Search

The underground pipeline is a critical component of urban water supply and drainage infrastructure. However, the absence of accurate pipe information frequently leads to construction delays and cost overruns, adversely impacting urban management and economic development. To address these challenges, the digital management of underground pipelines has become essential. Despite its importance, research on the structural analysis and reconstruction of underground pipelines remains limited, primarily due to the complexity of underground environments and the technical constraints of LiDAR technology. This study proposes a framework for reconstructing underground pipelines based on unstructured point cloud data, aiming to accurately identify and reconstruct pipe structures from complex scenes. The Random Sample Consensus (RANSAC) algorithm, enhanced with parameter-adaptive adjustments and subset-independent fitting strategies, is employed to fit centerline segments from the set of center points. These segments were used to reconstruct topological connections, and a Building Information Model (BIM) of the underground pipeline was generated based on the structural analysis. Experiments on actual underground scenes evaluated the method using recall rate, radius error, and deviation between point clouds and models. Results showed an 88.8% recall rate, an average relative radius error below 3%, and a deviation of 3.79 cm, demonstrating the framework's accuracy. This research provides crucial support for pipeline management and planning in smart city development. [ABSTRACT FROM AUTHOR]

In this paper, an exhaustive review and comprehensive analysis of recent and former deep learning methods in 3D semantic segmentation (3DSS) is presented. In the related literature, the taxonomy scheme used for the classification of 3DSS deep learning methods is ambiguous. Based on the taxonomy schemes of nine existing review papers, a new taxonomy scheme for 3DSS deep learning methods is proposed, aiming to standardize it and improve the comparability and clarity across related studies. Furthermore, an extensive overview of the available 3DSS indoor and outdoor datasets is provided along with their links. The core part of this review is the detailed presentation of recent and former 3DSS deep learning methods and their classification using the proposed taxonomy scheme along with their GitHub repositories. Additionally, a brief but informative analysis of the evaluation metrics and loss functions used in 3DSS is included. Finally, a fruitful discussion of the examined 3DSS methods and datasets is presented to foster new research directions and applications in the field of 3DSS. In addition to this review, a GitHub repository is provided, including an initial classification of over 400 3DSS methods, using the proposed taxonomy scheme. [ABSTRACT FROM AUTHOR]

Detecting the reflection symmetry plane of an object represented by a 3D point cloud is a fundamental problem in 3D computer vision and geometry processing due to its various applications, such as compression, object detection, robotic grasping, 3D surface reconstruction, etc. Several approaches exist to solve this problem for clean 3D point clouds. However, it is a challenging problem to solve in the presence of outliers and missing parts. The existing methods try to overcome this challenge primarily by voting-based techniques but do not work efficiently. In this work, we proposed a statistical estimator-based approach for the plane of reflection symmetry that is robust to outliers and missing parts. We pose the problem of finding the optimal estimator for the reflection symmetry as an optimization problem on a 2-sphere that quickly converges to the global solution for an approximate initialization. We further adapt the heat kernel signature for symmetry invariant matching of mirror symmetric points. This approach helps us to decouple the chicken-and-egg problem of finding the optimal symmetry plane and correspondences between the reflective symmetric points. The proposed approach achieves comparable mean ground-truth error and 4.5% increment in the F-score as compared to the state-of-the-art approaches on the benchmark dataset. [ABSTRACT FROM AUTHOR]

The convexity of a set can be generalized to the two weaker notions of positive reach and r-convexity; both describe the regularity of a set's boundary. For any compact subset of R d , we provide methods for computing upper bounds on these quantities from point cloud data. The bounds converge to the respective quantities as the sampling scale of the point cloud decreases, and the rate of convergence for the bound on the reach is given under a weak regularity condition. We also introduce the β -reach, a generalization of the reach that excludes small-scale features of size less than a parameter β ∈ [ 0 , ∞) . Numerical studies suggest how the β -reach can be used in high-dimension to infer the reach and other geometric properties of smooth submanifolds. [ABSTRACT FROM AUTHOR]

3D topographic point cloud change estimation produces fundamental inputs for understanding Earth surface process dynamics. In general, change estimation aims at detecting the largest possible number of points with significance (i.e., difference > uncertainty) and quantifying multiple types of topographic changes. However, several complex factors, including the inhomogeneous nature of point cloud data, the high uncertainty in positional changes, and the different types of quantifying difference, pose challenges for the reliable detection and quantification of 3D topographic changes. To address these limitations, the paper proposes a graph comparison-based method to estimate 3D topographic change from point clouds. First, a graph with both location and orientation representation is designed to aggregate local neighbors of topographic point clouds against the disordered and unstructured data nature. Second, the corresponding graphs between two topographic point clouds are identified and compared to quantify the differences and associated uncertainties in both location and orientation features. Particularly, the proposed method unites the significant changes derived from both features (i.e., location and orientation) and captures the location difference (i.e., distance) and the orientation difference (i.e., rotation) for each point with significant change. We tested the proposed method in a mountain region (Sellrain, Tyrol, Austria) covered by three airborne laser scanning point cloud pairs with different point densities and complex topographic changes at intervals of four, six, and ten years. Our method detected significant changes in 91.39 % − 93.03 % of the study area, while a state-of-the-art method (i.e., Multiscale Model-to-Model Cloud Comparison, M3C2) identified 36.81 % − 47.41 % significant changes for the same area. Especially for unchanged building roofs, our method measured lower change magnitudes than M3C2. Looking at the case of shallow landslides, our method identified 84 out of a total of 88 reference landslides by analysing change in distance or rotation. Therefore, our method not only detects a large number of significant changes but also quantifies two types of topographic changes (i.e., distance and rotation), and is more robust against registration errors. It shows large potential for estimation and interpretation of topographic changes in natural environments. [ABSTRACT FROM AUTHOR]

