Search Results (249)

At present, most of the switchgear partial discharge detection means are offline detection and cannot monitor multiple partial discharge sources online at the same time. Based on this, this paper investigates the application of ultrasonic technology in localized discharge fault localization in high-voltage switchgear, removes the background noise of localized discharge in switchgear by using soft and hard filtering; proposes a generalized cubic correlation algorithm on the basis of TODA, improves the accuracy of the time difference acquisition in the case of low signal-to-noise ratio; determines the number of multiple localized discharging power sources by using the single-channel signal blind source separation technique and singularity spectral analysis; and determines the number of multiple localized discharging power sources by using independent component analysis to separate them. As well as for the problem that TDOA cannot be directly applied to the localization of multiple partial discharge sources, independent component analysis is used to separate the mixed signals, and the disordered coordinate selection method is proposed to determine the coordinates of multiple partial discharge sources. The experimental results show that (1) the noise reduction method is able to remove the excess interference while preserving the localized discharge signals; (2) the improved generalized cubic inter-correlation algorithm is more resistant to interference and has less error than other time delay estimation algorithms. The localization error is reduced by 60 mm~68 mm compared to the basic correlation algorithm, 41 mm~47 mm compared to the twice correlation algorithm, and 17 mm~20 mm compared to the three times correlation algorithm, which is a big improvement compared to the pre-improved algorithm. (3) It is able to locate the multiple localized power sources, and the accuracy of the number of localized power sources reaches 88%. Full article

Microbiological analysis is crucial for identifying species, assessing infections, and diagnosing infectious diseases, thereby supporting both research studies and medical diagnosis. In response to these needs, accurate and efficient identification of bacterial colonies is essential. Conventionally, this process is performed through manual counting and visual inspection of colonies on agar plates. However, this approach is prone to several limitations arising from human error and external factors such as lighting conditions, surface reflections, and image resolution. To overcome these limitations, an automated bacterial colony counting and classification system was developed by integrating a custom-designed imaging device with advanced deep learning models. The imaging device incorporates controlled illumination, matte-coated surfaces, and a high-resolution camera to minimize reflections and external noise, thereby ensuring consistent and reliable image acquisition. Image-processing algorithms implemented in MATLAB were employed to detect bacterial colonies, remove background artifacts, and generate cropped colony images for subsequent classification. A dataset comprising nine bacterial species was compiled and systematically evaluated using five deep learning architectures: ResNet-18, ResNet-50, Inception V3, GoogLeNet, and the state-of-the-art EfficientNet-B0. Experimental results demonstrated high colony-counting accuracy, with a mean accuracy of 90.79% ± 5.25% compared to manual counting. The coefficient of determination (R² = 0.9083) indicated a strong correlation between automated and manual counting results. For colony classification, EfficientNet-B0 achieved the best performance, with an accuracy of 99.78% and a macro-F1 score of 0.99, demonstrating strong capability in distinguishing morphologically distinct colonies such as Serratia marcescens. Compared with previous studies, this research provides a time-efficient and scalable solution that balances high accuracy with computational efficiency. Overall, the findings highlight the potential of combining optimized imaging systems with modern lightweight deep learning models to advance microbiological diagnostics and improve routine laboratory workflows. Full article

Landslides endanger human safety and damage infrastructure, underscoring the importance of accurate extraction. However, landslide extraction is often hindered by the omission of sparsely distributed landslides and the difficulty of delineating their blurred boundaries. Large-scale landslide extraction faces two key challenges. The first is a severe sample imbalance between landslides and background objects, which biases the model toward background and omits landslides. The second is the confusion between landslides and background features, which leads to inaccurate boundary delineation and fragmented extraction results. To address these issues, this paper proposes a two-phase landslide extraction framework. First, we propose a PCA-based landslide candidate extraction module to remove salient background objects and reduce data imbalance. Second, we propose a Spike-inspired Landslide Extraction Model to further discriminate actual landslides from the candidates by incorporating a spike-inspired sparse attention module (SISA). It can enhance weak landslide features such as blurred boundaries while mitigating background noise through its adaptive spatial suppression mechanism. To integrate spatial details across scales, a mix-scale feature aggregation module (MSFA) is proposed, which aggregates hierarchical features to extract landslides of various scales. Experiments on the landslide datasets from the Hengduan Mountains and Hokkaido, Japan, show IoU improvements of 4.26% and 1.22% compared to the recently proposed methods, validating its effectiveness under both imbalanced and dense landslide conditions. Full article

