more occasions for tampering outbreaks. Lithography services at affordable prices, in aggregation with open software tools to influence changes in digital spaces, invigorated amateurs to oblige to fabricating and imitating. Tampering has reached a level of cleverness that leaves negligible trace with the pace of progress happening with editing technology, luring the next generation. However tampering with technology violates intellectual property rights that can be treated to cost dearly, including severe retributions. At the same time, accountability for evading, aiding the evasion of technology, and restricting access that could impede the infringement of the content are new protection activities, which were not present during the pre-digital age. Digitalization shrinks the cost of content development, nevertheless availability of pirated content is also on the rise due to ease of copy, transform and distribution. Surveillance is a big dataset, with arrangements from various sources at diverse scenarios that are critical to events, therefore, susceptible errors such as defocusing, occlusion and displacement. The forensic study is thus correlated, for prediction using multi-task joint model through convolutional neural network (CNN), as they are open to access in metadata, also its alteration through ease of EXIF tools provide innumerable opportunities to tamper contents, thus tough to identify, except after severe inquiries. In this study, we present a brief overview of recent status with respect to the content tampering using several advanced tools.
Logistic-map based fragile image watermarking scheme for tamper detection and localization
Sahu A.K., Hassaballah M., Rao R.S., Suresh G.
Article, Multimedia Tools and Applications, 2023, DOI Link
View abstract ⏷
In this paper, two logistic-map based fragile image watermarking schemes are proposed. The first scheme is a conventional irreversible, whereas the second scheme is a reversible one. The proposed first scheme considers a pair of two consecutive host image (HI) pixels for embedding the watermark bits. At the embedding end, each HI pixel observes a maximum of ±1 modifications to produce the watermarked pixels. At the same time, the second scheme utilizes the concept of mirrored images of the HI to reproduce the image as well as the watermark bits, with minimal distortion. The experimental results show that the quality of the watermarked image is superior with an average peak signal-to-noise ratio (PSNR) of more than 51 dB for both schemes. Also, the first scheme offers excellent tamper detection and localization ability as compared to the existing state-of-art schemes. Besides, promising results are obtained in favor of the proposed scheme for measures like accuracy, true positive (TP), true negative (TN), false positive (FP), false negative (FN), and precision.
Towards improving the performance of blind image steganalyzer using third-order SPAM features and ensemble classifier
Hemalatha J., Sekar M., Kumar C., Gutub A., Sahu A.K.
Article, Journal of Information Security and Applications, 2023, DOI Link
View abstract ⏷
The success rate for blind or universal steganalysis lies in the ability to extract the statistical footprints of image features. Further, the choice of machine learning (ML) algorithm is crucial to distinguish the stego image more precisely from the untouched clean images. Literature suggests that most steganalysis approaches report less favorable detection accuracy despite considering many features. This study presents a three-step process to accurately identify the clean and stego images to solve this issue. We used the curvelet denoising as an initial phase during the first step to suppress the natural noise residuals (NRs) by producing the stego NRs. Secondly, it extracts the Third-order Markov-chain sample transition probability matrices as features. Finally, the oblique decision tree ensemble using a multisurface proximal support vector machine (SVM) classifier has been utilized to achieve greater detection accuracy than the state-of-the-art classifiers. The experiments are performed on an extensive database comprising clean and stego images generated from nine embedding schemes with varying payloads. The experimental results suggest that an accuracy of 93.12 has been achieved using the proposed Third order subtractive pixel adjacency matrix (SPAM) features with an ensemble classifier.
Enhancing Cloud Communication Security: A Blockchain-Powered Framework with Attribute-Aware Encryption
Raghunandan K.R., Kallapu B., Dodmane R., Krishnaraj Rao N.S., Thota S., Sahu A.K.
Article, Electronics (Switzerland), 2023, DOI Link
View abstract ⏷
The global production of information continuously increases in quantity and variety. However, the tools and technologies developed to handle such large volumes of data have not adequately met the security and privacy requirements. Existing cloud security systems, often managed by a trusted third party, are susceptible to various security risks. To address these challenges and ensure the protection of personal information, blockchain technology emerges as a crucial solution with substantial potential. This research uses the blockchain-powered attribute-aware encryption method to establish a real-time secure communication approach over the cloud. By employing attribute-based encryption technology, data owners can implement fine-grained search permissions for data users. The proposed solution incorporates accessible encryption technology to enable secure access to encrypted data and facilitate keyword searches on the blockchain. This study provides a functional comparison of recently developed attribute-based encryption algorithms. The access control strategy comprises two access tree types and a linear secret-sharing system, serving as the main components. The elliptic curve’s base field was set to 512b, and the bilinear pairing parameter type used was Type-A. This approach involves storing keywords on a remote server and encrypting them using attribute-based encryption. Furthermore, the encrypted data blockchain and the corresponding ciphertext are stored in the blockchain. Numerical experiments were conducted to evaluate the system’s key generation, trapdoor building, and keyword retrieval capabilities.