Terrestrial laser scanning (TLS) and camera-equipped small unmanned aircraft systems (sUAS) are two methods that are often used to produce dense point clouds for several monitoring applications. This paper compares the two methods in their ability to provide accurate monitoring information for rockfill embankment dams. We compare the two methods in terms of their uncertainty, data completeness, and field data acquisition/processing challenges. For both datasets, we derive an error budget that considers registration and measurement uncertainty. We also proceed to merge the TLS and sUAS data and leverage the advantages of each method. Furthermore, we conduct an analysis of the multiscale model-to-model cloud comparison (M3C2) input parameters, namely projection scale, normal scale, and sub-sampling of the reference point cloud, to show their effect on the M3C2 distance estimation. The theoretical methodologies and practical considerations of this paper can assist surveyors, who conduct monitoring of rockfill embankment dams using point clouds, in establishing reliable change/deformation estimations. [ABSTRACT FROM AUTHOR]

In recent years, the growth of digital data has been unimaginable. This also applies to geospatial data. One of the largest data types is LiDAR point clouds. Their large volumes on disk, both at the acquisition and processing stages, and in the final versions translate into a high demand for disk space and therefore electricity. It is therefore obvious that in order to reduce energy consumption, lower the carbon footprint of the activity and sensitize sustainability in the digitization of the industry, lossless compression of the aforementioned datasets is a good solution. In this article, a new format for point clouds—3DL—is presented, the effectiveness of which is compared with 21 available formats that can contain LiDAR data. A total of 404 processes were carried out to validate the 3DL file format. The validation was based on four LiDAR point clouds stored in LAS files: two files derived from ALS (airborne laser scanning), one in the local coordinate system and the other in PL-2000; and two obtained by TLS (terrestrial laser scanning), also with the same georeferencing (local and national PL-2000). During research, each LAS file was saved 101 different ways in 22 different formats, and the results were then compared in several ways (according to the coordinate system, ALS and TLS data, both types of data within a single coordinate system and the time of processing). The validated solution (3DL) achieved CR (compression rate) results of around 32% for ALS data and around 42% for TLS data, while the best solutions reached 15% for ALS and 34% for TLS. On the other hand, the worst method compressed the file up to 424.92% (ALS_PL2000). This significant reduction in file size contributes to a significant reduction in energy consumption during the storage of LiDAR point clouds, their transmission over the internet and/or during copy/transfer. For all solutions, rankings were developed according to CR and CT (compression time) parameters. [ABSTRACT FROM AUTHOR]

This study presents a novel filtering methodology for Mobile Laser Scanning (MLS) data using robust iterative reweighting. Initially, 3D point clouds are projected onto a 2D grid to create surfaces from the lowest points. Weights are assigned based on the Height Above Ground (HAG) of these points. Ground points are distinguished by applying a surface function to the dataset via iterative reweighting. Among the tested four robust weight functions, the Denmark and Beaton-Tukey functions outperformed others, achieving total error values of 2.30 and 2.32 across three test areas, respectively. This method efficiently filters MLS data, irrespective of ground point proportions. [ABSTRACT FROM AUTHOR]

The occurrences of large rock slides often result in catastrophic debris flow within high mountain environments. Discontinuity intersected blocks meeting kinematic conditions stemming from deglaciation-related damage can be triggered by external factors, leading to massive rock slides with a significant downstream hazard. This study presents a comprehensive analysis underlining the mechanism and evolution of the failure during the 2017 Santa Lucía landslide, Patagonian Andes, Chile, utilizing remote sensing and numerical modelling. Due to the remote location, aerial photogrammetry was used to unravel the structural and geomorphological configuration, and four discontinuity sets were identified. Based on colour-shaded relief and slope kinematic analysis, it was found that the failure is governed by combinations of three different discontinuity sets. The failure in the crown portion is complex due to resulting planar and wedge surfaces, whereas in the toe region, the failure is governed by the wedge formation between bedding and other joint set. To further examine its mechanism and evolution, rigid block numerical models were developed in 3DEC to reproduce the failure with real topography and joint parameters. The maximum displacement was observed in the same topographical region where the actual failure occurred, thus conforming to the role of discontinuities in the evolution of the catastrophic failure. Acting on a reduced strength due to rock damage, the modelled slope boosts the instability leading to higher displacements along bonding surfaces with similar attributes as observed in the field. A detailed methodology is discussed regarding coupling remote sensing and 3D numerical modelling for detailed insights into the failure mechanism of the landslides. Overall, our results demonstrate that the Santa Lucía rock slide is a structurally controlled failure where joints provided kinematic freedom, favoured by long-term rock slope damage due to deglaciation. [ABSTRACT FROM AUTHOR]

