Search Results (39)

Accurate prediction of automotive aerodynamic wind noise is important for cabin comfort and early-stage styling, yet conventional CFD and wind-tunnel workflows are too expensive for rapid design iteration. This paper proposes a point cloud surrogate that combines farthest-point sampling with a Transolver-derived, physics-inspired slice-attention mechanism. Here, physics-inspired denotes a representation-level inductive bias; the model does not impose governing-equation residuals, conservation constraints, or physics-based losses. Exterior meshes are converted into 10,240-point geometric inputs and assembled into a controlled dataset of 867 sedan and SUV variants generated at 120 km/h and zero yaw. On the random test split, the model obtains RMSE values of 2.30 dB(A), 2.56 dB for SPL, and 0.0068 for the dimensionless articulation index (AI), with 0.80 s single-sample inference on an RTX 4090. Repeated-seed and grouped-split analyses indicate a favorable accuracy–latency trade-off while also showing a measurable performance decrease for held-out vehicle families. A single-vehicle wind-tunnel comparison confirms strong frequency-trend correlation but reveals a mean simulation over-prediction of 2.70 dB; therefore, the current surrogate should be interpreted primarily as an emulator of the simulation labels rather than a universally unbiased predictor of measured cabin noise. Full article

Generative data augmentation has been widely explored in Wi-Fi fingerprint-based indoor localization to reduce the cost of dense radiomap construction, with many studies reporting substantial localization improvements. However, these comparisons typically rely on default or weakly optimized baseline regressors, making it difficult to determine whether reported gains reflect genuine synthesis quality or merely compensate for suboptimal baselines. In this paper, we systematically investigate under which conditions generative augmentation is actually justified. We fine-tune four widely used localization regressors—kNN, SVR, XGBoost, and DNN—using Bayesian hyperparameter optimization and establish strong non-augmented baselines across radiomaps with controlled levels of spatial sparsity, constructed via farthest-point sampling. We then train five representative generative models—VAE, GAN, DDPM, DiT, and TDPM—within a unified augmentation pipeline that includes quality filtering and pseudo-labeling, and benchmark them against these baselines. Using two publicly available datasets, we show that none of the generative models consistently outperforms a non-augmented, fine-tuned baseline regressor such as XGBoost or kNN, across a wide range of sparsity levels. We further show that these conclusions are robust to three potential confounders: various proportions of synthetic data, the choice of localization regressor (ruling out circularity with the pseudo-labeling model), and the dataset itself, since the findings on the first dataset replicate on a second, structurally different building. These findings suggest that reported augmentation benefits in prior work may partly reflect under-optimized baselines rather than genuine synthesis quality, and that generative augmentation should be treated as a conditional last resort rather than a universal improvement strategy. Full article

Accurate reconstruction of complete tree point clouds is essential for estimating ecosystem structural characteristics from LiDAR data. In urban forestry environments, however, terrestrial laser scanning (TLS) and mobile laser scanning (MLS) frequently produce incomplete observations. Occlusion caused by neighboring trees, together with interference from surrounding urban objects such as buildings and vehicles, often leads to missing regions within scanned point clouds. These defects may further affect the reliability of tree structural analysis and parameter estimation. Although recent learning-based point cloud completion methods have improved reconstruction performance, several limitations remain when they are applied to complex tree structures. Many existing networks depend on farthest point sampling (FPS) for feature extraction, which can result in the loss of fine-scale branching information. Furthermore, local feature aggregation methods based on the traditional k-nearest neighbor (KNN) strategy are highly sensitive to regions with uneven point cloud distribution, such as the canopy region where density variations are significant in tree point clouds. To alleviate these issues, this study proposes GS-TreeAttn, an attention-guided framework specifically for tree point cloud completion. This network models density and structural representation as a coupled problem and employs a structure-guided density-adaptive attention mechanism to jointly capture global structural dependencies and local geometric features. We comprehensively evaluate the proposed method using publicly available datasets and urban forestry data collected under real-world scanning conditions. Experimental results show that even in complex scenarios with severe occlusion and uneven sampling density, GS-TreeAttn generates more complete reconstruction results. This improvement is particularly evident in regions where the canopy and branches mutually occlude each other, where information loss is very common in real-world urban forestry. Full article