A Novel and Secure Fake-Modulus Based Rabin-Ӡ Cryptosystem
Ramesh R.K., Dodmane R., Shetty S., Aithal G., Sahu M., Sahu A.K.
Article, Cryptography, 2023, DOI Link
View abstract ⏷
Electronic commerce(E-commerce) transactions require secure communication to protect sensitive information such as credit card numbers, personal identification, and financial data from unauthorized access and fraud. Encryption using public key cryptography is essential to ensure secure electronic commerce transactions. RSA and Rabin cryptosystem algorithms are widely used public key cryptography techniques, and their security is based on the assumption that it is computationally infeasible to factorize the product of two large prime numbers into its constituent primes. However, existing variants of RSA and Rabin cryptosystems suffer from issues like high computational complexity, low speed, and vulnerability to factorization attacks. To overcome the issue, this article proposes a new method that introduces the concept of fake-modulus during encryption. The proposed method aims to increase the security of the Rabin cryptosystem by introducing a fake-modulus during encryption, which is used to confuse attackers who attempt to factorize the public key. The fake-modulus is added to the original modulus during encryption, and the attacker is unable to distinguish between the two. As a result, the attacker is unable to factorize the public key and cannot access the sensitive information transmitted during electronic commerce transactions. The proposed method’s performance is evaluated using qualitative and quantitative measures. Qualitative measures such as visual analysis and histogram analysis are used to evaluate the proposed system’s quality. To quantify the performance of the proposed method, the entropy of a number of occurrences for the pixels of cipher text and differential analysis of plaintext and cipher text is used. When the proposed method’s complexity is compared to a recent variant of the Rabin cryptosystem, it can be seen that it is more complex to break the proposed method—represented as (Formula presented.) which is higher than Rabin-P ((Formula presented.) algorithms.
Intelligent data classification using optimized fuzzy neural network and improved cuckoo search optimization
Patro P., Kumar K., Kumar G.S., Sahu A.K.
Article, Iranian Journal of Fuzzy Systems, 2023, DOI Link
View abstract ⏷
In data mining, classification is one of the most critical steps in predicting the target class. It is performed by an improved model in existing work in which feature selection is performed based on the bat optimization method to increase the classification accuracy. This study uses an enhanced neural network for classification, including intuitive, interpretable correlated-contours fuzzy rules. Further, a practical model is created based on the extraction of fuzzy rules, where data partitioning is performed via a similarity-based directional component. However, the dataset used for experimentation is noisy and incomplete data values. Due to incompleteness, knowledge discovery is obstructed, and the classification results are affected. Here bat provides very slow convergence and easily falls into local optima. To solve this issue, an improved framework is introduced in which missing value imputation is performed by using k means clustering, and then for feature selection, an improved cuckoo search optimization is used. An enhanced classifier based on fuzzy logic and alex net neural network structure (F-ANNS) is used for classification, and hybrid ant colony particle swarm optimization (HASO) is used for optimizing parameters of the alex net neural network classifier. The results show that the proposed work is more effective in precision, recall, accuracy, and f-measure as shown by experimental results.
Data Hiding Using PVD and Improving Security Using RSA
Jeyaprakash H., Kartheeban K., Sahu A.K., Chokkalingam B.
Article, Journal of Applied Security Research, 2022, DOI Link
View abstract ⏷
Steganography deals with hiding information, which offers ultimate security in defense, profitable usages, thus sending the imperceptible information, will not be bare or distinguished by others. In this paper, a multidirectional PVD hiding scheme with RSA algorithm is proposed. Secret bits are embedded in three directions of a color image by dividing the non-overlapping blocks into R, G, b channels by selecting the minimum pixels of each block regrouping. In order to ensure the security of an stego-image, the work proposed by scheme is used, and in addition to that RSA algorithm is proposed.
Performance analysis of various image steganography techniques
Sahu M., Padhy N., Gantayat S.S., Sahu A.K.
