Search Results (1,282)

QR codes are widely used in digital authentication, mobile payment, access control, and information exchange, making QR code-based covert communication and hidden message delivery relevant to cybersecurity applications. This study proposes a QR code-based information-hiding method via module content modification based on corner-pixel grayscale adjustment. Secret bits are embedded into a QR code image through subtle grayscale adjustments of module corner pixels. To reduce visually noticeable embedding artifacts, embedding in adjacent module corners that share identical original grayscale values is prohibited, thereby preventing noticeable local contrast that could reveal hidden content. Gradient-based smoothing is further applied to suppress contrast artifacts within module interiors after embedding. A theoretical analysis of bilinear interpolation and quarter subregion averages is presented to justify the proposed extraction thresholds. Experiments across multiple QR code versions, error-correction levels, and embedding parameters show that the proposed method achieves an average embedding rate of approximately 1.9 bits per module while generally maintaining QR code readability across the tested commercial scanning devices. Exploratory random-noise experiments further show that hidden-message recovery remains possible up to the observed highest tolerable noise rates in the tested realizations. A comparison with four representative methods indicates that the proposed method achieves substantially higher embedding rates under the adopted comparison framework, demonstrating its potential for high-capacity QR code information-hiding while enabling successful hidden-message recovery under the evaluated conditions. Full article

Image compression has evolved from human-centric perceptual coding toward support for diverse machine vision applications, requiring modern codecs to serve both human viewing and downstream tasks in closed-set settings (where target tasks are incorporated during training) and open-set settings (where previously unseen tasks arise at test time). While recent learned compression methods jointly optimize perceptual quality and closed-set task performance, they often fail to generalize to unseen open-set tasks due to fixed training assumptions and objectives. Our prior work, LoRA-comp (Low-Rank Adaptation Compression), effectively addresses open-set challenges via instance-specific test-time fine-tuning (TTFT) without requiring task-specific pre-training. Nevertheless, its fixed LoRA architecture, which assigns a uniform rank across all layers, often leads to suboptimal instance-level performance. Moreover, allocating the same number of training epochs to every instance introduces unnecessary encoding-time overhead. To address these challenges, we propose Instance-Specific Early Stopping with Dynamic Rank Adaptation (ISER), which extends LoRA-comp. Building upon the LoRA-comp–based instance-specific adaptation framework, ISER introduces (i) instance-specific early stopping (ISES) combined with a multi-scale training strategy (MSTS) to reduce TTFT overhead and (ii) instance-specific dynamic rank adaptation (ISRA) to tailor the LoRA architecture per instance. Experiments demonstrate that ISER consistently outperforms competing methods. Compared to LoRA-comp, ISER achieves up to a 7% BD-Rate improvement and up to a 44% reduction in encoding time. Moreover, ISER achieves up to a 98% relative improvement in BD-Rate gain and up to a 19.2% reduction in decoding time over TransTIC. Full article

This work provides a structured synthesis of visual cryptography, a secret sharing technique that enables image reconstruction only when a specific number of shares are combined, with decryption performed visually by overlaying the shares. Although conceptually simple and distinctive in its reliance on the human visual system rather than complex computation, this method has been predominantly studied from theoretical and construction-oriented perspectives. The work consolidates fundamental concepts, mathematical foundations, operational principles, and security considerations of selected schemes, providing a common context for analyzing their characteristics. Particular attention is given to known attack vectors, including information leakage from individual shares and integrity violations caused by forged shares, together with corresponding mitigation approaches reported in the literature. In addition to this synthesis, the work presents an experimental investigation of the $\varphi$ correlation coefficient and its behavior for genuine and forged shares. Experiments conducted on a dataset of 100 images show that genuine–genuine share pairs consistently exhibit higher mean $\varphi$ correlation values than genuine–forged pairs, although the observed values depend on the underlying scheme. While these results suggest that $\varphi$ correlation may provide useful information for share authenticity analysis, the experiment used the same forged target image throughout the dataset, limiting the variation of the forged samples and potentially making the observed differences partly dependent on the selected target image. Finally, three Proof-of-Concept application scenarios demonstrate possible integrations of visual cryptography into QR code security, physical document verification, and IoT access control, illustrating its potential use in practical security systems. Full article

