Search Results (1,058)

To achieve efficient vineyard grape picking, a vision-based information processing framework integrating two-stage segmentation with morphological perception is proposed. In the first stage, an improved YOLOv8s-seg model is employed for coarse segmentation, incorporating two key enhancements: first, a dynamic deformation feature aggregation module (DDFAM), which facilitates the extraction of complex structural and morphological features; and second, an efficient asymmetric decoupled head (EADHead), which improves boundary awareness while reducing parameter redundancy. Compared with mainstream segmentation models, the improved model achieves superior performance, attaining the highest mAP@0.5 of 86.75%, a lightweight structure with 10.34 M parameters, and a real-time inference speed of 10.02 ms per image. In the second stage, the fine segmentation of fruit stems is performed using an improved OTSU thresholding algorithm, which is applied to a single-channel image derived from the hue component of the HSV color space, thereby enhancing robustness under complex lighting conditions. Morphological features extracted from the preprocessed fruit stem, including centroid coordinates and a skeleton constructed via medial axis transform (MAT), are further utilized to establish the spatial relationships with a picking point and cutting axis. The visualization analysis confirms the high feasibility and adaptability of the proposed framework, providing essential technical support for the automation of grape harvesting. Full article

To broaden the informational scope of existing fuzzy frameworks and enhance their flexibility in representing and processing uncertainty, we propose a novel p,q-rung triangular orthopair fuzzy number (p,q-RTOFN). To enhance the aggregation capability of fuzzy data, we develop a p,q-rung triangular orthopair fuzzy weighted power Bonferroni mean (p,q-RTOFWPBM) operator that integrates the strengths of the Bonferroni mean and power average operators. We formally establish its theorems, proofs, and key properties, including symmetry and idempotency. Furthermore, we extend the complex proportional assessment (COPRAS) method to the p,q-RTOF environment, resulting in a p,q-RTOF-PBM-COPRAS model. This model effectively incorporates both positive and negative evaluation information under uncertainty, thereby reducing information loss and improving decision accuracy. A case study on urban smart farm selection confirms the feasibility and superiority of the proposed approach. This study introduces the p,q-RTOFN framework with extended informational scope, develops a hybrid p,q-RTOFWPBM operator, and incorporates these advances into an extended COPRAS method to achieve more accurate multi-criteria decision-making under uncertainty. Full article

The principles of least effort and the illusion of control may influence the decision-making process. It is challenging for a decision-maker to maintain complete independence when assessing the membership and non-membership degrees of indicators. However, existing neutrosophic sets and q-rung orthopair fuzzy sets assume full independence of such information. In view of this, this paper proposes a new neutrosophic set, namely the q-type semi-dependent neutrosophic set (QTSDNS), based on the classical neutrosophic set, whose membership and non-membership degrees are interrelated. QTSDNS is a generalized form of classical semi-dependent fuzzy sets, such as the intuitionistic neutrosophic set. It contains a regulatory parameter, which allows for decision-makers to flexibly adjust the model. Furthermore, a multi-attribute group decision-making (MAGDM) algorithm is proposed by integrating QTSDNS with evidence theory to solve the supplier selection problem. The algorithm first utilizes QTSDNS to represent the preference information of experts, then employs the q-TSDNWAA (or q-TSDNWGA) operator to aggregate the evaluation information of individual experts. Following the analysis of the mathematical relationship between QTSDNS and evidence theory, evidence theory is used to aggregate the evidence from each expert to obtain the group trust interval. Then, the best supplier is determined using interval number ranking methods. Finally, a numerical example is provided to demonstrate the feasibility of the proposed method. Full article

