Search Results (1,394)

Hyperspectral reconstruction (SR) refers to the computational process of generating high-dimensional hyperspectral images (HSI) from low-dimensional observations. However, the superior performance of most supervised learning-based reconstruction algorithms is predicated on the availability of fully labeled three-dimensional data. In practice, this requirement demands complex optical paths with dual high-precision registrations and stringent calibration. To address this gap, we extend the fully supervised paradigm to a semi-supervised setting and propose SSHSR, a semi-supervised SR method for scenarios with limited spectral annotations. The core idea is to leverage spectrally aware mini-patches (SA-MP) as guidance and form region-level supervision from averaged spectra, so it can learn high-quality reconstruction without dense pixel-wise labels over the entire image. To improve reconstruction accuracy, we replace the conventional fixed-form Tikhonov physical layer with an optimizable version, which is then jointly trained with the deep network in an end-to-end manner. This enables the collaborative optimization of physical constraints and data-driven learning, thereby explicitly introducing learnable physical priors into the network. We also adopt a reconstruction network that combines spectral attention with spatial attention to strengthen spectral–spatial feature fusion and recover fine spectral details. Experimental results demonstrate that SSHSR outperforms existing state-of-the-art (SOTA) methods on several publicly available benchmark datasets, as well as on remote sensing and real-world scene data. On the GDFC remote sensing dataset, our method yields a 6.8% gain in PSNR and a 22.1% reduction in SAM. Furthermore, on our self-collected real-world scene dataset, our SSHSR achieves a 6.0% improvement in PSNR and a 11.9% decrease in SAM, confirming its effectiveness under practical conditions. Additionally, the model has only 1.59 M parameters, which makes it more lightweight than MST++ (1.62 M). This reduction in parameters lowers the deployment threshold while maintaining performance advantages, demonstrating its feasibility and practical value for real-world applications. Full article

Background/Objectives: The segmentation of three-dimensional radiological images constitutes a fundamental task in medical image processing for isolating tumors from complex datasets in computed tomography or magnetic resonance imaging. Precise visualization, volumetry, and treatment monitoring are enabled, which are critical for oncology diagnostics and planning. Volumetric analysis surpasses standard criteria by detecting subtle tumor changes, thereby aiding adaptive therapies. The objective of this study was to develop an enhanced, interactive Graphcut algorithm for 3D DICOM segmentation, specifically designed to improve boundary accuracy and 3D modeling of breast and brain tumors in datasets with heterogeneous tissue intensities. Methods: The standard Graphcut algorithm was augmented with a clustering mechanism (utilizing k = 2–5 clusters) to refine boundary detection in tissues with varying intensities. DICOM datasets were processed into 3D volumes using pixel spacing and slice thickness metadata. User-defined seeds were utilized for tumor and background initialization, constrained by bounding boxes. The method was implemented in Python 3.13 using the PyMaxflow library for graph optimization and pydicom for data transformation. Results: The proposed segmentation method outperformed standard thresholding and region growing techniques, demonstrating reduced noise sensitivity and improved boundary definition. An average Dice Similarity Coefficient (DSC) of 0.92 ± 0.07 was achieved for brain tumors and 0.90 ± 0.05 for breast tumors. These results were found to be comparable to state-of-the-art deep learning benchmarks (typically ranging from 0.84 to 0.95), achieved without the need for extensive pre-training. Boundary edge errors were reduced by a mean of 7.5% through the integration of clustering. Therapeutic changes were quantified accurately (e.g., a reduction from 22,106 mm³ to 14,270 mm³ post-treatment) with an average processing time of 12–15 s per stack. Conclusions: An efficient, precise 3D tumor segmentation tool suitable for diagnostics and planning is presented. This approach is demonstrated to be a robust, data-efficient alternative to deep learning, particularly advantageous in clinical settings where the large annotated datasets required for training neural networks are unavailable. Full article