Background: Brain MRI in patients with autism spectrum disorder (ASD) is generally considered when additional neurologic, developmental, or syndromic concerns warrant structural evaluation. In the study practice, ASD alone was not used as the reason to obtain MRI. We examined documented MRI use and broad result coding while explicitly separating selection for imaging from abnormal-result status. Methods: This retrospective specialty-practice cohort included 1884 patients with ASD. The primary outcome was the proportion of documented MRI examinations coded abnormal. Secondary analyses examined associations of age, sex, seizure history, EEG status, and genetic test status with abnormal versus normal MRI coding and with documented MRI utilization. MRI categories and the pediatric restriction were exploratory and sensitivity analyses. Results: An MRI result was recorded for 704 patients (37.4%; 95% CI 35.2–39.6); 322 were coded abnormal (45.7%; 95% CI 42.0–49.5). The abnormal-result models had low in-sample discrimination (AUC 0.528 and 0.553), and adjusted confidence intervals for all measured variables included 1.00; these results do not demonstrate equivalence between clinical groups. Seizure history was associated with documented imaging in the full cohort (aOR 2.47, 95% CI 1.94–3.14), and abnormal EEG was associated in the exploratory complete-case utilization model (aOR 2.83, 95% CI 1.93–4.14). Conclusions: Broad MRI abnormalities were frequent among patients selected for imaging because of additional clinical concerns, but a broad abnormal code is not equivalent to clinically actionable yield. The routinely captured variables available in this dataset were insufficient to distinguish abnormal from normal MRI coding. These findings favor individualized MRI decisions based on the clinical question prompting imaging and motivate future studies with richer neurologic and developmental phenotyping. Full article

Intelligent methods such as machine learning, deep learning, and related reasoning-based techniques are widely credited with transforming digital forensics, yet it remains unclear where that transformation has actually taken hold. This study presents a systematic mapping review, conducted in accordance with the PRISMA Extension for Scoping Reviews (PRISMA-ScR), that treats the adoption of intelligent methods not as an assumption but as a variable to be measured. Following searches across eight bibliographic databases, 82 primary studies were mapped onto six digital forensics domains and coded by intelligence method type, enabling a domain-by-method analysis of the field. The results reveal markedly uneven penetration. Intelligent methods dominate multimedia forensics (93% of included studies) and mobile and IoT forensics (77%), yet remain the exception in frameworks and governance (33%), imaging and acquisition (28%), memory forensics (25%), and data reduction (21%), where classical, deterministic techniques still prevail. Resolved onto a five-level maturity ladder, the domains differ not only in how much intelligence they have adopted but in its kind: some reach deep learning and explainable reasoning while others advance only to automation, and several skip intermediate stages entirely. These findings recast intelligence in digital forensics as a set of unevenly developed capabilities rather than a uniform pipeline, and identify where learning-based research has the furthest still to travel. Full article

Background: GLP-1 receptor agonists (GLP-1 RAs) are increasingly used for obesity and metabolic disease, but their use in people with chronic pancreatitis is not a chronic pancreatitis-directed indication and direct evidence is limited. We described recorded outcomes among non-diabetic adults carrying a chronic pancreatitis diagnosis who did or did not have recorded GLP-1 RA exposure. Methods: We performed a retrospective propensity score-matched cohort study using the TriNetX Collaborative Network. Chronic pancreatitis was identified from a recorded diagnosis; supporting imaging, histological, functional, or specialist-confirmation criteria were unavailable. The exposed cohort included patients receiving dulaglutide, semaglutide, or tirzepatide (n = 1441 before matching), and the comparator cohort included patients without recorded GLP-1 RA exposure (n = 142,047 before matching). One-to-one propensity score matching generated 1422 patients in each cohort. Outcomes were assessed from 1 to 1095 days after the index date using risk comparisons and time-to-event analyses. Results: Mean follow-up after matching was 458.8 days in the exposed cohort and 664.4 days in the comparator cohort. Recurrent acute pancreatitis was recorded in 19/799 (2.4%) versus 76/704 (10.8%) patients (HR 0.247, 95% CI 0.149–0.409), pancreatic cancer in 10/1373 (0.7%) versus 40/1345 (3.0%) (HR 0.255, 95% CI 0.128–0.511), and all-cause mortality in 21/1419 (1.5%) versus 157/1416 (11.1%) (HR 0.167, 95% CI 0.105–0.263). A similar direction was observed across several coded outcomes. Vitamin D deficiency showed a higher hazard (HR 1.556, 95% CI 1.129–2.145), although its risk comparison was not significant. Conclusions: These estimates describe outcomes in a selected treatment-exposed phenotype of chronic pancreatitis. The findings are hypothesis-generating and should not guide prescribing decisions. Further prospective studies are required to confirm this hypothesis. Full article