Underwater vessel localization using passive magnetic anomaly sensing is a challenging problem due to the variability in vessel magnetic signatures and operational conditions. Data-based approaches may fail to generalize even to slightly different conditions. Thus, we propose an Attentive Neural Process (ANP) approach, in order to take advantage of its few-shot capabilities to generalize, for robust localization of underwater vessels based on magnetic anomaly measurements. Our ANP models the mapping from multi-sensor magnetic readings to position as a stochastic function: it cross-attends to a variable-size set of context points and fuses these with a global latent code that captures trajectory-level factors. The decoder outputs a Gaussian over coordinates, providing both point estimates and well-calibrated predictive variance. We validate our approach using a comprehensive dataset of magnetic disturbance fields, covering 64 distinct vessel configurations (combinations of varying hull sizes, submersion depths (water-column height over a seabed array), and total numbers of available sensors). Six magnetometer sensors in a fixed circular arrangement record the magnetic field perturbations as a vessel traverses sinusoidal trajectories. We compare the ANP against baseline multilayer perceptron (MLP) models: (1) base MLPs trained separately on each vessel configuration, and (2) a domain-randomized search (DRS) MLP trained on the aggregate of all configurations to evaluate generalization across domains. The results demonstrate that the ANP achieves superior generalization to new vessel conditions, matching the accuracy of configuration-specific MLPs while providing well-calibrated uncertainty quantification. This uncertainty-aware prediction capability is crucial for real-world deployments, as it can inform adaptive sensing and decision-making. Across various in-distribution scenarios, the ANP halves the mean absolute error versus a domain-randomized MLP (0.43 m vs. 0.84 m). The model is even able to generalize to out-of-distribution data, which means that our approach has the potential to facilitate transferability from offline training to real-world conditions. Full article

Digital ink Chinese characters (DICCs) written by international students often contain various errors and irregularities, making the recognition of these characters a highly challenging pattern recognition problem. This paper designs a graph convolutional network with agent attention (GCNAA) for recognizing DICCs written by international students. Each sampling point is treated as a vertex in a graph, with connections between adjacent sampling points within the same stroke serving as edges to create a Chinese character graph structure. The GCNAA is used to process the data of the Chinese character graph structure, implemented by stacking Block modules. In each Block module, the graph agent attention module not only models the global context between graph nodes but also reduces computational complexity, shortens training time, and accelerates inference speed. The graph convolution block module models the local adjacency structure of the graph by aggregating local geometric information from neighboring nodes, while graph pooling is employed to learn multi-resolution features. Finally, the Softmax function is used to generate prediction results. Experiments conducted on public datasets such as CASIA-OLWHDB1.0-1.2, SCUT-COUCH2009 GB1&GB2, and HIT-OR3C-ONLINE demonstrate that the GCNAA performs well even on large-category datasets, showing strong generalization ability and robustness. The recognition accuracy for DICCs written by international students reaches 98.7%. Accurate and efficient handwritten Chinese character recognition technology can provide a solid technical foundation for computer-assisted Chinese character writing for international students, thereby promoting the development of international Chinese character education. Full article

The growing ubiquity of digital platforms has enabled unprecedented participation in large-scale group decision-making processes. Nevertheless, integrating subjective linguistically expressed opinions into structured decision protocols remains a significant challenge. This paper presents a novel framework that leverages the semantic and affective capabilities of large language models to support large-scale group decision-making tasks by extracting and quantifying experts’ communicative traits—specifically clarity and trust—from natural language input. Based on these traits, participants are clustered into behavioural groups, each of which is assigned a representative preference structure and a weight reflecting its internal cohesion and communicative quality. A sentiment-informed consensus mechanism then aggregates these group-level matrices to form a collective decision outcome. The method enhances scalability and interpretability while preserving the richness of human expression. The results suggest that incorporating behavioural dimensions into large-scale group decision-making via large language models fosters fairer, more balanced, and semantically grounded decisions, offering a promising avenue for next-generation decision-support systems. Full article

In the context of the rapid development of e-commerce, the selection of optimal land plots for the construction of warehouse complexes that meet environmental, technical, and political requirements has become increasingly relevant. This task requires a comprehensive approach that accounts for a wide range of factors, including transportation accessibility, environmental conditions, geographic features, legal constraints, and more. Such an approach enhances the efficiency and sustainability of decision-making processes. This article presents a solution to the aforementioned problem that employs the use of land suitability maps generated by aggregating multiple evaluation criteria. These criteria represent the degree to which each land plot satisfies the requirements of various stakeholders and are expressed as suitability functions based on attribute values. Attributes describe different characteristics of the land plots and are represented as layers on a digital terrain map. The criteria and their corresponding attributes are classified as either quantitative or binary. Binary criteria are aggregated using the minimum operator, which filters out plots that violate any constraints by assigning them a suitability score of zero. Quantitative criteria are aggregated using the second-order Choquet integral, a method that accounts for interdependencies among criteria while maintaining computational simplicity. The criteria were developed based on statistical and environmental data obtained from an analysis of the Samara region in Russia. The resulting suitability maps are visualized as gradient maps, where land plots are categorized according to their degree of suitability—from completely unsuitable to highly suitable. This visual representation facilitates intuitive interpretation and comparison of different location options. These maps serve as an effective tool for planners and stakeholders, providing comprehensive and objective insights into the potential of land plots while incorporating all relevant factors. The proposed approach supports spatial analysis and land use planning by integrating mathematical modeling with modern information technologies to address pressing challenges in sustainable development. Full article