Traditional water hazard monitoring often relies on manual inspection and water level sensors, typically lacking in accuracy and real-time capabilities. However, the method of using video surveillance for monitoring water hazard characteristics can compensate for these shortcomings. Therefore, this study proposes a method to detect water hazards in mines using video recognition technology, combining temporal and spatial descriptors to enhance recognition accuracy. This study employs residual preprocessing technology to effectively eliminate complex underground static backgrounds, focusing solely on dynamic water flow features, thereby addressing the issue of the absence of water inrush samples. The method involves analyzing dynamic water flow pixels and applying an iterative denoising algorithm to successfully remove discrete noise points while preserving connected water flow areas. Experimental results show that this method achieves a detection accuracy of 90.68% for gushing water, significantly surpassing methods that rely solely on temporal or spatial descriptors. Moreover, this method not only focuses on the temporal characteristics of water flow but also addresses the challenge of detection difficulties due to the lack of historical gushing water samples. This research provides an effective technical solution and new insights for future water gushing monitoring in mines. Full article

The safe and stable operation of distribution networks relies on the real-time monitoring, analysis, and feedback of power quality data. However, with the continuous advancement of distribution network construction, the number of distributed power electronic devices has increased significantly, leading to frequent power quality issues such as voltage fluctuations, harmonic pollution, and three-phase unbalance in distribution terminals. Therefore, the augmentation and processing of power quality data have become crucial for ensuring the stable operation of distribution networks. Traditional methods for augmenting and processing power quality data fail to consider the differentiated characteristics of burrs in signal sequences and neglect the comprehensive consideration of both time-domain and frequency-domain features in disturbance identification. This results in the distortion of high-frequency fault information, and insufficient robustness and accuracy in identifying Power Quality Disturbance (PQD) against the complex noise background of distribution networks. In response to these issues, we propose a power quality data augmentation and processing method for distribution terminals considering high-frequency sampling. Firstly, a burr removal method of the sampling waveform based on a high-frequency filter operator is proposed. By comprehensively considering the characteristics of concavity and convexity in both burr and normal waveforms, a high-frequency filtering operator is introduced. Additional constraints and parameters are applied to suppress sequences with burr characteristics, thereby accurately eliminating burrs while preserving the key features of valid information. This approach avoids distortion of high-frequency fault information after filtering, which supports subsequent PQD identification. Secondly, a PQD identification method based on a dual-channel time–frequency feature fusion network is proposed. The PQD signals undergo an S-transform and period reconfiguration to construct matrix image features in the time–frequency domain. Finally, these features are input into a Convolutional Neural Network (CNN) and a Transformer encoder to extract highly coupled global features, which are then fused through a cross-attention mechanism. The identification results of PQD are output through a classification layer, thereby enhancing the robustness and accuracy of disturbance identification against the complex noise background of distribution networks. Simulation results demonstrate that the proposed algorithm achieves optimal burr removal and disturbance identification accuracy. Full article

Accurate and timely pest monitoring is essential for sustainable agriculture and effective crop protection. While recent deep learning-based pest recognition systems have significantly improved accuracy, they are typically trained for fixed label sets and narrowly defined tasks. In this paper, we present RefPestSeg, a universal, language-promptable segmentation model specifically designed for pest monitoring. RefPestSeg can segment targets at any semantic level, such as species, genus, life stage, or damage type, conditioned on flexible natural language instructions. The model adopts a symmetric architecture with self-attention and cross-attention mechanisms to tightly align visual features with language embeddings in a unified feature space. To further enhance performance in challenging field conditions, we integrate an optimized super-resolution module to improve image quality and employ diverse data augmentation strategies to enrich the training distribution. A lightweight postprocessing step refines segmentation masks by suppressing highly overlapping regions and removing noise blobs introduced by cluttered backgrounds. Extensive experiments on a challenging pest dataset show that RefPestSeg achieves an Intersection over Union (IoU) of 69.08 while maintaining robustness in real-world scenarios. By enabling language-guided pest segmentation, RefPestSeg advances toward more intelligent, adaptable monitoring systems that can respond to real-time agricultural demands without costly model retraining. Full article

