Search Results (192)

Single image super-resolution is the inverse problem of reconstructing a high-resolution image from its low-resolution counterpart. Although recent Transformer-based architectures leverage global context integration to improve reconstruction quality, they often overlook frequency-specific characteristics, resulting in the loss of high-frequency information. To address this limitation, we propose the Dual Frequency Adaptive Network (DFAN). DFAN first decomposes the input into low- and high-frequency components via Stationary Wavelet Transform. In the low-frequency branch, Swin Transformer layers restore global structures and color consistency. In contrast, the high-frequency branch features a dedicated module that combines Directional Convolution with Residual Dense Blocks, precisely reinforcing edges and textures. A frequency fusion module then adaptively merges these complementary features using depthwise and pointwise convolutions, achieving a balanced reconstruction. During training, we introduce a frequency-aware multi-term loss alongside the standard pixel-wise loss to explicitly encourage high-frequency preservation. Extensive experiments on the Set5, Set14, BSD100, Urban100, and Manga109 benchmarks show that DFAN achieves up to +0.64 dBpeak signal-to-noise ratio, +0.01 structural similarity index measure, and −0.01learned perceptual image patch similarity over the strongest frequency-domain baselines, while also delivering visibly sharper textures and cleaner edges. By unifying spatial and frequency-domain advantages, DFAN effectively mitigates high-frequency degradation and enhances SISR performance. Full article

Modern aerospace engineering places increasing emphasis on materials that combine low weight with high mechanical performance. Fiber metal laminates (FMLs), which merge metal layers with fiber-reinforced composites, meet this demand by delivering improved fatigue resistance, impact tolerance, and environmental durability, often surpassing the performance of their constituents in demanding applications. Despite these advantages, inspecting such thin, layered structures remains a significant challenge, particularly when they are difficult or impossible to access. As with any new invention, they always come with challenges. This study examines the effectiveness of the fundamental anti-symmetric Lamb wave mode (A0) in detecting weak interfacial defects within Carall laminates, a type of hybrid fiber metal laminate (FML). Delamination detectability is analyzed in terms of strong wave dispersion observed downstream of the delaminated sublayer, within a region characterized by acoustic distortion. A three-dimensional finite element (FE) model is developed to simulate mode trapping and full-wavefield local displacement. The approach is validated by reproducing experimental results reported in prior studies, including the author’s own work. Results demonstrate that the A0 mode is sensitive to delamination; however, its lateral resolution depends on local position, ply orientation, and dispersion characteristics. Accurately resolving the depth and extent of delamination remains challenging due to the redistribution of peak amplitude in the frequency domain, likely caused by interference effects in the acoustically sensitive delaminated zone. Additionally, angular scattering analysis reveals a complex wave behavior, with most of the energy concentrated along the centerline, despite transmission losses at the metal-composite interfaces in the Carall laminate. The wave interaction with the leading and trailing edges of the delaminations is strongly influenced by the complex wave interference phenomenon and acoustic mismatched regions, leading to an increase in dispersion at the sublayers. Analytical dispersion calculations clarify how wave behavior influences the detectability and resolution of delaminations, though this resolution is constrained, being most effective for weak interfaces located closer to the surface. This study offers critical insights into how the fundamental anti-symmetric Lamb wave mode (A0) interacts with delaminations in highly attenuative, multilayered environments. It also highlights the challenges in resolving the spatial extent of damage in the long-wavelength limit. The findings support the practical application of A0 Lamb waves for structural health assessment of hybrid composites, enabling defect detection at inaccessible depths. Full article