Three-dimensional (3D) modeling technologies are increasingly vital in civil engineering, providing precise digital representations of infrastructure for analysis, supervision, and planning. This study presents a comparative assessment of Neural Radiance Fields (NeRFs) and digital photogrammetry using a real-world case study involving a terrace at the Civil Engineering School of the Pontificia Universidad Católica de Valparaíso. The comparison is motivated by the operational complexity of image acquisition campaigns, where large image datasets increase flight time, fieldwork effort, and survey costs. Both techniques were evaluated across varying levels of data availability to analyze reconstruction behavior under progressively constrained image acquisition conditions, rather than to propose new algorithms. NeRF and photogrammetry were compared based on visual quality, point cloud density, geometric accuracy, and processing time. Results indicate that NeRF delivers fast, photorealistic outputs even with reduced image input, enabling efficient coverage with fewer images, while photogrammetry remains superior in metric accuracy and structural completeness. The study concludes by proposing an application-oriented evaluation framework and potential hybrid workflows to guide the selection of 3D modeling technologies based on specific engineering objectives, survey design constraints, and resource availability while also highlighting how AI-based reconstruction methods can support emerging digital workflows in infrastructure monitoring under variable or limited data conditions. Full article

Facial information carries key personal privacy, and it is crucial to ensure its security through encryption. Traditional encryption for portrait images typically processes the entire image, despite the fact that most regions lack sensitive facial information. This approach is notably inefficient and imposes unnecessary computational burdens. To address this inefficiency while maintaining security, we propose a novel dual-region encryption model for portrait images. Firstly, a Multi-task Cascaded Convolutional Network (MTCNN) was adopted to efficiently segment facial images into two regions: facial and non-facial. Subsequently, given the high sensitivity of facial regions, a robust encryption scheme was designed by integrating a CNN-based key generator, the proposed three-dimensional Multi-module Nonlinear Feedback-coupled Chaotic System (3D-MNFC), DNA encoding, and bit reversal. The 3D-MNFC incorporating time-varying parameters, nonlinear terms and state feedback terms and coupling mechanisms has been proven to exhibit excellent chaotic performance. As for non-facial regions, the Logistic map combined with XOR operations is used to balance efficiency and basic security. Finally, the encrypted image is obtained by restoring the two ciphertext images to their original positions. Comprehensive security analyses confirm the exceptional performance of the regional model: large key space ($2^{536}$) and near-ideal information entropy (7.9995), NPCR and UACI values of 99.6055% and 33.4599%. It is worth noting that the model has been verified to improve efficiency by at least 37.82%. Full article

Effectively fusing the rich semantic information from camera images with the high-precision geometric measurements provided by LiDAR point clouds is a key challenge in mobile robot environmental perception. To address this problem, this paper proposes a highly extensible instance-aware fusion framework designed to achieve efficient alignment and unified modeling of heterogeneous sensory data. The proposed approach adopts a modular processing pipeline. First, semantic instance masks are extracted from RGB images using an instance segmentation network, and a projection mechanism is employed to establish spatial correspondences between image pixels and LiDAR point cloud measurements. Subsequently, three-dimensional bounding boxes are reconstructed through point cloud clustering and geometric fitting, and a reprojection-based validation mechanism is introduced to ensure consistency across modalities. Building upon this representation, the system integrates a data association module with a Kalman filter-based state estimator to form a closed-loop multi-object tracking framework. Experimental results on the KITTI dataset demonstrate that the proposed system achieves strong 2D and 3D detection performance across different difficulty levels. In multi-object tracking evaluation, the method attains a MOTA score of 47.8 and an IDF1 score of 71.93, validating the stability of the association strategy and the continuity of object trajectories in complex scenes. Furthermore, real-world experiments on a mobile computing platform show an average end-to-end latency of only 173.9 ms, while ablation studies further confirm the effectiveness of individual system components. Overall, the proposed framework exhibits strong performance in terms of geometric reconstruction accuracy and tracking robustness, and its lightweight design and low latency satisfy the stringent requirements of practical robotic deployment. Full article

