The advent of smart farms, enabled by information and communication technologies (ICT) and the Internet of Things (IoT), has improved productivity and sustainable agriculture. However, the large-scale implementation of smart farms is currently hampered by physical constraints. These constraints have led to the concept of open-field smart farming as a viable alternative. In this paradigm, data from unmanned aerial vehicles (UAVs) play a central role in effective and sustainable agricultural management. The quantitative analysis of such data requires highly reliable technological solutions. The objective of this study is to conduct a comparative analysis of image binarization algorithms for UAV-based soybean canopy extraction across growth stages and to contribute to the development of an image labeling methodology. UAVs were used to capture images of soybean fields at different growth stages, and a comparative analysis was performed using binarization image algorithms. The performance of each algorithm was evaluated using Normalized Cross Correlation (NCC) and Mean Absolute Error (MAE). The results indicate that the Excess Green (ExG) and Excess Green minus Excess Red (ExGR) vegetation indices provide accurate and stable soybean canopy extraction across growth stages when combined with Adaptive and Otsu binarization algorithms. These indices are particularly suitable for extracting soybean canopy from UAV-based data, thereby expanding the scope of precision analysis in the agricultural sector and providing data for advancing precision agriculture technology. This study contributes to the standardization and efficient use of UAV-based agricultural data processing. However, since manual weeding was performed prior to image acquisition to ensure that only soybean plants were present, reflecting standard agricultural practices in South Korea, additional validation would be required for application in fields where weeds are naturally present.

Flame image binarization statistics and pollutant emission analysis of hydrogen-enriched kerosene combustion in a turbulent swirl burner

WormTracer: A precise method for worm posture analysis using temporal continuity.

Semi-analytical penalized threshold dynamics method for binary image segmentation

Early prognosis prediction is challenging after endovascular treatment (EVT) for acute posterior circulation ischemic stroke (PCIS). We evaluated the frequency and prognostic impact of competitive blood flow from the contralateral vertebral artery (CBF-cVA), observed post-recanalization. We retrospectively screened patients with acute PCIS who underwent EVT with successful recanalization. CBF-cVA was defined as the rapid clearing of the basilar artery and posterior cerebral arteries previously opacified by antegrade reperfusion. The good functional outcomes are defined as a score of 0-3 on the modified Rankin scale (mRS) at 90 days. Logistic regression was used to investigate the association of CBF-cVA and good functional outcomes at 90 days. A total of 259 patients (median age, 64 years, 74.9% male) were included. CBF-cVA was observable in 44.0% of patients and more frequently in patients with good status of the non-operated vertebral artery which was categorized as good or bad based on the presence of hypoplasia, occlusion, slow flow, or lack of opacification (14.9% vs. 62.1%; p < 0.001) and better collateral score (median 6 vs. 4; p < 0.001). CBF-cVA was associated with good functional outcomes (adjusted OR [95% CI], 3.410 [1.636, 7.105]; p = 0.001), but not with 90-day mortality and symptomatic intracranial hemorrhage (both p > 0.05). CBF-cVA was associated with better functional outcomes in patients with PCIS who underwent EVT with successful recanalization. The presence of CBF-cVA was related to the status of the non-operated vertebral artery and better collateral flow. Question The prediction of early prognosis continues to pose a significant challenge in the clinical management of patients undergoing thrombectomy for posterior circulation strokes. Findings The presence of competitive blood flow from the contralateral vertebral artery (CBF-cVA) was significantly associated with more favorable 90-day functional outcomes. Clinical relevance CBF-cVA provided a practical, binary imaging assessment after endovascular treatment in patients with acute posterior circulation ischemic stroke. This straightforward tool improved the prediction of favorable clinical outcomes, aiding in guiding treatment strategies after recanalization.

Given GPS points on a transportation network, the Driving Pattern Detection (DPD) problem aims to classify drivers as normal or abnormal based on their driving behavior. The DPD problem is challenging due to the variability in trip lengths, routes, and spatial patterns, which complicates input standardization for deep learning models. In this paper, we introduce a novel spatial representation learning framework for the DPD problem by analyzing driving patterns using a real-world dataset. We propose using binary grid images to capture the spatial structure of driving trajectories and present a new driving behavior representation for input to a Vision Transformer (ViT) model for driver classification. The experimental results demonstrate the effectiveness of the proposed algorithm, achieving an F1 score of 94% that significantly outperforms the baseline models. The results indicate that binary grid representations can effectively encode interpretable spatial patterns in driving behavior, with direct relevance to improved driver classification, road safety, and cognitive health assessment.