Conference paper, 2022 2nd International Conference on Computer Science, Engineering and Applications, ICCSEA 2022, 2022, DOI Link
View abstract ⏷
Information hiding is the most interesting and prominent field of data security. With the evolution of computational infrastructure, covert communication techniques have received an inclusive acknowledgment among researchers as well as common participants of data communication. Among all, an image steganography technique which is a field of data hiding is one of the foremost choices among experts. The foremost defiance in scheming a steganographic system is to preserve an adequate equilibrium among the measures. The objective of this study is to present a all-inclusive survey of disparate existing images steganography techniques (ISTs) with regard to different performance assessment standards, such as (1) camouflage image (CI) quality, (2) capacity, and (3) robustness to different attacks. Further, the underlying challenges and future directions are also highlighted.
High fidelity based reversible data hiding using modified LSB matching and pixel difference
Article, Journal of King Saud University - Computer and Information Sciences, 2022, DOI Link
View abstract ⏷
Owing to the inefficiency to hide large volume of secret data for the reversible data hiding (RDH) image steganography approaches, we propose two improved RDH based approaches, such as (1) improved dual image based least significant bit (LSB) matching with reversibility, and (2) n-rightmost bit replacement (n-RBR) and modified pixel value differencing (MPVD). The first approach extends the ability of LSB matching with reversibility using dual images. Whereas the second approach utilizes four identical cover images for secret data embedding using two phases, such as (1) n-rightmost bit replacement (n-RBR) and (2) modified pixel value differencing (MPVD). In the n-RBR phase, n bits of secret data are embedded in the pair of two neighboring pixels of the first two identical images, where 1 ≤ n ≤ 4. Correspondingly, the MPVD phase uses the third and fourth identical images for hiding the secret data. Experimental results with respect to peak signal-to-noise ratio (PSNR), embedding capacity (EC), structural similarity index (SSIM), and the comparative analysis with recently proposed state-of-art approaches exhibit the superiority of the proposed approach. Besides reversibility, the proposed approach ensures high fidelity to salt and pepper (S&P) noise, RS analysis, and pixel difference histogram (PDH) analysis.
False-Positive-Free SVD Based Audio Watermarking with Integer Wavelet Transform
Suresh G., Narla V.L., Gangwar D.P., Sahu A.K.
Article, Circuits, Systems, and Signal Processing, 2022, DOI Link
View abstract ⏷
Singular Value Decomposition (SVD) became a promising approach for developing digital media watermarking techniques due to stability and higher energy packing nature of singular values. Nevertheless, SVD based watermarking techniques suffers from false positive problem (FPP) when singular vectors are shared for extraction. Eliminating FPP in the development of digital audio watermarking (DAW) is still a challenging task. In this work, SVD based schemes and their vulnerability to FPP are studied, analyzed, and elucidated in detail. Further, a false positive free SVD based DAW scheme has been devised in Integer Wavelet Transform (IWT) domain. Audio is partitioned into segments. Each audio segment is transformed using IWT and SVD is applied on Arnold transformed watermark. Principal Component (PC) is obtained with the product of singular vector matrix and singular values matrix. Transformed audio is modified based on PC of watermark image. The developed scheme has been tested on benchmark dataset and it maintains imperceptibility, robustness, and capacity as per standards. The developed scheme has achieved resilience against signal processing attacks. Consequently, this DAW scheme helps in forensic examination of audio recording for authentication purpose.
Improving grayscale steganography to protect personal information disclosure within hotel services
Article, Multimedia Tools and Applications, 2022, DOI Link
View abstract ⏷
An unauthorized leak of information that is not tolerable always implies inadequate or poor security measures of information. Globally, the hospitality industry has in the recent past been targeted by cybercrimes. Management of cybercrimes are divided into facets like policies of security frameworks, cyber-threats, and management appreciating the value of Information Technology investment. These aspects possess a notable influence on the information security of an organization. This study’s purpose is to examine cybersecurity activities of network threats, electronic information, and the techniques of preventing cybercrime in hotels. Helping Chief Information Officers (CIO) and the directors of information technology is the main aim of the research to improve policy for electronic information security in the hospitality industry and recommending several tools and techniques to stabilize the network of computers. Further, to protect personal information disclosure for the visitors, an information hiding technique has been proposed by utilizing the least significant bits (LSBs) of each pixel of grayscale image adopting XOR features of the host image (HI) pixels. Also, the proposed technique successfully withstands various steganalysis attacks like regular and singular (RS) attack, pixel difference histogram (PDH) attack, and subtractive pixel adjacency matrix (SPAM) steganalysis.