Background: Single coronary artery (SCA) is a rare and anatomically heterogeneous congenital coronary anomaly in which the entire coronary circulation arises from a single aortic ostium. Its clinical significance depends on the ostial and proximal morphology, coronary course, associated coronary disease, symptoms and, where indicated, functional findings. Methods: PubMed/MEDLINE was searched with free-text terms from 24 December 2025 to 12 July 2026 and rerun on 9 August 2026; relevant reference lists were screened manually. Results: SCA-specific evidence consisted mainly of case reports, small series, and descriptive imaging studies. A single ostium or Lipton code does not fully describe the relevant anatomy. Reports should state whether a true common trunk is present, map the course and supplied territory of each major branch, and document proximal morphology and acquired coronary disease. Coronary computed tomography angiography (CCTA) is central to anatomical assessment in many adults. Functional testing is most informative when the method addresses the suspected mechanism. Conclusions: SCA should be interpreted by integrating detailed anatomy, symptoms, coexisting coronary disease and, when appropriate, mechanism-matched functional findings. The three-domain structure used here supports reporting and multidisciplinary discussion, but it is not a validated risk model or treatment algorithm. Prospective evidence on SCA-specific outcomes and management remains limited. Full article

Screen content images (SCIs) exhibit complex structural heterogeneity, rendering traditional statistics-based natural scene image quality assessment (NR-IQA) metrics ineffective. Although deep learning models achieve high prediction accuracy, their prohibitive computational demands preclude deployment in latency-sensitive industrial scenarios. While existing handcrafted lightweight SCI-IQA metrics reduce computational overhead, most rely on unsegmented global feature pooling or holistic edge statistics (e.g., edge histograms or Fisher vector coding), thereby diluting locally critical text-edge distortions in vast homogeneous backgrounds. To address this limitation, we propose an ultra-lightweight, deep-learning-free NR-IQA framework centered on human visual masking. Unlike existing lightweight methods, our approach explicitly employs dual-scale Canny edge operators to partition SCIs into edge-sensitive and flat background regions. Guided by this visual prior, structural degradations and micro-compression textures are extracted region-wise using Sobel gradients and uniform local binary patterns (LBPs) and aggregated with global Commission Internationale de I’Eclairage L*a*b*(CIELAB) color statistics into a compact 60-dimensional descriptor. A grid-search-optimized Support Vector Regression (SVR) maps these features to subjective quality scores. Extensive cross-validation on the SIQAD and SCID datasets demonstrates that our metric outperforms existing handcrafted lightweight SCI metrics and traditional NSS models, while achieving accuracy competitive with representative full-reference metrics. Consuming only 79.3 ms per image on a standard CPU, it offers a practical accuracy–efficiency trade-off for resource-constrained periodic quality monitoring. Full article

Deep joint source-channel coding (DeepJSCC) has emerged as a promising paradigm for semantic-aware wireless image transmission, achieving strong performance under challenging channel conditions. However, MSE-trained DeepJSCC systems typically achieve high peak signal-to-noise ratio (PSNR) values but suppress high-frequency details, resulting in perceptually blurry reconstructions that fail to capture fine textures and edge information. Existing perceptual enhancement approaches for JSCC systems face significant practical limitations: full transceiver redesign methods require replacing both the transmitter and the receiver with large models (19–31 million parameters), incurring substantial deployment costs; diffusion-based refinement approaches require over 1700 million additional parameters and introduce inference latency exceeding 13 s, rendering them unsuitable for latency-constrained wireless applications; and generic image restoration networks lack channel state awareness and cannot adapt to varying signal-to-noise ratio (SNR) conditions. This paper proposes a lightweight receiver-only perceptual enhancer designed for use with frozen DeepJSCC backbones. The proposed module adopts residual learning with feature-wise linear modulation (FiLM)-based SNR-adaptive modulation to dynamically adjust the enhancement strength under varying channel conditions. A radially weighted FFT magnitude loss is further introduced to guide high-frequency recovery. The enhancer adds only 0.29 million trainable parameters (<1% of the backbone) and requires neither transmitter modification nor backbone retraining. Extensive experiments on the Kodak24 and DIV2K datasets demonstrate a 34.4–37.5% LPIPS reduction over the frozen DeepJSCC baseline under AWGN channels. Supplementary robustness evaluations further show a 30–33% LPIPS reduction under Rayleigh fading, and stable generalization to unseen SNR levels. The receiver-side decoder-plus-enhancer pipeline requires 43 ms at 768 × 512 resolution, corresponding to approximately 23 frames per second. Full article