Background/Objectives: Acquired sensorineural hearing loss (SNHL) can result from a wide range of insults, including ototoxic drugs, Meniere’s disease, noise-induced ototoxicity, and aging. The underlying pathophysiological mechanism arises through damage to the inner ear via oxidative stress and inflammation. Recent research suggests that natural antioxidants are promising solutions to prevent SNHL. Nigella sativa (NS), through its active compound thymoquinone (TQ), is a potent antioxidant that has shown promising results. The aim of this systematic review is to examine whether NS can offer protection against acquired SNHL. Methods: This study reviewed the literature on the protective effects of Nigella sativa oil (NSO) or TQ against acquired SNHL. We followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). A comprehensive literature search was conducted across multiple databases using keywords related to NS and hearing loss. Meta-analyses were performed on eligible studies. Risk of bias was assessed using the Systematic Review Center for Laboratory Animal Experiments (SYRCLE) tool for animal studies. Results: Out of a total of 76 records, 38 duplicates were removed. From the remaining 38, 13 studies met the inclusion criteria. Multiple studies reported a significant protective effect of NS, especially against ototoxicity. The risk of bias across the studies was moderate. Conclusions: Preclinical evidence indicates that NS provides significant protection against acquired SNHL. These protective effects are attributed to their antioxidant, anti-apoptotic, and anti-inflammatory properties. Overall, our systematic review highlights NS as a promising candidate for preventing SNHL. Full article

Background and Objectives: Speech-in-noise testing is essential for evaluating functional hearing abilities in clinical practice. Although the Quick Speech-in-Noise test (QuickSIN) is widely used, no equivalent tool existed for European Portuguese. This study aimed to develop a Speech-in-Noise Test for European Portuguese (SiN-EP), linguistically adapted and calibrated for native speakers, to support clinical assessment of speech perception in realistic listening environments. Materials and Methods: The SiN-EP was developed through a multi-stage process. Sentences were drafted to reflect natural speech patterns and reviewed by native speakers for clarity and grammatical accuracy. Selected sentences were recorded by a female native speaker in a controlled acoustic environment and mixed with multi-talker babble at signal-to-noise ratios (SNR (dB)) from 25 to 0 SNR (dB). A pre-test in a free-field setting at 65 dB SPL was conducted with fifteen normal-hearing young adults. Participants repeated each sentence, and their responses were analyzed to refine list composition, adjust difficulty, and ensure phonetic balance. Results: Intelligibility decreased systematically as SNR (dB) worsened, with ceiling effects at 25 and 20 SNR (dB). At 5 SNR (dB), high variability was observed, with set 5 showing disproportionate difficulty and set 14 containing an incomplete sentence; both were removed. At 0 SNR (dB), all sets demonstrated expected low intelligibility. The final test comprises thirteen lists of six sentences, each maintaining stable intelligibility, phonetic representativeness, and consistent difficulty across SNRs (dB). Conclusions: The SiN-EP provides a linguistically appropriate, phonetically balanced, and SNR (dB) calibrated instrument for assessing speech-in-noise perception in European Portuguese. The refinement process improved reliability and list equivalence, supporting the test’s clinical and research applicability. The SiN-EP fills a critical gap in assessing speech-in-noise perception in European Portuguese speakers, providing a reliable tool for both clinical and research applications. Full article

Microelectrode recording (MER) is commonly used to validate preoperative targeting during subthalamic nucleus (STN) deep brain stimulation (DBS) surgery for Parkinson’s Disease (PD). Although machine learning (ML) has been used to improve STN localization using MER data, the impact of preprocessing steps on the accuracy of classifiers has received little attention. We evaluated 24 distinct preprocessing pipelines combining four artifact removal strategies, three outlier handling methods, and optional feature normalization. The effect of each data processing procedure’s component of interest was evaluated in function of the performance obtained using three ML models. Artifact rejection methods (i.e., unsupervised variance-based algorithm (COV) and background noise estimation (BCK)), combined with optimized outlier management (i.e., statistical outlier identification per hemisphere (ORH)) consistently improved classification performance. In contrast, applying hemisphere-specific feature normalization prior to classification led to performance degradation across all metrics. SHAP (SHapley Additive exPlanations) analysis, performed to determine feature importance across pipelines, revealed stable agreement with regard to influential features across diverse preprocessing configurations. In conclusion, optimal artifact rejection and outlier treatment are essential in preprocessing MER for STN identification in DBS, whereas preliminary feature normalization strategies may impair model performance. Overall, the best classification performance was obtained by applying the Random Forest model to the dataset treated using COV artifact rejection and ORH outlier management (accuracy = 0.945). SHAP-based interpretability offers valuable guidance for refining ML pipelines. These insights can inform robust protocol development for MER-guided DBS targeting. Full article