Long-distance water pipelines are typically instrumented by engineering convention, producing dense and partially redundant networks. Using 7239 hourly snapshots from a 122 km trunk pipeline in northeast China (72 deployed pressure sensors; 60 retained after a >$30\%$ missing-rate filter), we ask how few sensors are needed to reconstruct the full pressure field. The field has effective rank 15 at the 99% cumulative-variance level, so the cleaned network is over-sampled by roughly $4\times$. We benchmark four selection strategies against a random baseline—spatial farthest-point, PCA leverage, Q-DEIM, and a proposed hybrid that adds a soft spatial-diversity penalty to the Q-DEIM residual score—under a uniform Tikhonov-regularised gappy POD reconstruction. Q-DEIM and the hybrid both reach $R^2=0.982$ (RMSE 0.96 mH₂O) with only 15 sensors (75% reduction of the 60-sensor cleaned network); the stricter $R^2\ge 0.99$ milestone requires $K=26$ for Q-DEIM and $K=42$ for the hybrid. PCA leverage, spatial and random sampling need 19, 32, and 43 sensors for $R^2\ge 0.95$. At $K=20$, the hybrid concedes 0.003 mH₂O of RMSE for a 3.3× improvement in worst-case fill distance (5.86 km vs. 19.37 km). Regime coverage of the training library is the binding deployment constraint. Full article

Radiotherapy simulation is essential for accurately targeting tumors while preserving healthy tissue, ensuring treatment precision and safety. This study aimed to develop an integrated radiotherapy simulation system capable of automated segmentation, dose estimation, and collision detection within a virtual planning environment to enhance efficiency and reduce costs in radiotherapy treatment planning. The Point Transformer model was applied to organ point cloud data derived from CT medical imaging for automated segmentation. Farthest point sampling (FPS) was employed to downsample the data before training. To enhance the accuracy and anatomical fidelity of the AI-generated segmentation results, reconstruction and refinement algorithms, including k-d tree, outlier removal, marching cubes, and surface smoothing, were implemented. Beam penetration simulation with the ray casting algorithm was employed for correction-based dose estimation. A collision detection module was incorporated to identify potential machine–machine or machine–patient interactions. The entire workflow was executed within a Unity 3D-based virtual simulation environment. As a result, the Point Transformer model demonstrated high segmentation accuracy, achieving Dice scores of 93.86 $\pm$ 1.50% for single-organ and 91.86 $\pm$ 3.25% for multi-organ cases, surpassing the performance of PointNet++. Applying ray casting for the refined surface meshes generated through post-processing enabled accurate dose estimation with discrepancies of 3.5% (brain), 5.9% (liver), and 13.8% (lung) compared to a Pinnacle TPS. The proposed method provides a low-cost and adaptable solution that enables easy modification and further development, making it particularly suitable for widespread applications in radiotherapy research, education, and clinical workflow optimization. Full article

Point cloud data, with its rich information and high-precision geometric details, holds significant value for urban road infrastructure surveying and management. To overcome the limitations of manual classification, this study employs deep learning techniques for automated point cloud feature extraction and classification, achieving high-precision object recognition in road scenes. By integrating the Princeton ModelNet40, ShapeNet, and Sydney Urban Objects datasets, we extracted 3D spatial coordinates from the Sydney Urban Objects Dataset and organized labeled point cloud files to build a comprehensive dataset reflecting real-world road scenarios. To address noise and occlusion-induced data gaps, three augmentation strategies were implemented: (1) Farthest Point Sampling (FPS): Preserves critical features while mitigating overfitting. (2) Random Z-axis rotation, translation, and scaling: Enhances model generalization. (3) Gaussian noise injection: Improves training sample realism. The PointNet++ framework was enhanced by integrating a point-filling method into the preprocessing module. Model training and prediction were conducted using its Multi-Scale Grouping (MSG) and Single-Scale Grouping (SSG) schemes. The model achieved an average training accuracy of 86.26% (peak single-instance accuracy: 98.54%; best category accuracy: 93.15%) and a test set accuracy of 97.41% (category accuracy: 84.50%). This study demonstrates successful road scene point cloud classification, providing valuable insights for point cloud data processing and related research. Full article

