Search Results (456)

Accurate building roof classification from airborne LiDAR point clouds is fundamental to reliable three-dimensional (3D) urban reconstruction. While supervoxel-based methods offer efficiency and resilience to uneven point density, their performance is critically undermined by cross-boundary segmentation errors—a direct consequence of random seed initialization that merges geometrically similar yet semantically distinct objects. To address this root cause, this study proposes Interior-Anchored Mean-Shift (IAMS), a novel supervoxel segmentation framework that rethinks seed placement as a geometry-aware interior localization problem. By integrating local geometric consistency point density, and spatial correlation into a unified kernel density estimator, supplemented by density-adaptive voxel weighting and a semi-variogram-driven bandwidth, IAMS reliably anchors seeds within object interiors, yielding highly homogeneous supervoxels without post-processing. Extensive experiments on three diverse airborne LiDAR datasets demonstrated that IAMS consistently outperformed state-of-the-art baselines. On the International Society for Photogrammetry and Remote Sensing (ISPRS) Vaihingen benchmark, our approach improved roof classification completeness, correctness, and quality by up to 7.1% (per-object) over the conventional Voxel Cloud Connectivity Segmentation (VCCS) algorithm while being significantly faster than recent boundary-preserving alternatives. Critically, IAMS maintains robust performance under challenging conditions, including sparse sampling and dense vegetation occlusion, making it a practical solution for real-world urban remote sensing. Full article

Rising atmospheric CO₂ levels have increased the demand for robust, scalable adsorbents for practical CO₂ capture and separation. Porous organic polymers (POPs) are attractive candidates because their pore architecture and binding site properties can be precisely tuned via building blocks and linkage formation. This review summarizes experimental and computational studies of azo-linked POPs and, more broadly, nitrogen–nitrogen (N–N) linked systems, emphasizing how synthetic routes, building blocks, and framework topology govern CO₂ uptake. We highlight key synthetic strategies and representative systems, including porphyrin–azo networks, and discuss the relatively sparse experimental literature on alternative N–N linked POPs incorporating azoxy and azodioxy motifs. Emphasis is placed on reversible nitroso/azodioxide chemistry as a potential pathway to ordered porous organic materials. Computational studies provide a practical route to connect structure with adsorption behavior in largely amorphous or partially ordered networks. We review hierarchical workflows combining periodic DFT and electrostatic potential properties, grand canonical Monte Carlo (GCMC) simulations, and binding energy calculations to rationalize trends and identify favorable binding environments. Computational findings demonstrate that pore accessibility and stacking models can strongly influence predicted CO₂ adsorption. This review provides guidelines for designing POPs with enhanced CO₂ adsorption, offering an outlook and discussing challenges for future studies. Full article

Metal–polymer hybrid joints are gaining importance as they combine high structural rigidity with a low weight. Additive manufacturing processes such as the laser powder bed fusion process (L-PBF) enable the production of complex metallic lattice structures that allow for form-fitting force transmission between the metal and polymer as mechanical interlock elements. In this work, metal–polymer hybrid compounds with additively manufactured transition zones are systematically investigated and mechanically evaluated. Three different lattice geometries (z4A, z8A, z8V) were fabricated from maraging steel (1.2709) using L-PBF and then hybridised with injection moulding using polypropylene (PP C7069-100NA). Mechanical characterisation was performed by tensile tests according to DIN EN ISO 527, in combination with statistical analyses and an analytical serial three-spring model to determine the homogenised elasticity modulus of the transition zone. The results show significant geometry-related differences in tensile strength, maximum force, and effective stiffness. The A-shaped transition zone geometry (z4A) achieves the highest mechanical performance and up to 82% of the tensile strength of the pure polymer, while the V-shaped transition zone geometry (z8V) has significantly lower load-bearing capacities. Variance analysis shows a dominant geometric influence with effect strength of η² ≈ 0.99. The analytically predicted stiffness values match the experimental results within 5–10%. This work demonstrates a reproducible, simulation-sparse approach to the analysis and design of metal–polymer hybrid connections. Full article