Distributed learning is the standard for training large-scale models across private data silos, offering privacy and efficiency but facing challenges in Byzantine robustness and communication efficiency. Existing Byzantine-robust and communication-efficient methods rely on full gradient information, and they only converge to an unnecessarily large neighborhood around the solution. Motivated by these issues, we propose a novel Byzantine-robust and communication-efficient stochastic distributed learning method that imposes no requirements on batch size and converges to a smaller neighborhood, aligning with the theoretical lower bound. Our key innovation is leveraging Polyak Momentum to mitigate the noise caused by both biased compressors and stochastic gradients, thus defending against Byzantine workers under information compression. We provide proof of tight complexity bounds for nonconvex smooth loss functions. Finally, we validate the practical significance of our algorithm through an extensive series of experiments, benchmarking its performance on both binary classification and image classification tasks.

Obfuscation detection using matrix complexity features of binary grayscale images

Reinforced Concrete (RC) bridges are prone to deterioration throughout their service life, which can compromise their structural integrity. Among the visible indicators of deterioration, cracks are particularly significant, underscoring the need for their early detection. Traditional inspections, primarily based on visual inspection, face challenges such as restricted access to certain bridge areas, complex logistics, and excessive operational costs and time. To address these challenges, the engineering community is working toward implementing more efficient technologies for bridge inspection. This paper proposes an automatic deep learning-based system for the segmentation and quantification of surface cracks in RC bridges using images captured by unmanned aerial vehicles (UAVs). The proposed framework combines patch-based and pixel-level segmentation using binary image classification and crack segmentation models. Additionally, an alternative method based on a laser module with a diffractive optical element (DOE) was implemented to obtain an automatic scale factor, enabling the conversion of pixel-level information into metric measurements for damage quantification. Experimental results show that the binary classification model achieved an F1-score of 0.93 on the test set, while the segmentation model achieved an F1-score of 0.83. Overall system performance yielded a coefficient of determination (R²) of 0.70 and an average inference time of 2.32 seconds per high-resolution image. Through integrating UAVs, lasers, and deep learning, the proposed approach offers an alternative for bridge inspections, reducing cost and time while supporting the rapid generation of preliminary reports.

As counterfeit components become increasingly prevalent, encoded surfaces, particularly physically unclonable functions (PUFs), have emerged as powerful tools for secure part authentication and reliable traceability. However, significant challenges remain in fabricating unclonable surface structures in a high-throughput, scalable, and cost-effective manner while also ensuring robust encryption and secure authentication. This work aims to address the existence gaps by introducing an innovative approach to PUF manufacturing utilizing the cold spray (CS) particle deposition technique, complemented by algorithmic feature extraction and cryptographic surface encoding. In our approach, a mixture of metal and fluorescent microparticles is deposited onto an aluminum (Al 5052) substrate by leveraging the process-specific two-phase (gas-solid) turbulent flow characteristic of the CS process. The inherent stochasticity of the CS flow leads to a random distribution of fluorescent particles, generating unique, physically unclonable luminescent patterns on the target surface. The spatial distribution of the optical fluorescent particles is then captured under UV light (365 nm) exposure and subsequently processed through image binarization. Features are extracted from this distribution by using Voronoi analysis. The extracted features are then encrypted using the SHA-256 cryptographic algorithm to generate a secure "certification key" for part authentication. Experimental results demonstrate the effectiveness of the proposed manufacturing approach for high-throughput, scalable PUF production, confirming its suitability for robust part authentication and its reliability under environmental stressors (e.g., thermal cycling, chemical exposure). The developed method shows strong potential for enabling tamper-evident part authentication solutions to address the growing threat of counterfeiting in critical sectors, such as aerospace, defense, and advanced manufacturing.

The effective care of skin cancer relies on the fine detection of skin lesions. Deep learning techniques are increasingly being used in medical diagnosis, ranging from the classification of skin lesions. Their ability to learn deep discriminative features from dermoscopic images is what makes them popular. In spite of the fact that deep learning approaches learn rich semantically rich information, the approaches currently being taken tend to suffer from poor generalization, high levels of redundancy, and KNN classifiers that assign identical weights to all neighbors. The paper proposes a new approach using machine learning for the classification of skin lesions, entailing deep feature extraction, techniques for dimensionality reduction, and approaches for optimization. Specifically, the ResNet50 architecture using Global Average Pooling for deep feature extraction from dermoscopic images will be employed. The most relevant and non-redundant features are identified through the Minimum Redundancy Maximum Relevance (mRMR) method. mRMR removes irrelevant class information and reduces the feature size considerably. A new approach for the KNN classifier substitutes the fully connected layer of ResNet50. The weights for instance and feature levels are computed with the Genetic Algorithm (GA) and the use of cosine similarity. The proposed approach attains a high accuracy of 90.61% on the classification task for the binary images of skin lesions. The experimental results show that the proposed optimized cosine weighted KNN approach is effective for the diagnosis of skin cancer.