Mammography is the main method for early detection of breast cancer, which is still a major global health concern. However, inter-reader variability and the inherent difficulty of interpreting subtle radiographic features frequently limit the accuracy of diagnosis. A thorough assessment of deep convolutional neural networks (CNNs) for automated mammogram classification is presented in this work, along with the introduction of two innovative multi-view integration techniques: Dual-Branch Ensemble (DBE) and Merged Dual-View (MDV). By setting aside two datasets for out-of-sample testing, we evaluate the generalizability of the model using six different mammography datasets that represent various populations and imaging systems. We compare a number of cutting-edge architectures on both individual and combined datasets, including ResNet, DenseNet, EfficientNet, MobileNet, Vision Transformers, and VGG19. Both MDV and DBE strategies improve classification performance, according to experimental results. VGG19 and DenseNet both obtained high ROC AUC scores of 0.9051 and 0.7960 under the MDV approach. DenseNet demonstrated strong performance in the DBE setting, achieving a ROC AUC of 0.8033, while ResNet50 recorded a ROC AUC of 0.8042. These enhancements demonstrate how beneficial multi-view fusion is for boosting model robustness. The impact of domain shift is further highlighted by generalization tests, which emphasize the need for diverse datasets in training. These results offer practical advice for improving CNN architectures and integration tactics, which will aid in the creation of trustworthy, broadly applicable AI-assisted breast cancer screening tools. Full article

Graph theory models relationships by representing entities as vertices and their interactionsas edges. To handle directionality and multiple head–tail assignments, various extensions—directed, bidirected, and multidirected graphs—have been introduced, with the multidirected graph unifying the first two. In this work, we further enrich this landscape by proposing the Multidirected hypergraph, which merges the flexibility of hypergraphs and superhypergraphs to describe higher-order and hierarchical connections. Building on this, we introduce five uncertainty-aware Multidirected frameworks—fuzzy, neutrosophic, plithogenic, rough, and soft multidirected graphs—by embedding classical uncertainty models into the Multidirected setting. We outline their formal definitions, examine key structural properties, and illustrate each with examples, thereby laying groundwork for future advances in uncertain graph analysis and decision-making. Full article

Traditional cattle health monitoring systems rely on centralized data collection, posing significant challenges related to data privacy, network connectivity, model reliability, and trust. This study introduces a novel, nature-inspired federated learning (FL) framework for cattle health monitoring, integrating blockchain to ensure model validation, system resilience, and reputation management. Inspired by the fission–fusion dynamics of elephant herds, the framework adaptively forms and merges subgroups of edge nodes based on six key parameters: health metrics, activity levels, geographical proximity, resource availability, temporal activity, and network connectivity. Graph attention networks (GATs) enable dynamic fission, while Density-Based Spatial Clustering of Applications with Noise (DBSCAN) supports subgroup fusion based on model similarity. Blockchain smart contracts validate model contributions and ensure that only high-performing models participate in global aggregation. A reputation-driven mechanism promotes reliable nodes and discourages unstable participants. Experimental results show the proposed framework achieves 94.3% model accuracy, faster convergence, and improved resource efficiency. This adaptive and privacy-preserving approach transforms cattle health monitoring into a more trustworthy, efficient, and resilient process. Full article

ECG monitoring is central to the early detection of cardiac abnormalities. We compared 28 manually selected signal-specific features with 159 automatically extracted signal-independent descriptors from the MIT BIH Arrhythmia Database. ANOVA reduced features to the 10 most informative attributes, which were evaluated alone and in combination with the signal-specific features using Random Forest, SVM, and deep neural networks (CNN, RNN, ANN, LSTM) under an interpatient 80/20 split. Merging the two feature groups delivered the best results: a 128-layer CNN achieved 100% accuracy. Power profiling revealed that deeper models improve accuracy at the cost of runtime, memory, and CPU load, underscoring the trade-off faced in edge deployments. The proposed hybrid feature strategy provides beat-by-beat classification with a reduction in the number of features, enabling real-time ECG screening on wearable and IoT devices. Full article