Ground-penetrating radar (GPR) has been proven effective for detecting subsurface pipes in a nondestructive way, typically with manual processing and decision-making. However, existing automatic models for segmenting the target hyperbolas often lack generalization across different pipe radii, varying subsurface media, and complex field conditions. This is especially reflected in B-scans with diverse or small-scale hyperbolas, often accompanied by cluttered and irregular noise. In this paper, an automatic preprocessing model is proposed to enhance the interpretation of B-scans under challenging conditions. The model includes a ground reflection removal algorithm (GRRA), the data gravitational force enhancement (DGFE) method, and a global–local thresholding technique consisting of dilation-based local thresholding and segmentation (DLTS). First, a frequency-domain filter based on the fast Fourier transform and a spatial filter are applied to the raw B-scan to remove obstructive ground reflection strips. Owing to the minimal intensity differences among the target hyperbola, multiples, and background, the DGFE approach is introduced to amplify the main body of the hyperbola, distinguishing it from the noise. Finally, the target hyperbola is extracted from the grayscale image by an integrated thresholding approach. The approach initially employs global thresholding to eliminate all information except for part of the hyperbola, followed by DLTS, which uses a dilation operation with local thresholding to fully segment the hyperbola. The proposed model is evaluated on two distinct datasets and compared with several state-of-the-art methods. The results demonstrate its effectiveness, particularly in terms of cross-dataset generalization. Full article

Background: The DASH risk assessment scheme is used across the UK to identify and manage instances of domestic abuse. Recent studies have questioned whether the scheme can identify offenders who go on to commit further acts of domestic abuse, in particular serious violence, and therefore whether it is fit for purpose. Methods: We therefore tested the ability of the DASH to predict future instances of deadly or persistent domestic abuse. From a database of ≈25,000 incidents, we compared DASH assessments which preceded an incident of “deadly violence” or was the first in a series of “persistent abuse”. These groups were compared to a control group where there was no further incident of domestic abuse. Results: The proportion of “high-risk” stratifications was approximately 5 times higher in the deadly violence group compared to the control group. Prediction accuracy assessed via signal detection theory showed the DASH was a moderate predictor of deadly violence (AUC = 0.67). The DASH also showed predictive accuracy in identifying persistent offenders (AUC = 0.62). While these results are encouraging and are similar in efficacy to other risk assessment schemes used in the prediction of domestic violence, the results identified that many individual items of the DASH were not predictive. The inclusion of non-predictive items within the DASH adds “noise” and error into the risk evaluation. The development of a shortened version of the DASH, removing these ineffectual items, was shown to have even higher predictive value for deadly violence (AUC = 0.80). Conclusions: We stress, however, that the role of risk assessment is not to predict violence per se, but to prevent violence via the accurate identification of dangerous perpetrators and via effective intervention and safeguarding of victims. Despite this, research such as this is imperative to evaluate if the risk assessment schemes selected by practitioners and police are fit for purpose. Full article

In the wild, snow is not merely additive noise; it is a non-stationary, semi-transparent veil whose spatial statistics vary with depth, illumination, and wind. Because conventional two-stage pipelines first detect a binary mask and then inpaint the occluded regions, any early mis-classification is irreversibly baked into the final result, leading to over-smoothed textures or ghosting artifacts. We propose BAT-Net, a Bidirectional Attention Transformer Network that frames desnowing as a coupled representation learning problem, jointly disentangling snow appearance and scene radiance in a single forward pass. Our core contributions are as follows: (1) A novel dual-decoder architecture where a background decoder and a snow decoder are coupled via a Bidirectional Attention Module (BAM). The BAM implements a continuous predict–verify–correct mechanism, allowing the background branch to dynamically accept, reject, or refine the snow branch’s occlusion hypotheses, dramatically reducing error accumulation. (2) A lightweight yet effective multi-scale feature fusion scheme comprising a Scale Conversion Module (SCM) and a Feature Aggregation Module (FAM), enabling the model to handle the large scale variance among snowflakes without a prohibitive computational cost. (3) The introduction of the FallingSnow dataset, curated to eliminate the label noise caused by irremovable ground snow in existing benchmarks, providing a cleaner benchmark for evaluating dynamic snow removal. Extensive experiments on synthetic and real-world datasets demonstrate that BAT-Net sets a new state of the art. It achieves a PSNR of 35.78 dB on the CSD dataset, outperforming the best prior model by 1.37 dB, and also achieves top results on SRRS (32.13 dB) and Snow100K (34.62 dB) datasets. The proposed method has significant practical applications in autonomous driving and surveillance systems, where accurate snow removal is crucial for maintaining visual clarity. Full article

