Search Results (1,629)

Comminution is the most energy-intensive stage of mineral processing, and microwave-assisted comminution (MAC) can reduce grinding energy by selectively heating microwave-absorbing minerals within transparent gangue, generating thermal microcracks that improve liberation. MAC performance, however, depends on the ore mineralogy and surface, which fixed-parameter operation cannot accommodate. An integrated mechatronic “intelligent crusher” is presented unifying actuation (microwave source, feed system, adjustable crusher geometry), sensing (thermal infrared and hyperspectral imaging, HSI), and control (offline reinforcement learning). HSI-derived mineralogical features and infrared thermal features form the state of a behavior-regularized actor–critic (BRAC) controller trained offline on logged operating data to adjust the power, exposure, feed rate, and crusher setting. A two-dimensional coupled electromagnetic–thermal–mechanical finite-element study underpins the process model. It is executed with temperature-independent dielectric properties in a single staggered coupling pass, and so calibrates the damage law qualitatively rather than predicting stress quantitatively. It reproduces cracking thresholds from the literature and shows that thermal gradients decay with exposure time as (1 + t/$\tau$)^−0.57, so that damage at a constant dose falls from 0.63 to 0.02 as exposure lengthens from 0.25 to 16 s. On this basis, the phenomenological damage law, which had been exposure-insensitive, is corrected. On the FEA-calibrated simulator, the BRAC policy reduces the mean size targeting error by 62% (1.43 to 0.54 mm) and the total specific energy by 4.2% (6.50 to 6.22 kWh/t), averaged over five training seeds, relative to fixed-parameter operation, outperforms rule-based and behavior-cloning baselines, and generalizes to a simulated ore batch excluded from the training. The framework establishes a validated control architecture for adaptive MAC ahead of three-dimensional model extension and experimental deployment. Full article

3D Gaussian Splatting (3DGS) enables real-time photorealistic scene reconstruction, yet its segmentation tasks suffer from two critical flaws: poor 3D consistency (e.g., blurred instance boundaries and unstable cross-view semantic association) and insufficient structural awareness near ambiguous object boundaries. To address these issues, this paper proposes a semantic-guided 3D Gaussian segmentation framework. We first process multi-view RGB images and semantic masks to generate 2D semantic codes and initial Gaussian parameters. We then build a Gaussian-level joint representation by combining flow-aligned appearance cues, CLIP-derived semantic descriptors, and explicit Gaussian geometry. Semantic-guided adaptive decomposition identifies boundary-sensitive Gaussians through cross-view boundary statistics and semantic uncertainty, while multi-view voting further refines instance boundaries. Finally, 3D global optimization unifies instance association and labeling. Experiments on ScanNet and SPIn-NeRF show improved results under the reported evaluation protocol and the stated subset definitions. In particular, on the ScanNet benchmark, compared with the evaluated 3DGS baseline InstanceGaussian, SG-AGS improves mAcc@0.25 by 16.2 points for category-agnostic 3D instance segmentation and by 25.6 points for text-query-based open-vocabulary labeling of segmented 3D instances. These results support the usefulness of SG-AGS on the two evaluated benchmarks and suggest potential value for object-level scene querying and digital-twin inspection, while broader generalization to external captures and other scene domains still requires further validation. Full article

Hydrological regulation and local engineering can modify the planform dynamics of wandering rivers, but separating their effects is difficult where pre-regulation morphology and synchronous hydraulic observations are unavailable. This study integrates discharge and water-level records (2000–2023), SRTM-derived surface topography, and nine December satellite-image composites (2015–2023) for the urban Lhasa River reach between Gates 1 and 5. Within the available 24-year discharge record, Pettitt tests identify a mid-2000s distributional shift at the three stations; sensitivity tests place the detected change in 2005–2006. Because the pre-change segment contains only six years and no precipitation-based natural-flow control was available, the result is interpreted as temporally consistent with, but not proof of an effect of, Zhikong Hydropower Station commissioning. The 2013 Pangduo project is treated separately as an engineering timeline marker because no distinct annual mean breakpoint was detected at that date. Mapped water area increased by 45.39% between 2015 and 2023, whereas exposed bar–island and riparian-land classes decreased by 36.29% and 61.13%, respectively. These values are reported as a water-surface expansion signal consistent with gate impoundment and/or acquisition-stage differences, not as confirmed sedimentary or geomorphic conversion. The comparison also contains sensor-dependent positional uncertainty: approximately ±15 m for the 2015 and 2018 Landsat-8 inputs and ±5 m for Sentinel-2 inputs. Centerline length and sinuosity varied by less than 1%. The nearest-neighbor bankline metric declined by 9.53%, whereas the node-mean channel-regime-center metric declined by 81.99%; the latter is retained only as an exploratory indicator because it is sensitive to centerline-node sampling. Overall, the observations are consistent with reduced lateral activity in an engineered reach, but the available evidence does not isolate reservoir regulation, gate operation, bank protection, climate, land use, or sediment-supply effects. Stronger causal inference requires a pre-regulation morphological baseline, synchronous discharge and water-level observations at image acquisition, fixed-interval centerline resampling, sediment data, and surveyed hydraulic geometry. Full article