Accurate plant phenotyping is crucial for gaining a deeper understanding of plant growth patterns and improving yield. However, the segmentation and measurement of 3D phenotypic data in maize remains challenging due to factors such as complex canopy structure, occlusion, and uneven point distribution. To address this, we propose a deep learning network, DUFA-Net, based on dual uncertainty-driven feature aggregation. This method employs a dual uncertainty-driven farthest point sampling (DU-FPS) strategy to mitigate errors caused by uneven point cloud density. Furthermore, for local feature encoding, we designed a Dynamic Feature Aggregation (DFA) module to model neighborhood structures and capture fine-grained geometric features, thereby effectively handling complex canopy structures. Experiments on a self-constructed maize dataset demonstrate that DUFA-Net achieves 95.82% segmentation accuracy and a mean IoU of 92.52%. Based on the segmentation results, six key phenotypic features were accurately extracted, showing high R² values ranging from 0.92 to 0.99. Further evaluation on the Syau Single Maize dataset confirms the generalization capability of the proposed method, achieving 92.52% accuracy and 91.23% mIoU, outperforming five state-of-the-art baselines, including PointNet++, PointMLP, and CurveNet. These results highlight the effectiveness and robustness of DUFA-Net for high-precision organ segmentation and phenotypic trait extraction in complex plant architectures. Full article

In surgical navigation-assisted pedicle screw fixation, cross-source pre- and intra-operative point clouds registration faces challenges like significant initial pose differences and low overlapping ratio. Classical algorithms based on feature descriptor have high computational complexity and are less robust to noise, leading to a decrease in accuracy and navigation performance. To address these problems, this paper proposes a coarse-to-fine registration framework. In the coarse registration stage, a Point Matching algorithm based on Curvature Feature Learning (CFL-PM) is proposed. Through CFL-PM and Farthest Point Sampling (FPS), the coarse registration of overlapping regions between the two point clouds is achieved. In the fine registration stage, the Iterative Closest Point (ICP) is used for further optimization. The proposed method effectively addresses the challenges of noise, initial pose and low overlapping ratio. In noise-free point cloud registration experiments, the average rotation and translation errors reached 0.34° and 0.27 mm. Under noisy conditions, the average rotation error of the coarse registration is 7.28°, and the average translation error is 9.08 mm. Experiments on pre- and intra-operative point cloud datasets demonstrate the proposed algorithm outperforms the compared algorithms in registration accuracy, speed, and robustness. Therefore, the proposed method can achieve the precise alignment of the surgical navigation-assisted pedicle screw fixation. Full article

Fractal theory provides a powerful tool for quantifying complex geometric patterns such as concrete cracks. The box-counting method is widely employed for fractal dimension (FD) calculation due to its intuitive principles and compatibility with image data. However, two critical limitations persist in existing studies: (1) the selection of scale parameters (including minimum measurement scale and cutoff scale) lacks systematization and exhibits significant arbitrariness; (2) insufficient attention to the sensitivity of counting origins compromises the stability and comparability of FDs, severely limiting reliable engineering application. To address these limitations, this study first employs classical fractal images and crack samples to systematically analyze the impact of four minimum measurement scales (2, $\sqrt{2}$, 3, $\sqrt{3}$) and three cutoff scale coefficients (cutoff-to-minimum image side ratios: 1, 1/2, 1/3) on computational accuracy. Subsequently, the farthest point sampling (FPS) method is adopted to select counting origins, comparing two optimization strategies—Count-FD-Mean (mean of fits from multiple origins) and Count-Min-FD (fit using minimal box counts across scales). Finally, the optimized approach is validated through static loading tests on concrete beams. Key findings demonstrate that: the optimal scale combination (minimum scale: $\sqrt{2}$; cutoff coefficient: 1) yields a mere 0.5% average error from theoretical FDs; the Count-Min-FD strategy delivers the highest stability and closest alignment with theoretical values; FDs of beam cracks increase continuously with loading, exhibiting an exponential correlation with midspan deflection that effectively captures crack evolution; uncalibrated scale parameters and counting strategies may induce >40% errors in inferred mechanical parameters; results stabilize with 40–45 counting origins across three tested fractal patterns. This work advances standardization in fractal analysis, enhances reliability in concrete crack assessment, and provides critical support for the practical application of fractal theory in structural health monitoring and damage evaluation. Full article