Compared with capillary electrophoresis (CE), gel electrophoresis (GE) is a traditional method for the analysis of nucleic acids because of its low cost, although the operation process is complicated. The electropherogram from CE can offer more information (e.g., DNA size and its concentration) for researchers. Based on the self-built integrated biochip GE system, we proposed a computational method that converts conventional agarose GE images into CE-like fluorescence profiles for enhanced DNA analysis. The gel images were processed using an image-based algorithm involving median filtering to remove background noise and pixel-wise intensity summation along the migration axis to generate one-dimensional records of electrophoretic separations. Each DNA band in the gel was thereby transformed into a distinct fluorescence peak, reflecting its migration distance and relative intensity. To further enhance resolution and peak separation, Gaussian modeling was applied to fit the fluorescence intensity distribution, providing smoother and more distinguishable spectral peaks. To validate the method, three periodontal pathogens—Porphyromonas gingivalis (P.g), Treponema denticola (T.d), and Tannerella forsythia (T.f)—were amplified using PCR and analyzed by gel electrophoresis. The method successfully identified distinct electrophoretic patterns for the three pathogens by using a 50 bp DNA ladder as an internal calibration reference. The results demonstrate that image-based reconstruction of electrophoretic data provides a reliable, quantitative, and visually interpretable representation of DNA migration, comparable to CE output. This approach bridges a gap between traditional GE and modern capillary systems, allowing for the semi-quantitative analysis of DNA fragments without specialized CE instrument. The proposed method offers a valuable analysis method for the separation of DNA, RNA, protein and polypeptides. Full article

Flange bolts are vital fasteners in civil, mechanical, and aerospace structures, where preload stability directly affects overall safety. Conventional methods for bolt loosening detection often suffer from missed detections, weak feature representation, and insufficient cross-scale fusion under complex backgrounds. This paper presents an integrated detection and angle estimation framework using a lightweight deep learning detection network. A MobileViT backbone is employed to balance local texture with global context. In the spatial pyramid pooling stage, large separable convolutional kernels are combined with a channel and spatial attention mechanism to highlight discriminative features while suppressing noise. Together with content-aware upsampling and bidirectional multi-scale feature fusion, the network achieves high accuracy in detecting small and low-contrast targets while maintaining real-time performance. For angle estimation, the framework adopts an efficient training-free pipeline consisting of oriented FAST and rotated BRIEF feature detection, approximate nearest neighbor matching, and robust sample consensus fitting. This approach reliably removes false correspondences and extracts stable rotation components, maintaining success rates between 85% and 93% with an average error close to one degree, even under reflection, blur, or moderate viewpoint changes. Experimental validation demonstrates strong stability in detection and angular estimation under varying illumination and texture conditions, with a favorable balance between computational efficiency and practical applicability. This study provides a practical, intelligent, and deployable solution for bolt loosening detection, supporting the safe operation of large-scale equipment and infrastructure. Full article

The Solar Magnetic Field Telescope (SMFT) at Huairou Solar Observing Station (HSOS) has conducted continuous observations of solar vector magnetic fields for nearly four decades, and while the primary optical system remains unchanged, critical components—including filters, polarizers, and detectors—have undergone multiple upgrades and replacements. Maintaining data consistency is essential for reliable long-term studies of magnetic field evolution and solar activity, as well as current helicity. In this study, we systematically analyze background bias and noise levels in SMFT observations from 1988 to 2019. Our dataset comprises 12,281 vector magnetograms of 1484 active regions. To quantify background bias, we computed mean values of Stokes $Q/I$, $U/I$ and $V/I$ over each entire magnetogram. The background bias of Stokes $V/I$ is small for the whole dataset. The background biases of Stokes $Q/I$ and $U/I$ fluctuate around zero during 1988–2000. From 2001 to 2011, however, the fluctuations in the background bias of both $Q/I$ and $U/I$ become significantly larger, exhibiting mixed positive and negative values. Between 2012 and 2019, the background biases shift to predominantly positive values for both Stokes $Q/I$ and $U/I$ parameters. To address this issue, we propose a potential method for removing the background bias and further discuss its impact on the estimation of current helicity. For each magnetogram, we quantify measurement noise by calculating the standard deviation ($\sigma$) of the longitudinal ($B_{\mathrm{l}}$) and transverse ($B_{\mathrm{t}}$) magnetic field components within a quiet-Sun region. The noise levels for $B_l$ and $B_{\mathrm{t}}$ components were approximately 15 Gauss (G) and 87 G, respectively, during 1988–2011. Since 2012, these values decreased significantly to ∼6 G for $B_{\mathrm{l}}$ and ∼55 G for $B_{\mathrm{t}}$, likely due to the installation of a new filter. Full article