This paper presents the RaDICAL sensing framework, a monostatic passive radar concept that combines a Sparse Uniform Circular Array (SUCA), deterministic multifrequency dither, and dictionary-based waveform recognition for target detection and classification. Rather than forming conventional spatial images or relying primarily on Doppler processing, RaDICAL encodes target geometry directly into a composite receiver waveform and performs hypothesis testing by matching measured signals to a library of predicted responses. This study develops the SUCA-based signal model for point and extended targets and formulates recognition as a waveform-domain matching problem against the representation dictionary using normalized complex correlation and QR-domain processing. A reproducible MATLAB-based study evaluates waveform separability, probability of detection versus dictionary SNR, physical power balance, receiver operating characteristic (ROC) behavior, and detection performance versus illuminator EIRP. The results show that deterministic dither produces distinctive composite waveforms with strong hypothesis separability. The ROC simulations characterize binary detection of the structured waveform, while the recognition simulations quantify discrimination among competing physical dictionary hypotheses. Because the proposed receiver coherently processes multiple spatial and temporal waveform samples, no direct single-sample SNR advantage over a classical matched-filter detector is claimed. These results support the feasibility of waveform-domain passive sensing using deterministic spatial–frequency encoding and dictionary-based recognition. Unlike conventional representation-learning methods that derive embeddings from image or feature datasets, the proposed waveform representations are generated deterministically from physical electromagnetic models. Full article

Forward-looking sonar (FLS) is robust to low illumination and turbidity, whereas its range–azimuth projection collapses elevation information, making single-view three-dimensional reconstruction inherently ambiguous. This study investigates how the geometric composition of synthetic training data influences reconstruction accuracy for cylindrical and spherical objects. Paired acoustic-view and front-view depth images were generated in Blender, and the synthetic acoustic-view images were translated toward the real-sonar image domain using CycleGAN. The CycleGAN-translated images were used to train the previously proposed acoustic-view-to-front-view network (A2FNet) to estimate pseudo-front-view inverse-depth maps. In addition to evaluating shape-specific training data, this study analyzes geometry-dependent large-error cases and uses the identified failure patterns to revise the synthetic cylinder training-scene configuration. Four models were trained using cylinder-only, sphere-only, mixed cylinder–sphere, or previous-study data, with 4000 training images per model. A total of 500 test images were prepared for each of four simulation-derived environments. Across the four environments, the best geometry-specific or mixed models reduced the mean scaled symmetric squared Chamfer distance (CD) from 77.3280–90.8582 for the previous-study model to 1.8341–7.7329. Large-error analysis revealed recurring incomplete target observations shared by the cylinder-specific and mixed models. Retraining with the revised cylinder dataset further reduced the mean CD to 1.0336–4.8036 and the maximum CD to 5.3346–11.9897 across the six evaluated model–environment combinations. These results indicate that reconstruction accuracy within the evaluated simulation-based pipeline is strongly influenced by the geometric composition and scene design of the synthetic training data. Systematic analysis of reconstruction failures can therefore provide useful guidance for revising synthetic training scenes and reducing large reconstruction errors. Full article

Underwater optical imaging plays a crucial role in marine exploration, ecological monitoring, underwater inspection, autonomous navigation, and scientific research, but it is heavily restricted by wavelength-dependent attenuation, scattering, refraction, and low visibility. This comprehensive review examines underwater optical imaging from a system-design perspective, integrating the physical principles of underwater light propagation with imaging modalities, illumination strategies, sensing technologies, computational methods, and system-level implementation considerations. A structured literature search was conducted using Scopus and Google Scholar, yielding 441 initially identified studies, of which 262 were retained after screening according to the defined inclusion and exclusion criteria, and seven additional recent studies were incorporated during revision, resulting in a final set of 269 publications. The reviewed literature is organized according to imaging modality and system architecture, including conventional RGB imaging, laser-based imaging, polarization imaging, multispectral and hyperspectral imaging, computational imaging, and AI-assisted approaches. Emphasis is placed on the relationship between environmental conditions, mission requirements, imaging range, illumination geometry, calibration, processing location, computational demand, and platform constraints. The analysis shows that no single imaging modality provides optimal performance across all underwater environments. Instead, effective system design requires matching the sensing and illumination strategy to water conditions, target distance, required image quality, available power, computational resources, and real-time requirements. The review therefore proposes a mission-oriented framework for understanding the principal trade-offs involved in the design and selection of underwater optical imaging systems and identifies adaptive acquisition, multimodal sensing, physics-informed AI, and optical–computational co-design as important directions for future development. Full article