Accurate weed mapping in cereal fields requires pixel-level segmentation from unmanned aerial vehicle (UAV) imagery that remains reliable across fields, seasons, and illumination. Existing multispectral pipelines often depend on thresholded vegetation indices, which are brittle under radiometric drift and mixed crop–weed pixels, or on single-stream convolutional neural network (CNN) and Transformer backbones that ingest stacked bands and indices, where radiance cues and normalized index cues interfere and reduce sensitivity to small weed clusters embedded in crop canopy. We propose VISA (Vegetation Index and Spectral Attention), a two-stream segmentation network that decouples these cues and fuses them at native resolution. The radiance stream learns from calibrated five-band reflectance using local residual convolutions, channel recalibration, spatial gating, and skip-connected decoding, which preserve fine textures, row boundaries, and small weed structures that are often weakened after ratio-based index compression. The index stream operates on vegetation-index maps with windowed self-attention to model local structure efficiently, state-space layers to propagate field-scale context without quadratic attention cost, and Slot Attention to form stable region descriptors that improve discrimination of sparse weeds under canopy mixing. To support supervised training and deployment-oriented evaluation, we introduce BAWSeg, a four-year UAV multispectral dataset collected over commercial barley paddocks in Western Australia, providing radiometrically calibrated blue, green, red, red edge, and near-infrared orthomosaics, derived vegetation indices, and dense crop, weed, and other labels with leakage-free block splits. On BAWSeg, VISA achieves 75.6% mean Intersection over Union (mIoU) and 63.5% weed Intersection over Union (IoU) with 22.8 M parameters, outperforming a multispectral SegFormer-B1 baseline by 1.2 mIoU and 1.9 weed IoU. Under cross-plot and cross-year protocols, VISA maintains 71.2% and 69.2% mIoU, respectively. The full BAWSeg benchmark dataset, VISA code, trained model weights, and protocol files will be released upon publication. Full article

Populus euphratica is a critical constructive species in arid desert regions, serving as a “natural barrier” for oasis protection. The sustainable management of Populus euphratica forests is directly related to regional ecological security, and the fine identification of sparse Populus euphratica forests is essential for the conservation of natural Populus euphratica forests. Currently, most mapping studies on Populus euphratica distribution focus on the extraction of dense, contiguous Populus euphratica forests, with insufficient attention paid to the identification of sparse Populus euphratica forests. This study utilizes Gaofen-2 (GF-2) satellite imagery as the data source and takes a typical sparse Populus euphratica forests distribution area in the Tarim River Basin as the study site. It systematically evaluates the performance of nine mainstream deep learning models, including U-Net, DeepLabV3+, and SegFormer, in the task of sparse Populus euphratica forests identification. The results indicate that: (1) The false-color sample set, synthesized from near-infrared, red, and green bands, contributes to improved model accuracy. Compared to the true-color (red, green, blue bands) dataset, the average Intersection over Union (IoU) of the nine models shows a relative improvement of approximately 20%. (2) For the sparse Populus euphratica forests identification task based on the false-color dataset, four models—U-Net, U-Net++, MA-Net, and DeepLabV3+—exhibited excellent performance, with IoU exceeding 75%. (3) Using U-Net as the baseline model, this study integrated the max-pooling indices mechanism, atrous spatial pyramid pooling, and residual connection modules to construct a semantic segmentation network tailored for sparse Populus euphratica forests, named Sparse Populus euphratica Segmentation Network (SPS-Net). This model achieved an IoU of 80%, a relative improvement of approximately 6.3% over the baseline model, and demonstrated good stability in large-scale classification tests. The identification scheme for sparse Populus euphratica forests constructed using GF-2 imagery and deep learning models proposed in this study can provide effective technical support for the refined monitoring and protection of natural Populus euphratica forests. Full article

Next Point-of-Interest (POI) recommendations are pivotal for enhancing location-based services; however, accurate prediction remains challenging due to the complex interplay between dynamic user preferences and spatiotemporal constraints. Existing graph-sequence hybrids often fail to unify these dimensions, typically treating temporal contexts as disjoint features or neglecting implicit collaborative signals within sparse user trajectories. This fragmentation limits the ability to capture high-order dependencies in user mobility. To address these challenges, we propose UPTRec, a unified framework that synergizes social, spatial, and temporal reasoning. UPTRec constructs a TF-IDF-weighted user similarity graph to recover latent social connections and a flow-based POI-transition graph to encode sequential mobility patterns. These structural priors are fused with fine-grained temporal-category embeddings (utilizing Time2Vec and periodic encoding) via a multi-layer Transformer encoder to comprehensively capture user behavior. Extensive experiments on three real-world datasets (NYC, TKY, and CA) demonstrate that UPTRec achieves state-of-the-art performance among the compared baselines under the same experimental settings. On the NYC dataset, UPTRec yields a Top-1 Accuracy of 25.76% and a Mean Reciprocal Rank (MRR) of 0.3879, representing a relative improvement of 5.8% and 7.1% over the strongest baseline (GETNext). These results validate the efficacy of jointly modeling collaborative and spatiotemporal dependencies. Full article