The simultaneous formation of stress transfer and structural optimization in deep underground engineering strongly influences the occurrence and severity of dynamic disasters, especially under complex stress–structure conditions. In western China, oil-rich coal is mainly distributed in complex geological structures, and the deformation, fracture, and failure of coal are essentially governed by fracture evolution under external loading. Therefore, the characteristics of the original fissure structure and the subsequent damage evolution process are critical for multi-source disaster prevention and control during oil-rich coal mining. In this study, the initiation, propagation, coalescence, and evolution of coal microfissures under multi-stage disturbances were investigated using CT scanning combined with mechanical testing. Differences in CT images among different regions were analyzed after initial oil and gas resource exploitation, and crack binary images were used to qualitatively describe fracture evolution. Meanwhile, coal CT number, crack number, trace width, gap length, and crack density were extracted to quantitatively characterize the spatio-temporal relationship between fracture structural features and multi-stage repeated disturbances, together with acoustic emission characteristics. The results provide a comprehensive description of fracture development and damage progression in disturbed coal samples, which is of great significance for understanding the fracture distribution and mechanical response mechanisms, thereby supporting safe and efficient coordinated mining of oil-rich coal in western China.

Background. Numerous studies have established that the roughness of the microrelief of the working surface of various parts of machines and mechanisms largely determines their operational reliability and durability. In this regard, the research and development of non-contact methods for the rapid assessment of the characteristics of microrepelops is an urgent task at present. Aim. The aim of the work is to study and develop an optoelectronic method, algorithms and software for processing raster images of the studied microreliefs obtained by small-sized video cameras, for the prompt assessment of its roughness directly in the course of the technodolic process. Methods. The method is based on computer processing of images of the studied microreliefs, as a result of which the roughness parameters of these microreliefs are determined. For this purpose, halftone images of microreliefs of reference samples, with certain GOST methods, roughness parameters, are first converted into binary images. Then, two-dimensional correlation functions are calculated from binary images using a quasi-optimal correlation algorithm. For these functions, the average amplitudes Ucp of the variable components of the correlation coefficients are calculated and the analytical dependence of the GOST roughness parameter Ra from Ucp is constructed. To depict the studied microrelief with unknown roughness parameters, the variable component of the two-dimensional correlation function is also determined and according to the obtained dependence Ra = (Ucp), the arithmetic mean deviation of the profile from the mean line Ra is found. Results. A new optoelectronic method for assessing the roughness parameters of microreliefs of precision surfaces is proposed, which is based on correlation processing of images of the studied microreliefs.The results of applying this method to the assessment of the roughness parameters of the surfaces of gas turbine engine. Defined fields roughness fields of the surfaces of the trough and the back of the blade of the 1st stage of gas turbine engine. Conclusion. The expediency of applying a correlation approach to processing images of microreliefs in order to determine the roughness parameters of precision surfaces is shown. At the same time, the negative influence of side uninformative factors on the results of measuring the parameters of the microrelief, in particular, fluctuations in the power of the light flux and the angle of its incidence on the surface under study, is also eliminated. In order to reduce the time spent on correlation image processing, a quasi-optimal correlation algorithm was used. Roughness fields for the working surfaces of the gas turbine engine blade were determined, on the basis of which it became possible to determine dangerous areas with unacceptable values of stress concentration.

Traditional thresholding methods are often tailored to specific histogram patterns, making it difficult to achieve robust segmentation across diverse images exhibiting non-modal, unimodal, bimodal, or multimodal distributions. To address this limitation, this paper proposes an automatic thresholding method guided by maximizing homologous isomeric similarity under a unified transformation toward unimodal distribution. The primary objective is to establish a generalized selection criterion that functions independently of the input histogram’s pattern. The methodology employs bilateral filtering, non-maximum suppression, and Sobel operators to transform diverse histogram patterns into a unified, right-skewed unimodal distribution. Subsequently, the optimal threshold is determined by maximizing the normalized Renyi mutual information between the transformed edge image and binary contour images extracted at varying levels. Experimental validation on both synthetic and real-world images demonstrates that the proposed method offers greater adaptability and higher accuracy compared to representative thresholding and non-thresholding techniques. The results show a significant reduction in misclassification errors and improved correlation metrics, confirming the method’s effectiveness as a unified thresholding solution for images with non-modal, unimodal, bimodal, or multimodal histogram patterns.