Procedural textures trade image storage for runtime shader evaluation, whereas baked bitmaps replace repeated procedural computation with stored image data. Controlled path-tracing comparisons that isolate this representation choice remain limited. This study compares four procedural Base Color graphs with their ${2048}^2$ baked bitmap counterparts in Blender Cycles 5.2 on a GIGABYTE NVIDIA GTX 1050 Ti (GIGA-BYTE Technology Co., Ltd., New Taipei City, Taiwan) while holding geometry, lighting, camera, sampling and BSDF parameters constant. Three texture graphs—checker, noise and multifractal—were examined in a predefined ordering by computational demand, while ambient occlusion was analyzed separately as a ray-traced effect. The primary experiment comprised 480 renders, of which 360 were analyzed using block means as the inferential unit. Block-level models showed that the procedural checker was 0.65% faster than its baked counterpart, whereas the bitmap representation was 1.73% faster for noise and 3.53% faster for multifractal; all four primary representation effects had 95% confidence intervals excluding zero after Holm adjustment. Baking ambient occlusion reduced render time by 67.74% under the tested configuration. The bitmap branch increased Cycles-reported internal peak memory by exactly 16 MiB at ${2048}^2$, while resolution tests produced increments of 1, 4 and 64 MiB at ${512}^2$, ${1024}^2$ and ${4096}^2$, respectively. Because the ${2048}^2$ baked multifractal showed lower visual agreement with its procedural source, a targeted ${4096}^2$ follow-up comprising a further 120 renders, of which 90 were analyzed, was conducted. Increasing the multifractal bake resolution improved SSIM from 0.886 to 0.936, increased the image-texture footprint from 16 to 64 MiB, and reduced the bitmap render-time advantage from 3.53% to 1.73%. Across the three primary texture graphs, the bitmap advantage increased along the predefined ordering by computational demand; however, computational demand was not isolated from differences in algorithm, coordinate system, and spatial structure. The results therefore demonstrate configuration-specific trade-offs among render time, image-memory footprint, and representation fidelity rather than a universal preference for either representation. Full article

In practical beef cattle monitoring scenarios, automated body condition scoring (BCS) based on single-frame visual information can be affected by posture variation, occlusion, and image quality changes. This study proposed an episode-level BCS assessment framework using synchronized RGB-D videos, integrating multi-frame visual information, attention-based multiple instance learning (Attention-MIL), and a dorsal geometry-based quality control mechanism to improve reliability under complex acquisition conditions. The publicly available Raramuri Criollo dataset was reorganized into a dataset containing 3815 RGB-D frame pairs, 52 animal identity groups, and 151 episodes, and evaluated using animal-wise three-fold cross-validation. Predictions were generated only when an episode contained at least three quality-filtered frames and reliable angle estimation. With an average coverage of 0.7345 after quality gating, the final model achieved 61.66% accuracy, a Macro-F1 score of 0.5680, and a quadratic weighted kappa (QWK) of 0.6339 for five-class BCS classification. Additionally, 89.71% of predictions were within one BCS category of the reference scores. Three-class management classification achieved a Macro-F1 score of 0.8059. Ablation experiments showed complementary contributions from RGB and depth information, and the combined angle and quality-statistic features improved several metrics in the original ablation. Additional matched experiments did not establish an independent angle benefit but supported multi-frame aggregation and within-Episode prediction consistency. The proposed RGB-D multi-frame fusion approach demonstrates potential for automated beef cattle body condition monitoring under controlled acquisition conditions. Full article