Due to the large volume of raw data in 3D point clouds, downsampling techniques are crucial for reducing computational load and memory usage to improve the training of 3D point cloud models. This paper plans to conduct research using the ModelNet40 dataset. Our proposed method is based on the PointNext architecture, an improved version of PointNet++ that significantly enhances performance through optimized training strategies and adjusted receptive fields. During the model training process, we employ the farthest point sampling method for downsampling. Specifically, we use an improved attention-based point cloud edge sampling (APES) method for downsampling, where we compute the density of each point and set the size of the neighbor K value to effectively retain feature points during downsampling. Our improved method captures edge points more effectively than the original APES method. By adjusting the architecture, our method, combined with the farthest point sampling method, not only reduced the average training time by nearly 15% compared to PointNext-s, but also improved accuracy from 93.11% to 93.57%. Full article

This paper addresses the critical challenge of point cloud down-sampling for digital twin creation, where reducing data volume while preserving geometric fidelity remains an ongoing research problem. We propose a novel down-sampling approach that combines Low-Discrepancy Sequences (LDS) with Hilbert curve ordering to create a method that preserves both global distribution characteristics and local geometric features. Unlike traditional methods that impose uniform density or rely on computationally intensive feature detection, our LDS-Hilbert approach leverages the complementary mathematical properties of Low-Discrepancy Sequences and space-filling curves to achieve balanced sampling that respects the original density distribution while ensuring comprehensive coverage. Through four comprehensive experiments covering parametric surface fitting, mesh reconstruction from basic closed geometries, complex CAD models, and real-world laser scans, we demonstrate that LDS-Hilbert consistently outperforms established methods, including Simple Random Sampling (SRS), Farthest Point Sampling (FPS), and Voxel Grid Filtering (Voxel). Results show parameter recovery improvements often exceeding 50% for parametric models compared to the FPS and Voxel methods, nearly 50% better shape preservation as measured by the Point-to-Mesh Distance (than FPS) and up to 160% as measured by the Viewpoint Feature Histogram Distance (than SRS) on complex real-world scans. The method achieves these improvements without requiring feature-specific calculations, extensive pre-processing, or task-specific training data, making it a practical advance for enhancing digital twin fidelity across diverse application domains. Full article

The accurate point cloud completion of individual tree crowns is critical for quantifying crown complexity and advancing precision forestry, yet it remains challenging in dense plantations due to canopy occlusion and LiDAR limitations. In this study, we extended the scope of conventional point cloud completion techniques to artificial planted forests by introducing a novel approach called Multi−feature Fusion Completion of Populus (MFCPopulus). Specifically designed for Populus Tomentosa plantations with uniform spacing, this method utilized a dataset of 1050 manually segmented trees with expert−validated trunk−canopy separation. Key innovations include the following: (1) a hierarchical adversarial framework that integrates multi−scale feature extraction (via Farthest Point Sampling at varying rates) and biologically informed normalization to address trunk−canopy density disparities; (2) a structural characteristics split−collocation (SCS−SCC) strategy that prioritizes crown reconstruction through adaptive sampling ratios, achieving a 94.5% canopy coverage in outputs; (3) a cross−layer feature integration enabling the simultaneous recovery of global contours and a fine−grained branch topology. Compared to state−of−the−art methods, MFCPopulus reduced the Chamfer distance variance by 23% and structural complexity discrepancies (ΔD_b) by 33% (mean, 0.12), while preserving species−specific morphological patterns. Octree analysis demonstrated an 89−94% spatial alignment with ground truth across height ratios (HR = 1.25−5.0). Although initially developed for artificial planted forests, the framework generalizes well to diverse species, accurately reconstructing 3D crown structures for both broadleaf (Fagus sylvatica, Acer campestre) and coniferous species (Pinus sylvestris) across public datasets, providing a precise and generalizable solution for cross−species trees’ phenotypic studies. Full article