Local binary pattern-based reversible data hiding
Sahu M., Padhy N., Gantayat S.S., Sahu A.K.
Article, CAAI Transactions on Intelligence Technology, 2022, DOI Link
View abstract ⏷
A novel local binary pattern-based reversible data hiding (LBP-RDH) technique has been suggested to maintain a fair symmetry between the perceptual transparency and hiding capacity. During embedding, the image is divided into various 3×3 blocks. Then, using the LBP-based image descriptor, the LBP codes for each block are computed. Next, the obtained LBP codes are XORed with the embedding bits and are concealed in the respective blocks using the proposed pixel readjustment process. Further, each cover image (CI) pixel produces two different stego-image pixels. Likewise, during extraction, the CI pixels are restored without the loss of a single bit of information. The outcome of the proposed technique with respect to perceptual transparency measures, such as peak signal-to-noise ratio and structural similarity index, is found to be superior to that of some of the recent and state-of-the-art techniques. In addition, the proposed technique has shown excellent resilience to various stego-attacks, such as pixel difference histogram as well as regular and singular analysis. Besides, the out-off boundary pixel problem, which endures in most of the contemporary data hiding techniques, has been successfully addressed.
A logistic map based blind and fragile watermarking for tamper detection and localization in images
Article, Journal of Ambient Intelligence and Humanized Computing, 2022, DOI Link
View abstract ⏷
In real-time data transmission, the protection of multimedia content from unauthorized access is pivotal. In this context, digital watermarking techniques have drawn significant attention from the past few decades. However, most of the reported techniques fail to achieve a good balance among the perceptual transparency, embedding capacity (EC), and robustness. Besides, tamper detection and localization are the two crucial aspects of any authentication based watermarking technique. This paper proposes a logistic map based fragile watermarking technique to efficiently detect and localize the tampered regions from the watermarked image (WI). The proposed technique takes advantage of the sensitivity property of the logistic map to generate the watermark bits. Next, these watermark bits are embedded in the rightmost least significant bits (LSBs) by performing the logical XOR operation between the first intermediate significant bits (ISBs) and the watermark bits. Simulation results show that the proposed technique can produce high quality WI with an average peak signal-to-noise ratio (PSNR) of 51.14 dB. Further, the proposed technique can efficiently detect and locate the tampering regions from the image with a high true positive rate, low false positive and negative rate. Additionally, the proposed technique exhibits an excellent ability to resist various intentional and unintentional attacks which makes it suitable for real-time applications.
Shadow Image Based Reversible Data Hiding Using Addition and Subtraction Logic on the LSB Planes
Sahu M., Padhy N., Gantayat S.S., Sahu A.K.
Article, Sensing and Imaging, 2021, DOI Link
View abstract ⏷
Image steganographic communication demands a fair trade-off among the three diametrically opposed metrics such as higher capacity, larger visual quality, and attack survival ability (ASA). Recently, some reversible data hiding (RDH) techniques using dual images have shown promising results to achieve the aforementioned needs. However, maintaining a balance among these metrics is still an open challenge. In this paper, using the concept of shadow image, which is basically the replica of the cover image (CI) and performing some simple addition and subtraction logic on the shadow image pixels, we propose an improved RDH technique that offers larger capacity, better stego-image (SI) quality and higher ASA. At first, during embedding, three shadow images of the CI are produced. Then, the shadow image pixels are adjusted based on their XOR features of the least significant bit planes. After embedding the secret bits, a maximum of ± 1 modification has been observed in the SI pixels. Later at the receiving end, the CI has been restored by applying the round function on the obtained SI pixels. Experimental results show that the proposed technique offers excellent visual quality with peak signal-to-noise ratio and structural similarity index (SSIM) of 52.47 dB, 53.91 dB, 52.48 dB and 0.9974, 0.9981, 0.9974 for the respective shadow images. Further, the proposed technique show exceptional anti-steganalysis ability to regular and singular analysis, pixel difference histogram analysis, and bit pair analysis. Additionally, the proposed technique successfully avoids the falling-off boundary problem.
Multi-directional block based PVD and modulus function image steganography to avoid FOBP and IEP
Sahu A.K., Swain G., Sahu M., Hemalatha J.