To address the challenges posed by rock bolt corrosion to engineering safety and service life, this study focuses on corrosion detection through integrated image processing, deep learning, and feature extraction methods. An automatic corrosion identification model was constructed based on computer-vision object-detection algorithms. By incorporating a Feature Pyramid Network, the model's multi-scale object-detection capability was significantly enhanced. The corrosion features were extracted via image binarization and grayscale matrix analysis. The binary image method accurately quantified pitting density, revealing an initial increase followed by a decrease over time. The corrosion morphology was simulated using a Fractional Brownian Motion model, validating the accuracy of fractal feature calculations. The fractal dimension increased significantly with prolonged corrosion time, which not only characterize surface roughness evolution and corrosion rate, but also provide a reliable quantitative indicator for metal corrosion assessment. This research offers a technical framework integrating image processing, deep learning, and fractal theory for rock bolt corrosion monitoring and maintenance.

Background: Vitreoretinal lymphoma (VRL) often presents with features resembling uveitis and is commonly associated with central nervous system lymphoma (CNSL). Intravitreal methotrexate (IVMTX) is widely used as local therapy; however, objective markers for treatment response and prognosis remain limited. This study investigated choroidal structural changes after IVMTX via enhanced depth imaging optical coherence tomography (EDI-OCT) and explored prognostic indicators for subsequent CNSL development. Methods: This retrospective study included 18 patients (27 eyes) with VRL treated with IVMTX at Tokushima University Hospital between 2006 and 2021. EDI-OCT was conducted at baseline and at 1 and 3 months after IVMTX. Choroidal thickness and luminal and stromal areas were quantified through image binarization. The stromal/choroidal area (S/C) ratio and its association with CNSL onset were statistically analyzed. Results: The mean number of IVMTX injections administered over 3 months was 5.9 ± 1.3. Foveal retinal thickness did not significantly change, whereas foveal choroidal thickness significantly decreased from 275.8 ± 15.8 µm at baseline to 257.5 ± 14.7 µm at 1 month (p < 0.01). Total choroidal and stromal areas, particularly in the outer choroidal layer, were significantly decreased after IVMTX (p < 0.0001), whereas the luminal area in the inner layer modestly reduced (p < 0.05). The S/C ratio significantly declined at 1 month post-treatment (p < 0.001). Patients who developed CNSL within 2 years of VRL onset demonstrated higher baseline S/C ratios (p < 0.05). Conclusions: IVMTX induces measurable reductions in choroidal areas and stromal proportion, indicating decreased inflammatory infiltration. The baseline S/C ratio observed on EDI-OCT is a potential noninvasive biomarker of VRL activity and a prognostic indicator for early CNSL development.

Malware remains a major cybersecurity concern, which demands effective techniques for accurate detection and classification. This study presents a novel framework that leverages binary image representations of malware to enhance classification performance. The process begins by transforming malware files from their hexadecimal form into binary data, which is then converted to grayscale images serving as input for deep learning models. The study also examines the distinctive visual characteristics of various malware families, revealing how structural patterns in binary images are correlated with classification outcomes. By examining the role of image processing and deep learning, the research provides valuable insight into the intersection of artificial intelligence and cybersecurity. The findings highlight the strength of CNNs for malware classification, while acknowledging the complementary potential of ResNet and Autoencoder-based approaches. As cyber threats become increasingly sophisticated, advancing detection methods is essential. This work demonstrates that combining deep learning with binary image analysis presents a promising approach to developing more resilient malware detection systems and enhanced protection for digital environments. Three architectures—Convolutional Neural Networks (CNN), Residual Networks (ResNet), and Autoencoders—are systematically evaluated using a dataset of 3,240 malware samples categorized into nine families. The dataset is carefully divided into training and testing sets, and all images are resized to maintain consistency between inputs. Among the evaluated models, CNN with image-scaling techniques shows a superior accuracy of 91%, outperforming the ResNet and Autoencoder models, which achieve accuracies of 86% and 85%, respectively.