Hyperspectral image (HSI) classification is pivotal in remote sensing, yet deep learning models, particularly Transformers, remain susceptible to spurious spectral–spatial correlations and suffer from limited interpretability. These issues stem from their inability to model the underlying causal structure in high-dimensional data. This paper introduces the Causal Attention Transformer (CAT), a novel architecture that integrates causal inference with a hierarchical CNN-Transformer backbone to address these limitations. CAT incorporates three key modules: (1) a Causal Attention Mechanism that enforces temporal and spatial causality via triangular masking and axial decomposition to eliminate spurious dependencies; (2) a Dual-Path Hierarchical Fusion module that adaptively integrates spectral and spatial causal features using learnable gating; and (3) a Linearized Causal Attention module that reduces the computational complexity from $O\left(N^2\right)$ to $O\left(N\right)$ via kernelized cumulative summation, enabling scalable high-resolution HSI processing. Extensive experiments on three benchmark datasets (Indian Pines, Pavia University, Houston2013) demonstrate that CAT achieves state-of-the-art performance, outperforming leading CNN and Transformer models in both accuracy and robustness. Furthermore, CAT provides inherently interpretable spectral–spatial causal maps, offering valuable insights for reliable remote sensing analysis. Full article

This study aimed to develop a robust morphometric-based framework for classifying Apis cerana populations using deep learning and machine learning approaches. Previous studies on Apis cerana population differentiation have primarily relied on manual morphometrics or genetic markers, which are labor-intensive and often lack scalability for large image-based datasets. Forewing landmarks were automatically detected through a deep learning model employing a heatmap regression and Hourglass Network architecture. The extracted coordinates were processed by Principal Component Analysis (PCA) for dimensionality reduction, and shape alignment was further refined through Procrustes ANOVA to minimize non-biological variation. Nine machine learning algorithms were trained and compared under identical preprocessing and validation settings. Among them, the Extra Trees classifier achieved the highest accuracy (99.7%) in distinguishing the three populations—A. cerana cerana from China and A. cerana indica from Thailand, the northern and southern populations. After applying error-based data filtering and retraining, classification accuracy improved further, with almost perfect population separation. The Procrustes ANOVA confirmed that individual variation significantly exceeded residual error (Pillai’s trace = 1.13, p < 0.0001), validating the biological basis of shape differences. Mahalanobis distance and permutation tests (10,000 rounds) revealed significant morphological divergence among populations (p < 0.0001). The integration of geometric alignment and ensemble learning demonstrated a highly reliable strategy for population identification, supporting morphometric and evolutionary studies in Apis cerana. Full article

Mass production in product design typically relies on standardized geometries and dimensions to accommodate a broad user population. However, when products are required to interface directly with the human body, such generalized design approaches often result in inadequate fit and reduced user comfort. This limitation highlights the necessity of fully personalized design methodologies based on individual anthropometric characteristics. This paper presents a novel application that automates the design of custom-fit sunglasses through the integration of Artificial Intelligence (AI) and Computational Design. The system is implemented using both textual (Python™ version 3.10.11) and visual (Grasshopper 3D™ version 1.0.0007) programming environments. The proposed workflow consists of the following four main stages: (a) acquisition of user facial images, (b) AI-based detection of facial landmarks, (c) three-dimensional reconstruction of facial features via an optimization process, and (d) generation of a personalized sunglass frame, exported as a three-dimensional model. The application demonstrates a robust performance across a diverse set of test images, consistently generating geometries that conformed closely to each user’s facial morphology. The accurate recognition of facial features enables the successful generation of customized sunglass frame designs. The system is further validated through the fabrication of a physical prototype using additive manufacturing, which confirms both the manufacturability and the fit of the final design. Overall, the results indicate that the combined use of AI-driven feature extraction and parametric Computational Design constitutes a powerful framework for the automated development of personalized wearable products. Full article

Using additive manufacturing for the design of inserts in injection moulding (IM) offers advantages in product development and customization. However, challenges related to operating temperature and mechanical resistance remain. This article presents a systematic screening methodology to evaluate the suitability of materials for specific applications. Ten commercial Material Extrusion (MEX) filaments were selected to produce test samples. Moldex3D simulation software was employed to model the IM process using two thermoplastics and to determine the temperature and pressure conditions that the printed inserts must withstand. Simulation results were critically interpreted and cross-referenced with the experimental material characterisations to evaluate material suitability. Nine of the ten MEX materials were suitable for IM with LDPE, and five with PP. Dimensional assessments revealed that six insert solutions required further post-processing for assembly, while three did not. All of the selected materials successfully survived 10 injection cycles without encountering any significant issues. The simulation results were validated by comparing temperature data from a thermal imaging camera during IM, revealing only minor deviations. The study concludes that combining targeted material characterization with CAE simulation provides an effective and low-cost strategy for selecting MEX filaments for injection moulding inserts, supporting rapid tooling applications in niche production. Full article