Article, Journal of Information Security and Applications, 2021, DOI Link
View abstract ⏷
Since the inception of pixel value differencing (PVD) image steganography, it has drawn considerable interest among the researchers of this field. However, most of the PVD based techniques suffer from either falling-off boundary problem (FOBP) or incorrect extraction problem (IEP). Therefore, to address these two issues, this paper proposes a multi-directional pixel value differencing and modulus function (MDPVDMF) based technique. During the embedding process, the original image (OI) is partitioned into 2 × 2 size pixel blocks. Then, data embedding is performed by exploiting the horizontal, vertical, and diagonal directions for each block. For a 2 × 2 pixel block, two difference values can be obtained in any of the three directions. Next, using the difference values and the remainders of the pixel pairs, the secret bits are embedded. The experiment has been conducted to compute the performance of the proposed technique with regards to the image quality metrics like peak signal-to-noise ratio (PSNR), embedding capacity (EC), and FOBP. Results show that PSNR is optimal for vertical pairs with 39.17 dB whereas the EC is optimal for the diagonal pairs with 3.10 bits per pixel (BPP). Further, the proposed technique has shown exceptional attack resistance ability to regular & singular (RS) attack, salt & pepper (S&P) noise, pixel difference histogram (PDH) analysis, and subtractive pixel adjacency matrix (SPAM) steganalysis.
Reversible Image Steganography Using Dual-Layer LSB Matching
Article, Sensing and Imaging, 2020, DOI Link
View abstract ⏷
Recently, reversible information hiding (RIH) methods have drawn substantial attention in many privacy-sensitive real-time applications, such as the Internet of Things (IoT) enabled communications, electronic health care infrastructure, and military applications. The RIH methods are proven to be competent in such hyper-sensitive infrastructures where the loss of a single bit of information is not acceptable. In this paper, dual-layered based RIH method using modified least significant bit (LSB) matching has been proposed. The objective of the proposed work is to enhance the embedding efficiency (EE) using dual-layer based embedding strategy and to curtail the distortion caused to the stego-image to improve its quality. At the first layer of embedding, each pixel conceals two bits of information using the proposed modified LSB matching method to produce the intermediate pixel pair (IPP). Further, the IPP is utilized to conceal four bits of information during the next layer of embedding. Experimental study reveals that, the proposed method can embed 1,572,864 bits of secret data with peak signal-to-noise ratio (PSNR) of 47.86 dB, 48.05 dB, 46.51 dB and 48.14 dB, for the respective images. Further, the image quality assessment parameters like structural similarity (SSIM) index and universal image quality index (Q) are as good as the existing literature. Additionally, the proposed method shows excellent anti-steganalysis ability to regular and singular (RS) and pixel difference histogram (PDH) analysis.
An improved method for high hiding capacity based on LSB and PVD
Book chapter, Digital Media Steganography: Principles, Algorithms, and Advances, 2020, DOI Link
View abstract ⏷
In this chapter, we propose an improved image steganography method using the concept of least significant bit (LSB) substitution and pixel value differencing (PVD). The major contributions of the proposed method are (i) increase of the hiding capacity (HC), (ii) avoiding fall off boundary problem (FOBP), and (iii) resistance to regular and singular (RS) and pixel difference histogram (PDH) attack. Initially, the cover image is partitioned into blocks of three consecutive pixels. A reference pixel is chosen from each block, and then the pixels are readjusted. Further, the difference values are found using the smallest pixel of the block. Finally, the stego-pixels are obtained using the difference value and the smallest pixel. Additionally, to avoid FOBP, the pixel shifting process has been carried out for the pixels that fall out of the boundary. The observed result outperforms other existing state-of-the-art methods in terms of the steganographic parameters like peak signal-to-noise ratio (PSNR), HC, the number of fall off boundary pixels, and structural similarity index metric (SSIM).
Digital image steganography and steganalysis: A journey of the past three decades
Review, Open Computer Science, 2020, DOI Link
View abstract ⏷
Steganography is the science and art of covert communication. Conversely, steganalysis is the study of uncovering the steganographic process. The evolution of steganography has been paralleled by the development of steganalysis. In this game of hide and seek, the two player's steganography and steganalysis always want to break the other down. Over the past three decades, research has produced a plethora of remarkable image steganography techniques (ISTs). The major challenge for most of these ISTs is to achieve a fair balance between the metrics such as high hiding capacity (HC), better imperceptibility, and improved security. This study aims to present an exhaustive scrutiny of various ISTs from the classical to recent developments in the spatial domain, with respect to various image steganographic metrics. Further, the current status, recent developments, open challenges, and promising directions in this field are also highlighted.