Motion and path planning are fundamental to intelligent robotic systems, enabling navigation. The objective is to generate collision-free trajectories in obstacle-rich configuration spaces (C-spaces) while meeting performance constraints. In environments with narrow passages planning becomes especially difficult, as feasible regions have low measure and are rarely reached by random sampling. Classical sampling-based planners are probabilistically complete but inefficient in such regions. Learning-based planners like MPNet offer fast inference but often produce infeasible paths in cluttered areas, requiring expensive postprocessing. To address this trade-off, we propose a hybrid framework that combines improved sampling, structural abstraction, and neural prediction. A modified bridge-test sampler applies directional perturbations and corridor checks to generate reliable narrow passage samples. These are clustered into a sparse set of representative bridge points, which serve as nodes in a global graph. At query time, a greedy heuristic search explores this graph, using a neural local segment generator to connect nodes. We validate the approach on 2D maze maps, 3D voxel environments, and a 12-DOF manipulator performing a plugging task inside a simulated nuclear steam generator. Across all tasks, our method significantly outperforms classical and learning-based baselines in terms of success rate and planning time in narrow-passage-dominated scenarios. The inclusion of the repair module, under relaxed assumptions, also allows the framework to retain a generalized form of probabilistic completeness. Full article

Fifth generation (5G) networks have significantly enhanced connectivity, speed, and reliability, transforming industries with faster and more efficient communication. The Internet of Things (IoT) has introduced unprecedented convenience and automation, revolutionizing sectors such as healthcare, finance, and smart infrastructure. However, both 5G networks and IoT environments are experiencing a high frequency of attacks. Intrusion detection systems (IDSs) built on federated learning (FL) are being proposed to boost data privacy and security. However, these IDSs are related with the inherent drawbacks of FL, namely the existence of non-independently and identically (non-IID) distributed features and the machine learning model complexity. To address these limitations, we present a study that integrates a Mixture of Experts (MoE) into an FL setting in the task of intrusion detection. Specifically, to mitigate the issues of model complexity within the FL setting, we use a sparsely gated MoE layer consisting of a router/gating network and a set of experts. Only a subset of experts is selected via applying noisy top-k gating. To alleviate the issue of non-IID data, we adopt the Label-based Dirichlet Partition method, utilizing Dirichlet sampling with a hyperparameter $\alpha$ to simulate a label-based non-IID data distribution. Four FL strategies are employed. We perform our experiments on the 5G-NIDD and BoT-IoT datasets. Findings show that the proposed approach achieves competitive performance across both datasets under heterogeneous federated settings. Full article

Causal inference in text data has been a demanding objective in the field of natural language processing, mainly due to the intrinsic ambiguity and context sensitivity inherent in data, inducing uncertainty. Diminishing this uncertainty is essential in identifying reliable causal connections and advancing predictive consistency. In this research, we introduce an uncertainty-aware ensemble architecture that combines multiple text embedding schemes with both linear and nonlinear classifiers to boost causal text detection. Both sparse and neural-level embeddings were employed, and then combined it with an ensemble weighting approach based on two uncertainty estimation techniques, namely entropy-based and KL divergence-based. Unlike conventional ensemble methods with uniform or fixed voting strategies, our approach assigns weights inversely proportional to classifier uncertainty, ensuring that confident models exert greater influence on the final decisions. Our results show that TF-IDF, through its effective word frequency weighting scheme, consistently outperforms other embedding techniques, achieving better performance across both linear and nonlinear classifiers on both datasets (News Corpus and CausalLM–Adjective group). The experimental results show that our uncertainty-aware ensemble approach enhances both calibration and confidence predictions. Entropy-based weighting improves confidence in the case of linear classifiers with accuracy, F1-score, entropy and prediction confidence values of 94.3%, 94.0%, 0.382 and 0.774, respectively, while in the case of nonlinear classifiers the KL divergence-based weighting acquires a better performance with an accuracy of 97.6%, F1-score of 97.2%, KL Mean value of around 0.055 and LogLoss of 0.221. Full article