This work presents a combined numerical and experimental investigation into the laser machining of aluminum alloy Al 1050 H14 using a high-power Continuous Wave (CW) fiber laser. Advanced three-dimensional, coupled thermal–structural Finite Element Method (FEM) simulations are developed to model key laser–material interaction processes, including laser-induced plastic deformation, laser etching, and engraving. Cases for both static single-shot and dynamic linear scanning laser beams are investigated. The developed numerical models incorporate a Gaussian heat source and the Johnson–Cook constitutive model to capture elastoplastic, damage, and thermal effects. The simulation results, which provide detailed insights into temperature gradients, displacement fields, and stress–strain evolution, are rigorously validated against experimental data. The experiments are conducted on an integrated setup comprising a 2 kW TRUMPF CW fiber laser hosted on a 3-axis CNC milling machine, with diagnostics including thermal imaging, thermocouples, white-light interferometry, and strain gauges. The strong agreement between simulations and measurements confirms the predictive capability of the developed FEM framework. Overall, this research establishes a reliable computational approach for optimizing laser parameters, such as power, dwell time, and scanning speed, to achieve precise control in metal surface treatment and modification applications. Full article

This systematic mapping review and bibliometric analysis examines Scopus-indexed research on computer vision, image processing, and deep learning applied to wood and timber materials and products. A rule-based Scopus search (TITLE-ABS-KEY, 9 December 2025), combining wood and timber terms with imaging and computer vision terminology, followed by duplicate removal and structured exclusions, retained 1019 papers (1983–2026, early access) covering surface inspection, internal imaging, species identification, processing operations (log-yard/sawmill/panels), automation, dimensional metrology, and image-based property/structure characterization. The papers were classified into nine application categories and three methodological classes using improved rule-based classification with weighted scoring and exclusion rules. Paper output continues to accelerate, with 63.7% of papers published since 2016; Wood Surface Quality Control dominates (48.3%), followed by 3D and Internal Wood Imaging (13.6%), Wood Microstructure and Characterization (10.1%), and Wood Species and Origin Identification (10.6%). Methodologically, classical computer vision prevails (73.6%). Deep learning accounts for 26.4% of the corpus overall and 48.8% of papers from 2023–2026 (early access), while classical computer vision remains prevalent (70.1%) across most categories; the dataset totals 11,961 citations (mean: 11.74 per paper). Validation on 97 papers showed 80.41% accuracy for methodological classification and 70.1% for application categories. We quantitatively map method evolution across the nine categories, introducing a tailored taxonomy and tracking the shift from classical vision to deep learning at the category level. The remaining gaps include dimensional measurement automation, warp detection, sawing optimization, and benchmark datasets, with future directions emphasizing Vision Transformers, multi-modal sensing, edge computing, and explainable AI for certification. Full article

Robot hand-to-eye calibration is a necessary process for a robot arm to perceive and interact with its environment. Past approaches required collecting multiple images using a calibration board placed at different locations relative to the robot. When the robot or camera is displaced from its calibrated position, hand–eye calibration must be redone using the same tedious process. In this research, we developed a novel method that uses a semi-automatic process to perform hand-to-eye calibration with a stereo camera, generating a transformation matrix from the world to the camera coordinate frame from a single image. We use a robot-pointer tool attached to the robot’s end-effector to manually establish a relationship between the world and the robot coordinate frame. Then, we establish the relationship between the camera and the robot using a transformation matrix that maps points observed in the stereo image frame from two-dimensional space to the robot’s three-dimensional coordinate frame. Our analysis of the stereo calibration showed a reprojection error of 0.26 pixels. An evaluation metric was developed to test the camera-to-robot transformation matrix, and the experimental results showed median root mean square errors of less than 1 mm in the x and y directions and less than 2 mm in the z directions in the robot coordinate frame. The results show that, with this work, we contribute a hand-to-eye calibration method that uses three non-collinear points in a single stereo image to map camera-to-robot coordinate-frame transformations. Full article