A Novel n-Rightmost Bit Replacement Image Steganography Technique
Article, 3D Research, 2019, DOI Link
View abstract ⏷
Image steganography is a technique for hiding the secret data in a carrier image. This paper proposes a novel n-right most bit replacement image steganography technique to hide the secret data in an image, where 1 ≤ n ≤ 4. The major objectives of the proposed technique are, (1) improving the peak signal to noise ratio (PSNR), (2) improving the embedding capacity (EC), (3) avoiding the fall of boundary problem (FOBP), and (4) robustness against salt and pepper noise and RS attack. Initially, the n-right most bits for each pixel and the n-bits of the secret data are converted to decimal values. Then, using the difference between these two decimal values the original pixels are readjusted to produce stego-pixels. From the experimental results it is observed that PSNR is higher for lower value of n and the EC is larger for the higher value of n. Furthermore, it is also experimentally investigated that the proposed technique is resistant to steganalytic attacks.
Data hiding using adaptive LSB and PVD technique resisting PDH and RS analysis
Article, International Journal of Electronic Security and Digital Forensics, 2019, DOI Link
View abstract ⏷
This paper proposes an improved data hiding technique using the principle of least significant bit (LSB) substitution and pixel value differencing (PVD). It addresses two issues: 1) the error block problem (EBP); 2) the fall of boundary problem (FOBP). The image is divided into non-overlapping blocks of two consecutive pixels. The blocks are divided into three levels depending upon the pixel value difference. The level of the block and the pixel difference range decides the hiding capacity of a block. The proposed technique has been compared with related existing techniques in terms of parameters like peak signal to noise ratio (PSNR), quality index (Q), hiding capacity, bits per pixel (BPP), and the count of the blocks suffering from FOBP. The experimental results prove that the proposed technique offers better PSNR and hiding capacity as compared to the related existing techniques. Furthermore, the proposed technique is resistant to pixel difference histogram (PDH) analysis and RS analysis.
Dual stego-imaging based reversible data hiding using improved LSB matching
Article, International Journal of Intelligent Engineering and Systems, 2019, DOI Link
View abstract ⏷
Since the inception of the reversible data hiding (RDH) concept, it has been a compelling topic in the field of data hiding. Being reversible, it has the ability to restore the original image followed by the successful retrieval of the secret data, at the receiving side. The concept of the dual stego-image based RDH technique utilizes two identical images of the original image for concealing the secret data, has gained wide compliance. Therefore, someone with both the stego-images can only extract the concealed data. In this paper, two improved dual imaging based RDH techniques, such as (1) dual stego-image based pixel pair LSB matching with reversibility, and (2) dual stego-image based modified LSB matching with reversibility, are proposed. In technique 1, at first two mirrored images are obtained from the original image. Then, using the pair of two consecutive pixels from the original image, the mirrored images pixels are modified using LSB matching technique. Later, these pixel pairs are readjusted to ensure reversibility at the receiving side. Similarly, technique 2 utilizes each original pixel to generate two distinct stegopixels using modified LSB matching. The experimental result shows that the technique 1 maintains excellent peak signal-to-noise ratio (PSNR) of 51.29 dB and 51.30 dB for the two stego-images with hiding capacity (HC) of 524288 bits. At the same time, technique 2 offers 51.19 dB and 49.44 dB of PSNR while exhibiting the equal HC. Further, investigation with various image quality assessment (IQA) metrics like quality index (QI), and structural similarity index (SSIIM) are proven to be competent over the other existing works considered in this paper. In addition, both the proposed techniques have shown excellent anti-steganalytic ability against RS and pixel difference histogram (PDH) attack.
A novel multi stego-image based data hiding method for gray scale image
Article, Pertanika Journal of Science and Technology, 2019,
View abstract ⏷
In this paper, we present a novel multi stego-image based data hiding method using the principle of the modified least significant bit (LSB) matching to improve the embedding capacity (EC) as well as image quality. Initially, each original pixel produces four new pixels. The secret data is hidden in all the four produced pixels. Then the pixels are readjusted to improve the quality of the stego-images. There are four separate stego-images developed from the four different readjusted pixels. Each stego-image hides one bit per pixel. The average peak signal-to-noise ratios (PSNR) for the stego-images are 36.06 dB, 37.88 dB, 39.60 dB and 41.00 dB respectively. Furthermore, the proposed method successfully withstand against RS-steganalysis.