Target detection in resource-constrained environments faces multiple challenges such as the use of camouflage, diverse target sizes, and harsh environmental conditions. Moreover, the need for solutions suitable for edge computing environments, which have limited computational resources, adds complexity to the task. To meet these challenges, we propose MSCDNet (Multi-Scale Context Detail Network), an innovative and lightweight architecture designed specifically for efficient target detection in such environments. MSCDNet integrates three key components: the Multi-Scale Fusion Module, which improves the representation of features at various target scales; the Context Merge Module, which enables adaptive feature integration across scales to handle a wide range of target conditions; and the Detail Enhance Module, which emphasizes preserving crucial edge and texture details for detecting camouflaged targets. Extensive evaluations highlight the effectiveness of MSCDNet, which achieves 40.1% mAP50-95, 86.1% precision, and 68.1% recall while maintaining a low computational load with only 2.22 M parameters and 6.0 G FLOPs. When compared to other models, MSCDNet outperforms YOLO-family variants by 1.9% in mAP50-95 and uses 14% fewer parameters. Additional generalization tests on VisDrone2019 and BDD100K further validate its robustness, with improvements of 1.1% in mAP50 on VisDrone and 1.2% in mAP50-95 on BDD100K over baseline models. These results affirm that MSCDNet is well suited for tactical deployment in scenarios with limited computational resources, where reliable target detection is paramount. Full article

Prominent target detection in optical remote sensing images (RSI-SOD) focuses on segmenting key targets that capture human attention. However, most SOD methods prioritize detection accuracy at the cost of memory. Complex backgrounds, occlusions, and noise distort segmented target boundaries, while large memory demands increase computational cost, and reduced memory impairs segmentation accuracy. To address these challenges, we integrate edge enhancement and attention mechanisms with multi-path complementary features for salient object detection in remote sensing images (IEAM), aiming to improve salient target accuracy, boundary detection, and memory efficiency. The architecture utilizes a structured feature fusion strategy, combining spatial channel attention mechanisms with adaptive merging to enhance multi-scale feature representation and suppress background noise. The Spatially Adaptive Edge Embedded Module (SAEM) refines object boundary perception, the SCAAP module dynamically selects relevant spatial and channel features while balancing adaptive and maximal pooling, and the Spatial Adaptive Guidance (SAG) module enhances feature localization in cluttered environments to mitigate semantic dilution in U-shaped networks. Extensive experiments on the EORSSD and ORSSD benchmark datasets demonstrate that IEAM outperforms 21 state-of-the-art methods, achieving an inference speed of 48 FPS at 103.2 G FLOP, making it suitable for real-time applications. The proposed model is robust and excels in multiple aspects. Full article

Bringing together the Internet of Things (IoT), LLMs, and Federated Learning (FL) offers exciting possibilities, creating a synergy to build smarter, privacy-preserving distributed systems. This review explores the merging of these technologies, particularly within edge computing environments. We examine current architectures and practical methods enabling this fusion, such as efficient low-rank adaptation (LoRA) for fine-tuning large models and memory-efficient Split Federated Learning (SFL) for collaborative edge training. However, this integration faces significant hurdles: the resource limitations of IoT devices, unreliable network communication, data heterogeneity, diverse security threats, fairness considerations, and regulatory demands. While other surveys cover pairwise combinations, this review distinctively analyzes the three-way synergy, highlighting how IoT, LLMs, and FL working in concert unlock capabilities unattainable otherwise. Our analysis compares various strategies proposed to tackle these issues (e.g., federated vs. centralized, SFL vs. standard FL, DP vs. cryptographic privacy), outlining their practical trade-offs. We showcase real-world progress and potential applications in domains like Industrial IoT and smart cities, considering both opportunities and limitations. Finally, this review identifies critical open questions and promising future research paths, including ultra-lightweight models, robust algorithms for heterogeneity, machine unlearning, standardized benchmarks, novel FL paradigms, and next-generation security. Addressing these areas is essential for responsibly harnessing this powerful technological blend. Full article