Under moderate-to-high wind conditions, low-frequency wind direction modulation signals in X-band marine radar images are strongly coupled with wave textures, sea clutter, and blind-zone interference, which degrades wind direction retrieval accuracy. To address this problem, this study proposes a wind direction retrieval method based on two-dimensional dual-tree complex wavelet transform (2D-DTCWT), convolutional sparse coding (CSC), and maximum-energy radial rings. First, 2D-DTCWT is used to suppress wave textures and local noise in the wavelet domain while enhancing low-frequency wind direction modulation signals. Then, K–singular value decomposition (K-SVD) learns the energy distribution characteristics of wind signals, and CSC obtains the spatial response distribution of wind energy in radar images. Finally, the maximum-energy radial ring is adaptively identified, and azimuthal energy statistics within this ring are fitted using a cosine-squared function. The proposed method was evaluated using X-band marine radar data collected during sea trials in the coastal waters of Zhejiang, China. On the 900-sample main validation dataset, the proposed method achieved the highest correlation coefficient (CC) of 0.85 and an overall root mean square error (RMSE) of 4.24°, reducing the RMSE by 43.0% and 66.7% compared with conventional single-curve fitting and extended-bow-heading DWT, respectively. The results demonstrate improved robustness under both upwind and downwind blind-zone conditions. Full article

Background: Assessment of inflammatory activity in Crohn’s disease (CD) remains challenging. Bowel wall vascularization is an established marker of transmural inflammation and is typically evaluated using semiquantitative ultrasound scoring systems. In contrast, quantitative vascularization analysis is not yet widely used in clinical practice. To compare these approaches, this study evaluated semiquantitative and quantitative methods in the same patient cohort. Methods: This prospective single-center study included 50 patients with CD and sonographically detectable bowel wall thickening. Bowel wall vascularization was assessed using power Doppler (PD), color-coded superb microvascular imaging (cSMI) and monochrome mode superb microvascular imaging (mSMI). Video sequences were independently evaluated by three blinded readers using the Limberg classification. Quantitative vascularization was assessed using a vendor-derived vascular index and retrospective ImageJ analysis. Crohn’s Disease Activity Index (CDAI) and laboratory parameters were documented. Results: Interobserver agreement differed between imaging modalities. Krippendorff’s α values were 0.735 (PD), 0.655 (cSMI) and 0.527 (mSMI), indicating higher reliability for PD than for the SMI techniques. According to Fleiss’ κ (0.480 (PD), 0.376 (cSMI) and 0.367 (mSMI)), agreement was moderate for PD and fair for both SMI techniques. Complete concordance among all three readers was observed in 24 (PD), 18 (cSMI), and 22 (mSMI) patients, respectively. SMI-based techniques consistently resulted in higher Limberg grades compared to PD. Exploratory analyses showed strong correlations between different quantitative vascularization parameters (ρ up to 0.898, p < 0.001), but only moderate correlations with fecal calprotectin (FC) (ρ up to 0.389 with p = 0.017). Significant negative correlations were observed between vascular indices and body mass index (BMI) and skin-to-bowel distance (ρ up to −0.614, p < 0.001). The same trend was observed in the semiquantitative Limberg classification. Conclusions: Semiquantitative assessment of bowel wall vascularization seems to be limited by interobserver variability and Doppler technique, with SMI yielding higher Limberg grades than conventional Doppler. Standardized quantitative approaches may improve objectivity, but their reproducibility requires further validation. Patient-related factors should be considered when interpreting vascularization findings. Full article

In the contemporary digital era, image watermarking is essential for protecting intellectual property due to the widespread unauthorized distribution of digital content. In this work, a robust and efficient image watermarking scheme for copyright protection is proposed. The method integrates wavelet packet decomposition (WPD) with an ensemble of bagged tree classifiers, forming the BT-WPD framework. In the proposed approach, wavelet packet coefficients extracted from each color channel are reorganized into structured batches that capture spatial frequency characteristics, enabling effective watermark embedding in the WPD domain guided by the bagged tree ensemble model. Experimental results demonstrate that the proposed method achieves high imperceptibility, with a peak signal-to-noise ratio (PSNR) exceeding 60 dB, while maintaining strong robustness against various image processing attacks. The method also exhibits low computational complexity during watermark extraction, making it suitable for practical applications. Furthermore, the framework is extended to support Quick Response (QR) code watermark embedding, demonstrating enhanced robustness and versatility for copyright protection in digital media systems. Full article