An Optimal Information Hiding Approach Based on Pixel Value Differencing and Modulus Function
Article, Wireless Personal Communications, 2019, DOI Link
View abstract ⏷
This paper proposes an image steganography approach based on pixel value differencing and modulus function (PVDMF) to improve the peak signal-to-noise ratio (PSNR) and hiding capacity (HC). The proposed approach has two variants, (1) PVDMF 1 and (2) PVDMF 2. Both the variants use the difference between a pair of consecutive pixels to embed the secret data based on an adaptive range table. The modulus operations with pixel readjustment have been utilized to reduce the distortion in the stego-image. The experimental results prove that the PVDMF 1 offer higher PSNR and PVDMF 2 offers larger HC as compared to the existing approaches. In addition, the fall off boundary problem which exists in most of the pixel value differencing approaches has been avoided. Furthermore, it has been experimentally verified that the proposed approach is resistant against RS attack.
Pixel Overlapping Image Steganography Using PVD and Modulus Function
Article, 3D Research, 2018, DOI Link
View abstract ⏷
Abstract: This paper proposes an image steganography technique based on the principle of pixel overlapping to improve the embedding capacity (EC) and peak signal-to-noise ratio (PSNR). The proposed technique has two variants: (1) overlapped pixel value differencing with modulus function (OPVDMF), and (2) overlapped pixel value differencing (OPVD). Both the variants operate on pixel blocks of size 1 × 5. The OPVDMF uses the difference between the first four pixels with the 5th pixel for data embedding. Again the pixel adjustment is done to minimize the distortion. The OPVD method divides the block into four sub-blocks with 1st and 5th, 5th and 2nd, 3rd and 5th, 5th and 4th pixels. The proposed technique has been compared with the existing techniques in terms of PSNR, EC, bits per pixel, and execution time. Further, the security of the proposed technique has been verified using RS analysis. Graphical Abstract: [Figure not available: see fulltext.].
An improved data hiding technique using bit differencing and LSB matching
Article, Internetworking Indonesia Journal, 2018,
View abstract ⏷
This paper proposes an improved image steganographic technique based on the principle of modified least significant bit (LSB) substitution and LSB matching, to improve the capacity and peak signal to noise ratio (PSNR). The proposed technique has been divided into 3 variants such as variant-1, variant-2, and variant-3. A block consisting of 2 pixels has been considered for data embedding in all the three variants. The variant-1 initially uses the 6th and 7th bit to hide 2 bits of secret data in the first pixel of a block. Further modification to the pixel is done by ± 1 or 0 in order to hide 2 more bits in a block. Similarly, the variant-2 hides 3 bits and variant-3 hides 2 bits respectively in a block. The experimental results prove that the variant-1 offers better capacity whereas variant-3 offers better peak signal to noise ratio (PSNR). The results with respect to the steganographic parameters such as embedding capacity, PSNR, and the universal image quality index (Q) has been presented and it is found that the performance of the proposed technique is superior compared to existing techniques.
Digital image steganography using PVD and modulo operation
Article, Internetworking Indonesia Journal, 2018,
View abstract ⏷
This paper proposes an image steganographic approach using the principle of pixel value differencing (PVD) and modulo operation (MO). The major contributions of the proposed approach are: (i) increase in peak signal-to-noise ratio (PSNR), (ii) increase in hiding capacity, and (iii) avoidance of fall off boundary problem (FOBP). At first, the image is partitioned into non-overlapping blocks consisting of three consecutive pixels. Then, the secret data is embedded in a block using two phases, (i) pixel difference modulo operation (PDMO) phase, and (ii) average PVD (APVD) readjustment phase. In the first phase, the difference between two consecutive pixels of a block is found and using an adaptive range table and modulo operation the secret data are embedded. In the second phase, the average of the first two stego-pixels of the block and the third pixel is considered for data embedding using PVD approach. The result of the proposed approach has been compared with existing approaches and found to be improved.
Digital image steganography using bit flipping
Sahu A.K., Swain G., Suresh Babu E.