Rapid urbanisation and increasing heat extremes pose significant challenges for megacities in the Global South. This study develops a configuration-sensitive assessment of blue-green space (BGS) cooling in Delhi, a Global South megacity facing intensified heat. Using satellite imagery and statistical modelling, we quantify how land cover and patch structure regulate land surface temperature (LST). Satellite imagery was used to derive LST, and six land-cover classes were mapped using supervised classification. Spectral indices and proximity metrics were calculated, land-cover patches were delineated, and their thermal behaviour was analysed using patch-level LST statistics. Delhi exhibits a heterogeneous urban heat island (UHI) surface, with LST spanning 19.8–38.6 °C and built-up land dominating (743.50 km²), while BGS remains limited and fragmented. Warming scaled almost linearly with built-up patch size (R² = 0.98), with mean LST rising from 22.6 °C (<20,000 m²) to 27.4 °C (>500,000 m²). Cooling strengthened with BGS spatial dominance as dense vegetation declined from 23.8 to 22.1 °C (R² = 0.98), sparse vegetation from 24.3 to 22.2 °C, and water bodies from 21.4 to 18.8 °C (R² = 0.89) across increasing size classes. Correlations identified impervious surfaces as primary warming controls, while moisture and vegetation were cooling indicators. Random Forest-SHAP confirmed modified bare soil index (MBSI) and normalised difference built-up index (NDBI) as dominant predictors, with cooling from modified normalised difference water index (MNDWI) and comparatively conditional effects of normalised difference vegetation index (NDVI). Impervious and exposed surfaces govern Delhi’s thermal baseline, while BGS acts as a modifier whose benefits emerge when patches are large, connected, and integrated. These findings support shifting from area-based greening targets to morphology-based planning that protects connected blue-green corridors. Full article

Automated optical inspection (AOI) for printed circuit boards (PCBs) requires localizing small, sparse defects under illumination drift and minor placement misalignment, while supporting fast, auditable pass/fail decisions. This paper presents a training-free, reference-based digital image processing framework with no learning/training stage that compares each defective query image with a small library of defect-free reference templates (for the same PCB layout/revision) using a small set of interpretable control parameters. A reference is selected by coarse-to-fine matching (fast pre-screening followed by SSIM refinement on a central region), and an optional global alignment is applied only when it increases SSIM to limit defect-driven over-correction. Defects are highlighted by a defect-likelihood field that fuses an SSIM-derived structural dissimilarity map with a normalized absolute-difference map, followed by connected-component extraction to produce confidence-ranked bounding boxes. The method achieves Precision = 0.9663, Recall = 0.9987, and F1 = 0.9822 at the best-F1 operating point (0.149 false positives per image). Under the adopted box-matching protocol, average precision reaches 0.984. Precision–recall and FROC curves are reported to support threshold selection under different false-alarm budgets. Full article

This study develops a conceptual framework for characterizing reservoir architecture in multi-component, discrete systems using pressure transient analysis (PTA), aimed at calibrating inflow geometry prior to full-field dynamic simulation for subsurface gas storage applications such as CO₂ and hydrogen. A secondary objective is to identify variations in permeability over time by analyzing flow capacity trends and evaluating the dynamic influence of faults and fractures. The analysis is based on a gas-condensate field comprising seven wells and four zones (A, B, C, D), using integrated dynamic datasets including extended well tests (EWTs), mud loss, production logs, and production data. Detailed interpretation of PX-1’s EWT indicated delayed re-pressurization and persistent under-pressure, suggesting a compartmentalized or transient system with limited gas-in-place connectivity. Four reservoir architecture concepts were developed: (1) lithology-dominated inflow, (2) structurally controlled inflow, (3) discrete, weakly connected compartments, and (4) transient-dominated systems with tight matrix GIIP. These concepts informed four reservoir models: matrix-only (M), areal heterogeneity (A), sparse bodies (B), and sparse networks (S). Application of these models across other wells revealed consistent localized KH (permeability–thickness product) behavior, with all models fitting short-duration data comparably. However, only sparse drainage models (B/S) adequately matched PX-1’s EWT response. PTA results confirm that well tests constrain KH locally but provide limited insight into large-scale reservoir architecture. EWTs may reach ~1 km, while shorter tests are confined to ~200–400 m, typically within one to two simulation grid blocks. This study demonstrates how integrating PTA with multi-scale data improves characterization of naturally fractured, tight carbonate reservoirs and supports reservoir simulation and history matching for hydrogen storage evaluation. Based on reservoir simulations, this study concluded that naturally fractured carbonate gas reservoirs can provide significant storage and injection capacities for underground hydrogen storage. This study exemplifies how to characterize the naturally fractured tight carbonate reservoirs by integrating multi-scale and multi-dimensional data such as PTA. Furthermore, this study assists in gridding for full-field reservoir models, for history matching and quantifying the potential of hydrogen storage in these complex reservoirs. The proposed workflow provides an uncertainty-bounded reservoir characterization framework and should not be interpreted as a complete field-design methodology for hydrogen storage. The modeling does not explicitly couple geomechanical fracture growth, hydrogen diffusion, long-term geochemical reactions, or caprock integrity degradation. Therefore, the presented storage scenarios represent technically feasible cases under defined assumptions. Comprehensive site-specific geomechanical and containment assessments are required prior to field-scale implementation. Full article