Unlike natural image deblurring, which primarily emphasizes perceptual quality and pixel-level fidelity, Quick Response (QR) code deblurring must retain decoding-critical structures to guarantee successful decoding. QR codes contain regular binary module grids and functional patterns with sharp boundaries, providing a strong structural prior for restoration. However, most existing learning-based QR restoration methods capture QR-specific structural information via implicit feature learning. To address this limitation, we propose an Edge-Guided Attention Block (EGAB), which explicitly extracts multi-directional edge priors and injects them into the query–key correlations of Transformer attention. Based on EGAB, we develop an Edge-Guided Restormer (EG-Restormer) for restoring severely blurred QR codes. For mildly blurred inputs, we introduce a Lightweight and Efficient Network (LENet) that performs fast restoration with low computational overhead. We further integrate EG-Restormer and LENet into an Adaptive Dual-network (ADNet), which selects the appropriate restoration branch according to the input blur level. Extensive experiments demonstrate the effectiveness of the proposed framework. EG-Restormer boosts the decoding rate by 8.67 percentage points under GoPro-only training and achieves the highest decoding rate among the evaluated methods after QRData fine-tuning. Moreover, ADNet reduces average inference latency by 19% while maintaining comparable decoding performance. These results suggest that explicit edge prior modeling enhances the recovery of structures critical for decoding, while adaptive routing provides an effective balance between decoding accuracy and computational efficiency. Full article

Snapshot compressive multi-view spectral imaging (SC-MVSI) multiplexes view-spectral information into a single coded measurement, enabling compact acquisition with a two-dimensional detector. Because each reconstructed channel corresponds to both a selected spectral response and a view direction, direct cross-channel modeling at identical pixel coordinates can introduce structural mismatch caused by view-dependent displacement. This paper proposes a reference-guided view-aligned nonlocal low-rank tensor reconstruction method for SC-MVSI. The reconstruction is formulated as a coded inverse problem and solved using the alternating direction method of multipliers (ADMM) in a variable-splitting framework. In the prior update, a reference tensor guides block-level patch alignment before nonlocal tensor grouping, and the resulting fourth-order tensor groups are regularized by canonical polyadic (CP) low-rank approximation. Experiments on eight synthesized multispectral light-field scenes show that the proposed method achieves the highest average PSNR of 33.61 dB and the lowest average CAE of 5.69 degrees among the compared baselines, while obtaining the second-highest average SSIM of 0.8823. Real-system experiments further provide a qualitative demonstration of applying the proposed reconstruction framework to captured coded measurements. Full article

Artificial Intelligence (AI) is being integrated into the architectural profession via processes such as generating images based on design parameters; providing support to help create written descriptions of designs; assisting in the visualisation of buildings before they are built; allowing for greater use of parameters in construction; providing documentation to help complete projects in record time; and assisting in making design decisions. The success of integrating AI into practice depends not only on the use of tools but also on firms’ maturity in integrating AI into their workflows, employees’ capabilities, project teams’ operational efficiency, the impact of investment decisions, and the overall governance of the profession. This research evaluates the maturity of AI integration in architectural practice in the Kurdistan Region of Iraq. To do so, it develops and operationalises the AI Integration Maturity Index (AIMI), an eight-component formative composite index scored 0–37 and organised into five maturity bands (Non-adopter, Exploratory, Occasional, Integrated, and Advanced Strategic). The index comprises adoption, usage, diversity of tools, breadth of workflows, project penetration, staff involvement, training/capacity building, and governance/strategic focus. The AIMI is treated as a literature-derived formative diagnostic tool rather than a universal weighting standard, and was developed through a structured literature synthesis, expert pilot review, and internal-structure validation. Accordingly, its component logic and internal statistics are reported as a transparency and coherence check rather than as reflective reliability claims: Cronbach’s alpha (0.922) is presented descriptively to show component co-movement given the formative specification, while inter-coder reliability (kappa = 0.96) supports the qualitative benchmark coding. The study employs a mixed-methods descriptive comparative methodology consisting of a structured survey instrument administered to 100 architectural firms operating in the local market and structured asynchronous text-based interviews with 10 international architectural firms, comparing the resulting profiles with selected, generally accepted benchmarks and with a sample of leading firms engaged in architectural practice worldwide. On average, total AIMI scores in the local sample were 18.14 out of 37, indicating that local firms are broadly adopting and using AI (82% currently use AI on a regular or occasional basis). By contrast, the international sample yielded a mean AIMI score of 28.60, indicating that local firms exhibit a significantly lower level of maturity than the international benchmark, with the largest gaps in staff involvement, project penetration, and overall engagement with AI use. The paper concludes that the primary challenge facing architectural firms in the Kurdistan Region is no longer basic awareness or technological infrastructure, but rather the transition from broad and superficial AI adoption to a systematic, structured, project-based, and well-governed integration of AI within the architectural profession. Full article