High dynamic range (HDR) tone mapping techniques have been widely studied to effectively represent the broad dynamic range of real-world scenes. However, generating an HDR image from multiple low dynamic range (LDR) images captured at different exposure levels can introduce ghosting artifacts in dynamic scenes. Moreover, methods that estimate HDR information from a single LDR image often suffer from inherent accuracy limitations. To overcome these limitations, this study proposes a novel image processing technique that extends the dynamic range of a single LDR image. This technique achieves the goal through leveraging a Convolutional Neural Network (CNN) to generate a synthetic Near-Infrared (NIR) image—one that emulates the characteristic of real NIR imagery being less susceptible to diffraction, thus preserving sharper outlines and clearer details. This synthetic NIR image is then fused with the original LDR image, which contains color information, to create a tone-distributed HDR-like image. The synthetic NIR image is produced using a lightweight U-Net-based autoencoder, where the encoder extracts features from the LDR image, and the decoder synthesizes a synthetic NIR image that replicates the characteristics of a real NIR image. To enhance feature fusion, a cardinality structure inspired by Extended-Efficient Layer Aggregation Networks (E-ELAN) in You Only Look Once Version 7 (YOLOv7) and a modified convolutional block attention module (CBAM) incorporating a difference map are applied. The loss function integrates a discriminator to enforce adversarial loss, while VGG, structural similarity index, and mean squared error losses contribute to overall image fidelity. Additionally, non-reference image quality assessment losses based on BRISQUE and NIQE are incorporated to further refine image quality. Experimental results demonstrate that the proposed method outperforms conventional HDR techniques in both qualitative and quantitative evaluations. Full article

Digital transformation in healthcare has introduced data privacy challenges, as hospitals struggle to protect patient information while adopting digital technologies such as AI, IoT, and cloud more rapidly than ever before. The adoption of powerful third-party Machine Learning as a Service (MLaaS) solutions for disease prediction has become a common practice. However, these solutions offer significant privacy risks when sensitive healthcare data are shared externally to a third-party server. This raises compliance concerns under regulations like HIPAA, GDPR, and Australia’s Privacy Act. To address these challenges, this paper explores a decentralized, privacy-preserving approach to train the models among multiple healthcare stakeholders, integrating Federated Learning (FL) with Homomorphic Encryption (HE), ensuring model parameters remain protected throughout the learning process. This paper proposes a novel Homomorphic Encryption-based Adaptive Tuning for Federated Learning (HEAT-FL) framework to select encryption parameters based on model layer sensitivity. The proposed framework leverages the CKKS scheme to encrypt model parameters on the client side before sharing. This enables secure aggregation at the central server without requiring decryption, providing an additional layer of security through model-layer-wise parameter management. The proposed adaptive encryption approach significantly improves runtime efficiency while maintaining a balanced level of security. Compared to the existing frameworks (non-adaptive) using 256-bit security settings, the proposed framework offers a 56.5% reduction in encryption time for 10 clients and 54.6% for four clients per epoch. Full article

Graph convolutional networks (GCNs) have demonstrated remarkable effectiveness in processing graph-structured data and have been widely adopted across various domains. Existing methods mitigate over-smoothing through selective aggregation strategies such as attention mechanisms, edge dropout, and neighbor sampling. While some approaches incorporate global structural context, they often underexplore category-aware representations and inter-category differences, which are crucial for enhancing node discriminability. To address these limitations, a novel framework, CA-NodeNet, is proposed for semi-supervised node classification. CA-NodeNet comprises three key components: (1) coarse-grained node feature learning, (2) category-decoupled multi-branch attention, and (3) inter-category difference feature learning. Initially, a GCN-based encoder is employed to aggregate neighborhood information and learn coarse-grained representations. Subsequently, the category-decoupled multi-branch attention module employs a hierarchical multi-branch architecture, in which each branch incorporates category-specific attention mechanisms to project coarse-grained features into disentangled semantic subspaces. Furthermore, a layer-wise intermediate supervision strategy is adopted to facilitate the learning of discriminative category-specific features within each branch. To further enhance node feature discriminability, we introduce an inter-category difference feature learning module. This module first encodes pairwise differences between the category-specific features obtained from the previous stage and then integrates complementary information across multiple feature pairs to refine node representations. Finally, we design a dual-component optimization function that synergistically combines intermediate supervision loss with the final classification objective, encouraging the network to learn robust and fine-grained node representations. Extensive experiments on multiple real-world benchmark datasets demonstrate the superior performance of CA-NodeNet over existing state-of-the-art methods. Ablation studies further validate the effectiveness of each module in contributing to overall performance gains. Full article