Ecological network optimization can enhance ecological connectivity, regional ecological stability, and carbon sink capacity. Current research on ecological networks employs single-perspective optimization, which overlooks the synergistic requirements between network topological characteristics and the dual carbon goals. It lacks a comprehensive, systemic optimization framework. Focusing on the Beijing–Tianjin–Hebei region, the work constructs an ecological network by integrating ecosystem services, morphological spatial pattern analysis (MSPA), and circuit theory. A framework integrating barrier removal, structural robustness, and carbon sink enhancement is proposed, incorporating ecological barrier identification, complex network theory, and carbon offset patterns for multi-objective structural and functional optimization. The optimized network is evaluated using structural metrics, robustness analysis, and carbon sequestration validation. The network comprises 41 ecological sources and 102 corridors, exhibiting a dense northwest and sparse southeast distribution. Ecological barriers totaling 565.56 km² are removed to improve connectivity in the region. An edge-addition strategy introduces 12 nodes and 49 edges, enhancing connectivity, stability, and carbon sink capacity. Restoration priorities are set with the phased objectives of removing barriers, connecting topological weak points, and optimizing low-value carbon offset areas. Shifting the focus from structural connectivity to integrated function, the work contributes a methodological framework for advancing ecological security and carbon neutrality in urban agglomerations. Full article

Segmenting CAD models into developable surface patches is a fundamental problem in geometric modeling and manufacturing-oriented applications. Existing approaches often rely on discrete Gaussian curvature estimation or Gauss map analysis; however, their performance on CAD meshes is frequently hindered by numerical instability, sensitivity to mesh tessellation, and complex parameter tuning. In this work, we propose a simple and robust method for developable surface segmentation based on a sparse normal discontinuity prior. Our key observation is that industrial CAD models are typically composed of large developable regions separated by a sparse set of sharp creases and edges. Consequently, segmentation boundaries correspond to sparse discontinuities in the surface normal field rather than continuous variations in curvature. Based on this perspective, we formulate developable surface segmentation as the detection of sparse normal jump discontinuities. In the discrete setting of triangle meshes, this formulation naturally leads to a dihedral angle-based approach that avoids explicit curvature estimation and admits an efficient graph-based solution. The proposed algorithm consists of face normal computation, dihedral angle-based boundary detection, and connected component extraction on a thresholded face adjacency graph. The method requires only a single geometrically interpretable parameter and naturally aligns segmentation boundaries with sharp features commonly found in CAD models. Experimental results on a diverse set of industrial CAD meshes, including standard benchmarks widely used in related research, demonstrate that the proposed approach achieves robust and accurate segmentation, as validated by both visual coherence and quantitative developability metrics. Full article

Satellite–terrestrial spectrum sensing plays a crucial role in enhancing spectrum efficiency through reusing spectra. However, in a satellite–terrestrial converged system, the large SNR range, non-Gaussian signal characteristics and noise uncertainty pose significant challenges for spectrum sensing. In this paper, we investigate a downlink spectrum sensing framework where multi-terrestrial BSs act as a secondary system to sense idle satellite spectra through a multi-domain feature-level sensing signal fusion. To enhance the characterization of signal/noise features, we provide a fusion strategy of multi-features including energy, power spectral density, cyclic autocorrelation function, higher-order moments, sparse ratio, and I/Q samples, constructing two feature tensors of statistical features and an I/Q component. Then, we propose a deep-learning-enabled multi-feature fusion spectrum sensing method (DL-MFFSSnet) based on a dual-branch deep neural network architecture with the constructed two feature tensors as inputs. In the statistical feature processing branch, CNN and channel self-attention are incorporated to capture intra-channel correlations and inter-channel relative contributions of different feature modalities. In the I/Q branch, multi-scale dilated convolutions and spatial self-attention are introduced to analyze dependencies across different temporal positions and multi-scale spatial features. The feature map extracted from both branches passed through fully connected layers for deepwise feature fusion, achieving accurate spectrum sensing. Extensive simulation results demonstrate that the DL-MFFSSnet method outperforms the existing state-of-the-art algorithms. Full article