In this study, based on total column ozone observations from eight Antarctic stations, we evaluate the applicability of ERA5, C3S-MSR, MERRA-2, and JRA-55 reanalysis datasets and the NIWA-BS merged satellite dataset, in terms of interannual variation and long-term trend, using the correlation coefficient (R), root-mean-square error (RMSE), interannual variability skill score (IVS), and linear trend bias (TrBias). The results show that for interannual variation, C3S-MSR performs well at multiple stations, while JRA-55 performs poorly at most stations, especially Marambio, Rothera, and Faraday/Vernadsky, which are located at lower latitudes on the Antarctic Peninsula. Additionally, all datasets show significantly higher RMSE at Dumont D’Urville and Arrival Heights, which generally are located around the edge of the Antarctic stratospheric vortex where total column ozone values are more variable and on average larger than in the core of the vortex. The comprehensive ranking results show that C3S-MSR performs the best, followed by ERA5 and NIWA-BS, with MERRA-2 and JRA-55 ranking lower. For the long-term trend, each of the datasets has large bias values at Arrival Heights, and the absolute TrBias values of JRA-55 are larger at three stations on the Antarctic Peninsula. The overall averaged results show that C3S-MSR and NIWA-BS have the smallest absolute TrBias, and perform best in reflecting the Antarctic ozone trends, while ERA5 and JRA-55 significantly overestimate the Antarctic ozone recovery trend and perform poorly. Based on our analysis, the C3S-MSR dataset can be recommended to be prioritized when analyzing the interannual variations in Antarctic stratospheric ozone, and both the C3S-MSR reanalysis and NIWA-BS datasets should be prioritized for trend analysis. Full article

To improve the performance of the image semantic segmentation algorithm and make the algorithm achieve a better balance between accuracy and real-time performance when segmenting images, this paper proposes a real-time image semantic segmentation model based on an improved DeepLabv3+ network. First, the MobileNetV2 model with less computational overhead and number of parameters is selected as the backbone network to improve the segmentation speed; then, the Feature Enhancement Module (FEM) is introduced to several shallow features with different scale sizes in MobileNetV2, and then these shallow features are fused to improve the utilization rate of the model encoder on the edge information, to retain more detailed information and to improve the network’s feature representation ability for complex scenes; finally, to address the problem that the output feature maps of Atrous Spatial Pyramid Pooling (ASPP) module do not pay enough attention to detailed information after merging, the FEM attention mechanism is introduced on the feature maps processed by the ASPP module. The algorithm in this study achieves 76.45% for mean intersection over union (mIoU) accuracy with 29.18 FPS real-time performance in the PASCAL VOC2012 Augmented dataset; and 37.31% mIoU accuracy with 23.31 FPS real-time performance in the ADE20K dataset. The experimental results show that the algorithm in this study achieves a good balance between accuracy and real-time performance, and its image semantic segmentation performance is significantly improved compared to DeepLabv3+ and other existing algorithms. Full article

The Middle Yangtze River Urban Agglomeration, a critical ecological barrier in China, faces escalating pressures from rapid urbanization and climate change, leading to fragmented landscapes and degraded ecosystem services. To address the synergistic challenges of ecological protection and risk management, this paper takes the urban agglomeration in the middle reaches of the Yangtze River as the study area, and obtains the source patches through morphological spatial pattern analysis. Based on the spatial distribution of risky source areas, ecological blind zones are cut down by optimizing buffer zones and merging fragmented patches. Finally, a composite ecological network is constructed through circuit theory superimposed on the dual network method. The results showed that (1) there are 16 ecological source patches and 16 risk source patches in the study area. Six complementary ecological sources and four new ecological sources were obtained through the blind zone reduction strategy. The percentage of ecological blind zones reduced from 58.4% to 39.5%. (2) The integrated nodes with 11,366 connecting edges were identified. The integrated nodes are distributed around the central Jiuling-Mafushan Mountains, mainly in the western and southern areas of the Dongting Lake Plain. (3) Primary integration nodes are critical for network stability, with a 75% node failure threshold triggering systemic collapse. The proposed strategy of “mountain protection–plain control–railway monitoring” is consistent with China’s territorial and spatial planning. By incorporating the risk network into the conservation framework, this study provides feasible insights for balancing development and sustainability in ecologically fragile areas. Full article