Multi-modality image fusion (MIF) aims to integrate complementary information from diverse imaging modalities into a single comprehensive representation and serves as an essential processing step for downstream high-level computer vision tasks. The existing deep unfolding-based processes demonstrate promising results; however, they often rely on deterministic priors with limited generalization ability and usually decouple from the training process of object detection. In this paper, we propose Semantic-Aware Deep Unfolded Network with Diffusion Prior (SEND), a novel framework designed for transparent and effective multi-modality fusion and object detection. SEND consists of a Denoising Prior Guided Fusion Module and a Fusion Object Detection Module. The Denoising Prior Guided Fusion Module does not utilize the traditional deterministic prior but combines the diffusion prior with deep unfolding, leading to improved multi-modal fusion performance and generalization ability. It is designed with a model-based optimization formulation for multi-modal image fusion, which is unfolded into two cascaded blocks: a Diffusion Denoising Fusion Block to generate informative diffusion priors and a Data Consistency Enhancement Block that explicitly aggregates complementary features from both the diffusion priors and input modalities. Additionally, SEND incorporates the Fusion Object Detection Module with the Denoising Prior Guided Fusion Module for object detection task optimization using a carefully designed two-stage training strategy. Experiments demonstrate that the proposed SEND method outperforms state-of-the-art methods, achieving superior fusion quality with improved efficiency and interpretability. Full article

Images captured under complex lighting conditions often suffer from local under/ overexposure and detail loss. Existing methods typically process illumination and texture information in a mixed manner, making it difficult to simultaneously achieve precise exposure adjustment and preservation of detail. To address this challenge, we propose LapECNet, an enhanced Laplacian pyramid network architecture for image exposure correction and detail reconstruction. Specifically, it decomposes the input image into different frequency bands of a Laplacian pyramid, enabling separate handling of illumination adjustment and detail enhancement. The framework first decomposes the image into three feature levels. At each level, we introduce a feature enhancement module that adaptively processes image features across different frequency bands using spatial and channel attention mechanisms. After enhancing the features at each level, we further propose a dynamic aggregation module that learns adaptive weights to hierarchically fuse multi-scale features, achieving context-aware recombination of the enhanced features. Extensive experiments with public benchmarks on the MSEC dataset demonstrated that our method gave improvements of 15.4% in PSNR and 7.2% in SSIM over previous methods. On the LCDP dataset, our method demonstrated improvements of 7.2% in PSNR and 13.9% in SSIM over previous methods. Full article

Underwater image enhancement (UIE) is a key technology in the fields of underwater robot navigation, marine resources development, and ecological environment monitoring. Due to the absorption and scattering of different wavelengths of light in water, the quality of the original underwater images usually deteriorates. In recent years, UIE methods based on deep neural networks have made significant progress, but there still exist some problems, such as insufficient local detail recovery and difficulty in effectively capturing multi-scale contextual information. To solve the above problems, a Multi-Scale Contextual Fusion Residual Network (MCFR-Net) for underwater image enhancement is proposed in this paper. Firstly, we propose an Adaptive Feature Aggregation Enhancement (AFAE) module, which adaptively strengthens the key regions in the input images and improves the feature expression ability by fusing multi-scale convolutional features and a self-attention mechanism. Secondly, we design a Residual Dual Attention Module (RDAM), which captures and strengthens features in key regions through twice self-attention calculation and residual connection, while effectively retaining the original information. Thirdly, a Multi-Scale Feature Fusion Decoding (MFFD) module is designed to obtain rich contexts at multiple scales, improving the model’s understanding of details and global features. We conducted extensive experiments on four datasets, and the results show that MCFR-Net effectively improves the visual quality of underwater images and outperforms many existing methods in both full-reference and no-reference metrics. Compared with the existing methods, the proposed MCFR-Net can not only capture the local details and global contexts more comprehensively, but also show obvious advantages in visual quality and generalization performance. It provides a new technical route and benchmark for subsequent research in the field of underwater vision processing, which has important academic and application values. Full article