Objective: to identify key patterns of calcification in explanted bioprosthetic heart valves (BHVs) using cluster analysis of computed tomography-derived graphical data. Materials and methods. The study included 11 UniLine BHVs that were routinely explanted during reoperations for structural valve dysfunction. Computed tomography was used to obtain DICOM images of each sample, followed by generation of maximum intensity projections and segmentation of the valves into individual leaflets (n = 33). The images were pre-processed using binary thresholding to differentiate calcified regions from non-calcified biological tissue. Cluster analysis was performed using various algorithms: Gaussian mixture models, Ordering Points To Identify the Clustering Structure (OPTICS), k-means clustering, agglomerative (hierarchical) clustering, and spectral clustering. A basic quantitative method assessing the proportion of pixels corresponding to calcified areas was used for comparison. The performance of clustering algorithms was evaluated using the silhouette score. The presence of calcium deposits in the valves and the accuracy of binary thresholding were further verified histologically by alizarin red S staining of valve cryosections. Results. Data preprocessing based on image binarization yielded a maximum silhouette score of 0.55. Among the clustering algorithms, the highest silhouette scores were achieved with the agglomerative (0.55) and k-means (0.54) methods; however, both demonstrated substantial data imbalance, with up to 85% of samples grouped within a single cluster, limiting their practical applicability. The most balanced clustering was achieved using spectral clustering (silhouette score 0.45) and the basic quantitative approach (0.44). Both methods identified three distinct patterns of bioprosthetic valve leaflet calcification: (1) non-calcified leaflets, (2) partial calcification, and (3) total calcification. Conclusion. Three key calcification patterns were identified in explanted BHVs – absence of calcium, partial calcification, and total calcification. Spectral clustering and the basic quantitative method demonstrated the most balanced results, while other algorithms showed pronounced cluster imbalance. Heat map analysis revealed that in partial calcification, mineral deposition typically begins in the commissural and dome regions of the leaflets, near the free edge, and in total calcification, extends across the entire dome and leaflet base.

Background/Objectives: There is an increasing need for methods that provide improved insight into the molecular basis of colorectal cancer (CRC) and thus a better understanding of its morphological heterogeneity. The objectives of this study were to evaluate the use of imaging mass spectrometry (IMS) to examine tumor margins and gain insight into the molecular heterogeneity of CRC. Methods: An observational study involving 10 cases was conducted. Native tissue samples were collected during the subject's surgery, and consecutively taken tissue sections were immediately prepared for light microscopic and IMS analysis. IMS was performed across the 200-1000 Da mass range, divided into four sub-ranges, using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI TOF MS) in both negative and positive modes of ionization. For tumor margin and tissue heterogeneity assessment, image segmentation was used. Segmented MS images were analyzed against the respective light microscopy images of hematoxylin-eosin-stained tissue sections. Results: Quantitative analysis of the sample collection indicated that IMS enabled correct recognition of tumor margin in the 800-1000 Da range using binary segmentation. Denary image segmentation depicted tissue heterogeneity in greater detail. The strongest m/z signals specific to tumor, peritumor, and margins were identified and tentatively annotated: aside from dCTP, all other compounds were patient-specific, indicating interindividual variations in the molecular composition of CRC. Conclusions: IMS provides new insights into the morphological and biochemical properties of CRC: binary segmented MS images can clearly depict the tumor margin in the 800-1000 Da range, while denary segmented MS images depict intra- and inter-individual molecular heterogeneity of CRC.

This study explores the potential of fractal and complexity metrics in classifying open and globular stellar clusters, providing a nuanced perspective on their structural differences. Using a dataset of 20 open clusters and 20 globular clusters selected from the Messier Catalogue – due to their historical significance, availability in standardized formats, and high image quality — eight grayscale and eight binary metrics were applied to assess their effectiveness in distinguishing between these two classes. Although the sample size is modest, it was deliberately curated to ensure balance, comparability, and suitability for a first-stage methodological evaluation of fractal analysis in astronomy. This framework lays the groundwork for broader applications in future research. Among the grayscale metrics, Higuchi 2D Dimension emerged as the most effective, achieving complete separation with no overlap between the two classes. For binary images, the Fractal Fragmentation Index (FFI) demonstrated similar robustness, with minimal overlap. Complementary metrics, such as Lacunarity and Correlation Dimension, provided additional insights into the texture and spatial distribution of stellar clusters, though their utility for standalone classification was limited by moderate overlaps. Traditional metrics, such as FFT Dimension and Minkowski Dimension, showed reduced sensitivity in capturing subtle structural differences. The study employed overlap analysis and box-plot visualizations to identify key discriminative metrics, supported by statistical evaluations. Highly significant results (p¡0.001) were obtained for Higuchi Dimensions, Lacunarity, and FFI, underscoring their potential for precise classification. The findings align with previous research on fractal distributions in other complex systems, demonstrating the utility of fractal metrics in complementing traditional photometric and spectroscopic approaches.