Modern Earth observation combines high spatial resolution, wide swath, and dense temporal sampling, producing image grids and sequences far beyond the regime of standard vision benchmarks. Convolutional networks remain strong baselines but struggle to aggregate kilometre-scale context and long temporal dependencies without heavy tiling and downsampling, while Transformers incur quadratic costs in token count and often rely on aggressive patching or windowing. Recently proposed visual state-space models, typified by Mamba, offer linear-time sequence processing with selective recurrence and have therefore attracted rapid interest in remote sensing. This survey analyses how far that promise is realised in practice. We first review the theoretical substrates of state-space models and the role of scanning and serialization when mapping two- and three-dimensional EO data onto one-dimensional sequences. A taxonomy of scan paths and architectural hybrids is then developed, covering centre-focused and geometry-aware trajectories, CNN– and Transformer–Mamba backbones, and multimodal designs for hyperspectral, multisource fusion, segmentation, detection, restoration, and domain-specific scientific applications. Building on this evidence, we delineate the task regimes in which Mamba is empirically warranted—very long sequences, large tiles, or complex degradations—and those in which simpler operators or conventional attention remain competitive. Finally, we discuss green computing, numerical stability, and reproducibility, and outline directions for physics-informed state-space models and remote-sensing-specific foundation architectures. Overall, the survey argues that Mamba should be used as a targeted, scan-aware component in EO pipelines rather than a drop-in replacement for existing backbones, and aims to provide concrete design principles for future remote sensing research and operational practice. Full article

New-generation geostationary meteorological satellite radiometric imagers widely employ multi-column redundant long-linear-array detectors, for which the Best Detector Selection (BDS) strategy is crucial for enhancing the quality of remote sensing data. Addressing the limitation of current BDS methods that often rely on a single metric and thus fail to fully exploit the detector’s comprehensive performance, this paper proposes a detector evaluation method based on a modified Z-score. This method systematically categorizes detector metrics into three types: positive, negative, and uniformity. It introduces, for the first time, spectral response deviation (SRD) as an effective quantitative measure for the Spectral Response Function (SRF) and employs a robust normalization strategy using the Interquartile Range (IQR) instead of standard deviation, enabling multi-dimensional detector evaluation and selection. Validation using laboratory data from the FY-4C/AGRI long-wave infrared band demonstrates that, compared to traditional single-metric optimization strategies, the best detectors selected by our method show significant improvement across multiple performance indicators, markedly enhancing both data quality and overall system performance. The proposed method features low computational complexity and strong adaptability, supporting on-orbit real-time detector optimization and dynamic updates, thereby providing reliable technical support for high-quality processing of remote sensing data from geostationary meteorological satellites. Full article

This work presents an analytical homogenization model developed to predict the tensile and bending behavior of double-wall corrugated cardboard. The proposed approach replaces the complex three-dimensional geometry, composed of five paper layers, with an equivalent two-dimensional homogenized plate. Based on lamination theory and enhanced by sandwich structure theory, the model accurately captures the orthotropic behavior of the material. To achieve this objective, three configurations of double-wall corrugated cardboard were investigated: KRAFT LINER (KL), DUOSAICA (DS), and AUSTRO LINER (AL). A comprehensive experimental characterization campaign was conducted, including physical analyses (density measurement, SEM imaging, and XRD analysis) and mechanical testing (tensile tests), to determine the input parameters required for the homogenization process. The proposed model significantly reduces geometric complexity and computational cost while maintaining excellent predictive accuracy. Validation was performed by comparing the results of a 3D finite element model (ANSYS-19.2) with those obtained from the homogenized H-2D model. The differences between both approaches remained systematically below 2%, confirming the ability of the H-2D model to accurately reproduce the axial and flexural stiffnesses of double-wall corrugated cardboard. The methodology provides a reliable and efficient framework specifically dedicated to the mechanical analysis and optimization of corrugated cardboard structures used in packaging applications. Full article