Point cloud completion reconstructs incomplete, sparse inputs into complete 3D shapes. However, in the current 3D completion task, it is difficult to effectively extract the local details of an incomplete one, resulting in poor restoration of local details and low accuracy of the completed point clouds. To address this problem, this paper proposes a multi-scale point cloud completion method based on local neighborhood dynamic fusion (LNDF: adaptive aggregation of multi-scale local features through dynamic range and weight adjustment). Firstly, the farthest point sampling (FPS) strategy is applied to the original incomplete and defective point clouds for down-sampling to obtain three types of point clouds at different scales. When extracting features from point clouds of different scales, the local neighborhood aggregation of key points is dynamically adjusted, and the Transformer architecture is integrated to further enhance the correlation of local feature extraction information. Secondly, by combining the method of generating point clouds layer by layer in a pyramid-like manner, the local details of the point clouds are gradually enriched from coarse to fine to achieve point cloud completion. Finally, when designing the decoder, inspired by the concept of generative adversarial networks (GANs), an attention discriminator designed in series with a feature extraction layer and an attention layer is added to further optimize the completion performance of the network. Experimental results show that LNDM-Net reduces the average Chamfer Distance (CD) by 5.78% on PCN and 4.54% on ShapeNet compared to SOTA. The visualization of completion results demonstrates the superior performance of our method in both point cloud completion accuracy and local detail preservation. When handling diverse samples and incomplete point clouds in real-world 3D scenarios from the KITTI dataset, the approach exhibits enhanced generalization capability and completion fidelity. Full article

Based on the single nucleotide polymorphism (SNP) markers developed by whole genome resequencing (WGRS), the relationship and population genetic structure of 53 common apricot (P. armeniaca) varieties were analyzed to provide a theoretical basis for revealing the phylogenetic relationship and classification of the common apricot. WGRS was performed on 53 common apricot varieties, and high-quality SNP sites were obtained after alignment with the “Yinxiangbai” apricot genome as a reference. Phylogenetic analysis, G matrix analysis, principal component analysis, and population structure analysis were performed using Genome-wide Complex Trait Analysis (GCTA), FastTree, Admixture, and other software. The average comparison ratio between the sequencing results and the reference genome was 97.66%. After strict screening, 88,332,238 high-quality SNP sites were finally obtained. Based on the statistical SNP variation type, it was found that LNLJX had the largest number of variations (3,951,322) and the lowest base transition/base transversion ratio (ts/tv = 1.77), indicating that its gene exchange events occurred less frequently. Based on the SNP point estimation of the relationship and genetic distance between samples, the relationship between species was 1.41–0.01, among which PLDJX and BK1 had the closest relationship of 1.41, and YZH and LGWSX had the farthest relationship of 0.01. The genetic distance between species was 0.00367–0.264344, the genetic distance between HMX and JM was the closest, and the genetic distance between WYX and YX was the farthest, which was the largest. Phylogenetic tree, PCA, and genetic structure analysis results all divided 53 common apricot varieties into four groups, and the classification results were consistent. The SNP markers mined using WGRS technology are useful not only to analyze the variation of common apricots, but also to effectively identify their kinship and genetic structure, which plays a critical role in the classification and utilization of common apricot germplasm resources. Full article

The three-dimensional model of geographic elements serves as the primary medium for digital visualization. However, the original point cloud model is often vast and includes considerable redundant data, resulting in inefficiencies during the three-dimensional modeling process. To address this issue, this paper proposes a point cloud reduction algorithm that leverages domain normal vectors and fuzzy-c means (FCM) clustering for feature extraction. The algorithm first extracts the edge points of the model and then utilizes domain normal vectors to extract the overall feature points of the model. Next, utilizing point cloud curvature, coordinate information, and geometric attributes, the algorithm applies the FCM clustering method to isolate local feature points. Non-feature points are then sampled using an enhanced farthest point sampling technique. Finally, the algorithm integrates edge points, feature points, and non-feature points to generate simplified point cloud data. This paper compares the proposed algorithm with traditional methods, including the uniform grid method, random sampling method, and curvature sampling method, and evaluates the simplified point cloud in terms of reduction level and reconstruction time. This approach effectively preserves critical feature information from the majority of point cloud data, thereby addressing the complexities inherent in original point cloud models. Full article