Detecting small targets in infrared search and track (IRST) systems in complex backgrounds poses a significant challenge. This study introduces a novel detection framework that integrates directional derivative correlation filtering (DDCF) with a local relative intensity contrast measure (LRICM) to effectively handle diverse background disturbances, including cloud edges and structural corners. This approach involves converting the original infrared image into an infrared gradient vector field (IGVF) using a facet model. Exploiting the distinctive characteristics of small targets in second-order derivative computations, four directional filters are designed to emphasize target features while suppressing edge clutter. The DDCF map is then constructed by merging the results of the second-order derivative filters applied in four distinct orientations. Subsequently, the LRICM is determined by analyzing the gray-level contrast between the target and its immediate surroundings, effectively minimizing interference from background elements like corners. The final detection step involves fusing the DDCF and LRICM maps to generate a comprehensive saliency representation, which is then processed using an adaptive thresholding technique to extract small targets accurately. Experimental evaluations across multiple datasets verify that the proposed method substantially improves the signal-to-clutter ratio (SCR). Compared to existing advanced techniques, the proposed approach demonstrates superior detection reliability in challenging environments, including ground surfaces, cloudy conditions, forested areas, and urban structures. Moreover, the framework maintains low computational complexity, achieving a favorable balance between detection accuracy and efficiency, thereby demonstrating promising potential for deployment in practical IRST scenarios. Full article

Because of their distinctive toroidal blade configuration, toroidal propellers can improve propulsion efficiency, reduce underwater noise, and enhance blade stability and strength. In recent years, they have emerged as an extremely promising novel underwater propulsion technology. To investigate their working mechanism, a geometric model of the toroidal propeller was initially established, and an unsteady numerical calculation model was constructed based on the sliding mesh technique. Subsequently, with the E779A conventional propeller as the research subject, the numerical model was verified, and a grid independence test was accomplished. Thereafter, the hydrodynamic performance of the toroidal propeller under diverse advance coefficients was analyzed based on the numerical model, and open water characteristic curves were established. Eventually, the surface pressure distribution, velocity field, and vorticity field of the toroidal propeller under various working conditions were studied. The outcomes demonstrate that the toroidal propeller attains the maximum propulsion efficiency at high advance coefficients, possesses a broad range of working condition adaptability, and is more applicable to high-speed vessels. At low advance coefficients, the toroidal propeller exhibits a relatively strong thrust performance, with the thrust generated by the front propeller being greater than that generated by the rear propeller, and the pressure peak emerges at the leading edge of the transition section of the front blade. The analysis of the velocity field indicates that its acceleration effect is superior to that of the conventional propeller. The analysis of the vorticity field reveals that the trailing vortices shed from the leading edge of the transition section of the front propeller merge and develop with the tip vortices, resulting in a more complex vortex structure. This research clarifies the working mechanism of the toroidal propeller through numerical simulation methods, providing an important basis for its performance optimization. Full article

Dragonflies exhibit remarkable flight capabilities, and their wings feature corrugated structures that are distinct from conventional airfoils. This study investigates the aerodynamic effects of three corrugation parameters on gliding performance at a Reynolds number of 1350 and angles of attack ranging from 0° to 20°: (1) chordwise corrugation position, (2) linear variation in corrugation amplitude toward the trailing edge, and (3) the number of trailing-edge corrugations. The results show that when corrugation structures are positioned closer to the trailing edge, they generate localized vortices in the mid-forward region of the upper surface, thereby enhancing aerodynamic performance. Further studies show that a linear increase in corrugation amplitude toward the trailing edge significantly delays the shedding of the leading-edge vortex (LEV), produces a more coherent LEV, and reduces the number of vortices within the corrugation grooves on the lower surface. Consequently, the lift coefficient is maximized with an enhancement of 28.99%. Additionally, reducing the number of trailing-edge corrugations makes the localized vortices on the upper surface approach the trailing edge and merge into larger, more continuous LEVs. The vortices on the lower surface grooves also decrease in number, and the lift coefficient is maximally increased by 20.09%. Full article