Hippophae rhamnoides subsp. sinensis Rousi has high ecological and medicinal value, and it is an important plant resource unique to the arid regions of Northwest China. Exploring the influence of climate characteristics and soil factors on the composition, diversity, and function of the rhizosphere bacterial community of Chinese seabuckthorn is of great value for developing and popularizing characteristic plant resources in the arid regions of Northwest China. In this study, the rhizosphere soil of 13 Chinese seabuckthorn distribution areas in the northwest of China was taken as the research object, the bacterial community map was constructed based on 16S rRNA gene high-throughput sequencing technology, and the species abundance composition, structural diversity, molecular co-occurrence network, and phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt), as well as the function of rhizosphere soil bacterial community, were systematically studied. Combined with Mantel test and redundancy analysis (RDA), the key habitat factors driving the rhizosphere soil bacterial community structure of Chinese seabuckthorn were explored. The results showed that: (1) The number of amplicon sequence variants (ASVs) in rhizosphere soil bacterial community of Chinese seabuckthorn was the highest in S2(3072) and the S12(3637), and the lowest in the S11(1358) and S13(1996). The rhizosphere soil bacterial community was primarily composed of Proteobacteria, Actinobacteriota, and Acidobacteriota. Except for the S6 and S11 habitats, the dominant bacterial genera were mainly Achromobacter, Acidobacter (RB41), and Sphingomonas. (2) The α and β diversity of rhizosphere soil bacterial communities of Chinese seabuckthorn across 13 distribution areas were significantly different. The number of operational taxonomic units (OTUs), Ace index, and Chao 1 index of soil bacterial community in the S12 distribution area are the highest, and they are the lowest in S11 distribution area, with significant differences. The aggregation of soil bacterial communities in the S5 and S10 distribution areas is the highest, while it is the lowest in the S6 and S11 distribution areas. (3) PICRUSt function classification of soil bacteria showed that Metabolism and Genetic Information Processing functions were the strongest across all distribution areas, with S10 exhibiting higher functional capacity than other areas and S11 showing the weakest. (4) Cluster analysis revealed that soil bacteria across the 13 distribution areas were clustered into two groups, with S10 and S12 distribution areas as one group (Group 1) and the remaining 11 distribution areas as another group (Group 2). (5) Redundancy analysis revealed that pH was the key soil environmental factor driving the rhizosphere soil bacterial community α-diversity of Chinese seabuckthorn, followed by altitude (ALT) and soil water content (SWC). In summary, Chinese seabuckthorn prefers neutral to alkaline soils, and environmental factors play an important role in driving bacterial diversity, community structure, functional profiles, and co-occurrence networks in rhizosphere soil of Chinese seabuckthorn. Full article

As a core task in remote sensing image processing, change detection plays a vital role in dynamic surface monitoring for environmental management, urban planning, and agricultural supervision. However, existing methods often suffer from missed detection of small targets and pseudo-change interference, stemming from insufficient modeling of multi-scale feature coupling and spatio-temporal differences due to factors such as background complexity and appearance variations. To this end, we propose a Differential-Perception-Enhanced Multi-Scale Attention Network for Remote Sensing Image Change Detection (MDNet), an optimized framework integrating multi-scale feature extraction, cross-scale aggregation, difference enhancement, and context modeling. Through the parallel collaborative mechanism of the designed Multi-Scale Feature Extraction Module (EMF) and Cross-Scale Adjacent Semantic Information Aggregation Module (CASAM), multi-scale semantic learning is strengthened, enabling fine-grained modeling of change targets of different sizes and improving small-target-detection capability. Meanwhile, the Differential-Perception-Enhanced Module (DPEM) and Transformer structure are introduced for global–local coupled modeling of spatio-temporal differences. They enhance spectral–structural differences to form discriminative features, use self-attention to capture long-range dependencies, and construct multi-level features from local differences to global associations, significantly suppressing pseudo-change interference. Experimental results show that, on three public datasets (LEVIR-CD, WHU-CD, and CLCD), the proposed model exhibits superior detection performance and robustness in terms of quantitative metrics and qualitative analysis compared with existing advanced methods. Full article