Digital twin technologies provide new opportunities for visualizing, analyzing, and managing complex physical systems; however, effective methods are needed to transform digital twin data into immersive environments that support human interaction and decision-making. This study presents the development of a Digital Twin Immersive Environment (DTIE) prototype for facility visualization and analysis at the U.S. Army Corps of Engineers Engineering Research and Development Center (ERDC) in Vicksburg, Mississippi. A photogrammetry-based workflow was used to capture the facility interior and generate a dense three-dimensional point cloud, while a separate Gaussian splat reconstruction was generated from the same image dataset. The Gaussian splat representation was spatially aligned with Revit-derived CAD geometry to provide complementary geometric and photorealistic representations within the immersive environment. A virtual reality (VR) environment was developed with a user interface (UI) supporting ray-cast interaction, speech-to-text data entry, multi-user collaboration, switching between CAD and Gaussian splat representations, in-scene web browsing, and occupant movement simulation. The resulting DTIE prototype demonstrates the technical integration of complementary CAD and Gaussian splat representations with immersive interaction capabilities for facility digital twins. The environment operated at approximately 71.8 FPS under the reported Meta Quest 2 configuration. This work establishes a prototype framework for future investigation of collaborative visualization, scenario analysis, and decision-support applications for complex infrastructure systems. Full article

Accurate measurement of steel-strip anchor holes is crucial for intelligent roadway support drilling in underground coal mines, where severe illumination degradation, airborne dust scattering, and weak structural textures often lead to unreliable visual perception and drilling collision risks. To address these challenges, a geometry-guided monocular vision framework is proposed for anchor-hole segmentation and three-dimensional localization under low-visibility conditions. First, a physics-inspired image enhancement module combining atmospheric scattering correction and adaptive illumination compensation is developed to improve image quality. Second, an ellipse-constrained geometry-aware U-Net (EGU-Net) is proposed by embedding elliptical shape priors into feature learning and boundary refinement for robust anchor-hole segmentation. Third, an ellipse fitting-based monocular geometric reconstruction model is established to recover the three-dimensional coordinates of anchor-hole centers from calibrated imaging parameters and steel-strip planar constraints. Moreover, an analytical uncertainty propagation model is derived to quantify the influence of segmentation errors and calibration uncertainty on localization accuracy. Experimental results on a self-built underground support drilling dataset demonstrate that the proposed method achieves an IoU of 94.33% and a localization MAE of 3.27 mm, outperforming existing approaches and exhibiting strong robustness under low-visibility environments. The proposed framework provides an effective visual measurement solution for intelligent and safe roadway support drilling. Full article

Introduction: Objective, image-based comparisons of anterior capsule visualization across conventional microscopy and digital heads-up visualization systems remain limited. This exploratory retrospective study quantitatively compared capsule contrast and capsulorhexis geometry among conventional microscopy and two successive generations of a digital heads-up system, both of which apply yellow digital filtering. Methods: This retrospective, non-randomized observational study included 67 eyes of 60 patients undergoing phacoemulsification using a conventional microscope (21 eyes) or NGENUITY with yellow filtering [version 1.4 (N1.4; 25 eyes) or 1.5 (N1.5; 21 eyes)], with visualization modality determined by clinical availability at the time of surgery. Color contrast ratio (CCR), ovality index, and concentricity were quantified from masked video analysis using linear mixed-effects models (patient as random intercept) with Holm–Bonferroni-adjusted comparisons. Results: Mean CCR was significantly higher in both NGENUITY groups than the Microscope group (N1.4: 1.55 ± 0.55; N1.5: 1.70 ± 0.29; Microscope: 1.14 ± 0.09; both p ≤ 0.008), with no difference between versions. The ovality index was significantly lower in the N1.5 group than the Microscope group (2.90 ± 1.78% vs. 5.08 ± 3.38%; p = 0.022), though this difference did not reach significance in the per-patient sensitivity analysis (p = 0.071); no significant difference was observed between the Microscope and N1.4 groups. Concentricity did not differ significantly among groups. Conclusions: Anterior capsule contrast, as captured in recorded surgical video, was objectively higher with NGENUITY than with conventional microscopy, whereas a modest improvement in capsulorhexis circularity was observed only with version 1.5. These exploratory, hypothesis-generating findings are intended to motivate, rather than substitute for, a prospective, randomized, within-system study; they provide a standardized measurement framework showing that contrast parameters differ across visualization modalities, while geometric advantages across system generations warrant confirmation in such future studies. Full article