Article, Cybernetics and Information Technologies, 2018, DOI Link
View abstract ⏷
This article proposes bit flipping method to conceal secret data in the original image. Here a block consists of 2 pixels and thereby flipping one or two LSBs of the pixels to hide secret information in it. It exists in two variants. Variant-1 and Variant-2 both use 7th and 8th bit of a pixel to conceal the secret data. Variant-1 hides 3 bits per a pair of pixels and the Variant-2 hides 4 bits per a pair of pixels. Our proposed method notably raises the capacity as well as bits per pixel that can be hidden in the image compared to existing bit flipping method. The image steganographic parameters such as, Peak Signal to Noise Ratio (PSNR), hiding capacity, and the Quality Index (Q.I) of the proposed techniques has been compared with the results of the existing bit flipping technique and some of the state of art article.
Information hiding using group of bits substitution
Article, International Journal on Communications Antenna and Propagation, 2017, DOI Link
View abstract ⏷
The importance of any steganographic approach is based on its capacity and security in data transmission. This article proposes an image steganographic method, called three group of bits substitution (3GBS) which hides 3 bits of secret data in a pixel. Each pixel of an image can hide 3 bits of secret data. The Peak Signal-to-Noise Ratio (PSNR) and hiding capacity are two important parameters to evaluate the strength of any steganographic method. This article compares the PSNR and hiding capacity of the proposed 3GBS method with the existing GBS methods. The experimental results of the proposed method gives a conclusive evidence that the hiding capacity increased significantly with an acceptable visual fidelity of the produced-image.
Performance evaluation parameters of image steganography techniques
Conference paper, International Conference on Research Advances in Integrated Navigation Systems, RAINS 2016, 2016, DOI Link
View abstract ⏷
This paper illustrates the various performance evaluation parameters of image steganography techniques. The performance of a steganographic technique can be rated by three parameters; (i) hiding capacity, (ii) distortion measure and (iii) security. The hiding capacity means the maximum amount of information that can be hidden in an image. It can also be represented as the number of bits per pixel. The distortion is measured by using various metrics like mean square error, root mean square error, PSNR, quality index, correlation, structural similarity index etc. Each of these metrics can be represented mathematically. The security can be evaluated by testing the steganography technique with the steganalysis schemes like pixel difference histogram analysis, RS analysis etc. All these metrics are illustrated with mathematical equations. Finally, some future directions are also highlighted at the end of the paper.
Digital image steganography techniques in spatial domain: A study
Review, International Journal of Pharmacy and Technology, 2016,
View abstract ⏷
Secure digital data communication is always a concern. Cryptography and Steganography are the prominent fields in secure digital data communication. By seeing an innocent image it is very difficult to imagine that, the image is doing the task of a messenger. When an image carries information without the knowledge of an outsider, is called as Image steganography. This paper discusses about various image steganographic techniques in spatial domain such as least significant bit (LSB), pixel value differencing (PVD), Combination of LSB and PVD, Modulus function etc. The comparison among various parameters has been made to determine the efficient technique.
Performance evaluation of energy efficient power models for digital cloud
Jena S.R., Vijayaraja V., Sahu A.K.
Article, Indian Journal of Science and Technology, 2016, DOI Link
View abstract ⏷
Background: To improve quality of service energy-efficiency is one of the key parameters of Cloud service providers. Every year huge amounts of electrical energy consume by Cloud data center which leads to more expense in costs and emission of CO2 to the environment which is unhealthy for us. In this case the need of Green Cloud computing solutions to minimize emission of carbon footprints as well as operational costs is the utmost desire. Objectives: In our research work we have implemented four different power models such as linear model, cubic model, square model and square root model on an Infrastructure-as-a-Service (IaaS) Cloud environment to find out the best one. Methods: Here we considered the CPU utilization and power consumption by enabling virtual machine migration. Then to validate the accuracy of these power models R-squared, Mean Square Error (MSE) have been performed. Finding: We found out that the cubic polynomial model is the most efficient one and consume less power in comparison to the other three models. Application: Hence this model can be used in energy saving applications over Cloud data centers.
A review on LSB substitution and PVD based image steganography techniques
Review, Indonesian Journal of Electrical Engineering and Computer Science, 2016, DOI Link
View abstract ⏷
There has been a tremendous growth in Information and Communication technologies during the last decade. Internet has become the dominant media for data communication. But the secrecy of the data is to be taken care. Steganography is a technique for achieving secrecy for the data communicated in Internet. This paper presents a review of the steganography techniques based on least significant bit (LSB) substitution and pixel value differencing (PVD). The various techniques proposed in the literature are discussed and possible comparison is done along with their respective merits. The comparison parameters considered are, (i) hiding capacity, (ii) distortion measure, (iii) security, and (iv) computational complexity.