Precision agriculture increasingly relies on UAV-based multispectral sensors, but cross-platform data fusion remains limited by differences in spectral response, image geometry, and structural reconstruction accuracy. This challenge is particularly important in dense crop canopies, where vegetation indices and canopy height estimates may vary substantially between sensor platforms. This study compared the multi-lens DJI Phantom 4 Multispectral (P4M) and the dual-lens Sentera Double 4K in a maize hybrid trial (DKC 4596) in Debrecen, Hungary, using aerial surveys conducted at 35.5 m to assess DSM, DTM, NDVI, and NDRE outputs across 15 synchronized sampling plots processed in WebODM and RStudio. Agreement between platforms was evaluated using Bland–Altman analysis and Pearson correlation. The results showed a substantial systematic bias for NDVI, with Sentera producing 36% higher mean values than P4M and showing almost no linear association between platforms (r = 0.07). In contrast, NDRE demonstrated stronger and more manageable cross-platform agreement, with a significant positive correlation (r = 0.66, p = 0.007). Dense canopy closure reduced the reliability of absolute canopy height estimation for both platforms, but the P4M performed particularly poorly because of lens parallax and lower effective pixel density, while the higher-resolution Sentera imagery provided a more representative structural trend. Overall, NDRE is more suitable than NDVI for cross-platform vegetation index fusion between P4M and Sentera, while reliable structural assessment in tall, dense maize requires higher-resolution imagery, rigorous ground control, or complementary technologies such as LiDAR. Full article

Not all data derived from structural analysis or experimental testing can be effectively encoded using Industry Foundation Classes (IFC) properties or attributes, even if custom-made. In many cases, visual outputs—such as images—represent the only practical means of conveying complex analytical results. However, the current literature provides limited guidance on systematically integrating images within IFC models, with practitioners often relying on external formats or auxiliary technologies. This study investigates how the IFC standard supports the visualization of textures applied to geometric entities, focusing on the relevant classes and their interrelationships. Building on this theoretical framework, a prototype is developed within the Bonsai environment, leveraging Blender 4.5 and its add-on architecture. The prototype enables the creation of planar geometries in 3D space and the application of image-based textures, enabling visual information to be associated with IFC geometry through standard texture-mapping entities and an external image reference. A case study involving the visualization of structural testing results demonstrates the approach. By keeping the visual output registered to the model geometry, the approach has the potential to enhance analytical interpretation, communication among stakeholders, and data visualization within the built environment domain. Full article

This study proposes a local six-DoF pose-refinement method for small-diameter, pen-shaped instruments using geometric inconsistencies in unwrapped images of a paper-based cylindrical marker with an outer diameter of 8.3 mm after attachment. Full-circumference bands refine the cylinder-center position and longitudinal-axis orientation (five DoFs), while dot-edge image coordinates refine roll about the cylinder axis. The fixed 5+1-DoFscheme was evaluated in Geometry and Detector modes using 240 reference poses excluded from identification and tuning and 16 combined initial-error conditions per pose (3840 trajectories per mode). In the independent CG local-refinement evaluation, 17.60% of Detector-mode trajectories failed under the CG evaluation criteria; among 3164 algorithmically completed trajectories, 20 (0.63%) ended with greater normalized pose error than initially. In a separate end-to-end CG evaluation of 240 images, the current coarse initializer produced 157 coarse poses, 111 correct IDs, and 81 P0 completions; only 1 of 240 outputs also satisfied translation and rotation error thresholds of 0.10 mm and 0.10°, respectively. Thus, local-refinement and front-end/global capture performance must be evaluated separately. In K-transfer tests, matched rendering and processing intrinsic matrices showed no substantial deterioration, whereas intentional endpoint mismatch caused correction failure. For previously acquired real images, post-hoc nominal-setting analysis showed reduced XYZ RMSE after refinement, but did not constitute absolute-accuracy validation against an independent external reference. Full article

Accurate underwater fish perception is a prerequisite for intelligent aquaculture, providing essential support for automated growth monitoring and precision feeding. Nevertheless, underwater imaging is severely affected by light scattering and cluttered backgrounds, making reliable fish perception and image-based non-contact fish size estimation particularly challenging. To overcome these limitations, this paper presents a geometry-aware underwater fish perception framework based on YOLOv5 for simultaneous object detection, pose estimation, and fish body size measurement. The proposed framework incorporates a Fish Geometry Coordinate Attention (FGCA) module into the backbone to enhance geometric feature representation by jointly modeling horizontal and vertical spatial dependencies, thereby improving the discrimination of elongated fish structures. In addition, a Dynamic Edge Relation Modeling (DERM) module is integrated into the neck to explicitly capture edge cues and structural relationships among key body regions, leading to more robust feature aggregation and more accurate key point localization under complex underwater conditions. Based on the predicted key points, the length and width of the fish can be estimated in a non-contact manner. Experimental results on a public underwater fish dataset demonstrate that the proposed framework achieves a Box mAP_0.5 of 93.07% and a Pose mAP_0.5 of 93.10%, consistently outperforming several representative YOLO-based detectors. Moreover, the proposed method maintains low body size estimation errors across different fish size categories. These results indicate that the proposed framework provides an accurate and practical solution for integrated underwater fish perception and image-based non-contact fish size estimation, offering significant potential for intelligent aquaculture and automated fish growth monitoring. Full article