Three-dimensional registration is an established yet challenging problem that is key in many different applications, such as mapping the environment for autonomous vehicles, or modeling people for avatar creation, among others. Registration refers to the process of mapping multiple data into the same coordinate system by means of matching correspondences and transformation estimation. Novel proposals exploit the benefits of deep learning architectures for this purpose, as they learn the best features for the data, providing better matches and hence results. However, the state of the art is usually focused on cases of relatively small transformations, although in certain applications and in a real and practical environment, large transformations are very common. In this paper, we present ReLaTo (Registration for Large Transformations), an architecture that addresses the cases where large transformations happen while maintaining good performance for local transformations. This proposal uses a novel Softmax pooling layer to find correspondences in a bilateral consensus manner between two point sets, sampling the most confident matches. These matches estimate a coarse and global registration using weighted Singular Value Decomposition (SVD). A target-guided denoising step is applied to both the obtained matches and latent features to estimate the final fine registration considering the local geometry. All these steps are carried out following an end-to-end approach, which has been shown to perform better than 7 state-of-the-art registration methods in two datasets commonly used for this task (ModelNet40 and the Karlsruhe Institute of Technology and Toyota Technological Institute dataset, KITTI), especially in the case of large transformations. [ABSTRACT FROM AUTHOR]

Terrestrial laser scanning (TLS) provides highly detailed 3D information of forest environments but is limited to small spatial scales, as data collection is time consuming compared to other remote sensing techniques. Furthermore, TLS data collection is heavily dependent on wind conditions, as the movement of trees negatively impacts the acquired data. Hardware advancements resulting in faster data acquisition times have the potential to be valuable in upscaling efforts but might impact overall data quality. In this study, we investigated the impact of the pulse repetition rate (PRR), or pulse frequency, which is the number of laser pulses emitted per second by the scanner. Increasing the PRR reduces the scan time required for a single scan but decreases the power (amplitude) of the emitted laser pulses commensurately. This trade-off could potentially impact the quality of the acquired data. We used a RIEGL VZ400i laser scanner to test the impact of different PRR settings on the point cloud quality and derived tree structural metrics from individual tree point clouds (diameter, tree height, crown projected area) as well as quantitative structure models (total branch length, tree volume). We investigated this impact across five field plots of different forest complexity and canopy density for three different PRR settings (300, 600 and 1200 kHz). The scan time for a single scan was 180, 90 and 45 s for 300, 600 and 1200 kHz, respectively. Differences among the raw acquired scans from different PRR replicates were largely removed by several necessary data processing steps, notably the removal of uncertain points with a low reflectance attribute. We found strong agreement between the individual tree structural metrics derived from each of the PRR replicates, independent of the forest complexity. This was the case for both point cloud-based metrics and those derived from quantitative structural models (QSMs). The results demonstrate that the PRR in high-end TLS instruments can be increased for data collection with negligible impact on a selection of derived structural metrics that are commonly used in the context of aboveground biomass estimation. [ABSTRACT FROM AUTHOR]

The sudden development of systems capable of rapidly acquiring dense point clouds has underscored the importance of data processing and pre-processing prior to modeling. This work presents the implementation of a denoising algorithm for point clouds acquired with LiDAR SLAM systems, aimed at optimizing data processing and the reconstruction of surveyed object geometries for graphical rendering and modeling. Implemented in a MATLAB environment, the algorithm utilizes an approximate modeling of a reference surface with Poisson's model and a statistical analysis of the distances between the original point cloud and the reconstructed surface. Tested on point clouds from historically significant buildings with complex geometries scanned with three different SLAM systems, the results demonstrate a satisfactory reduction in point density to approximately one third of the original. The filtering process effectively removed about 50% of the points while preserving essential details, facilitating improved restitution and modeling of architectural and structural elements. This approach serves as a valuable tool for noise removal in SLAM-derived datasets, enhancing the accuracy of architectural surveying and heritage documentation. [ABSTRACT FROM AUTHOR]

Accurate digital elevation models (DEMs) derived from airborne light detection and ranging (LiDAR) data are crucial for terrain analysis applications. As established in the literature, higher point density improves terrain representation but requires greater data storage and processing capacities. Therefore, point cloud sampling is necessary to reduce densities while preserving DEM accuracy as much as possible. However, there has been a limited examination directly comparing the effects of various sampling algorithms on DEM accuracy. This study aimed to help fill this gap by evaluating and comparing the performance of three common point cloud sampling methods octree, spatial, and random sampling methods in high terrain. DEMs were then generated from the sampled point clouds using three different interpolation algorithms: inverse distance weighting (IDW), natural neighbor (NN), and ordinary kriging (OK). The results showed that octree sampling consistently produced the most accurate DEMs across all metrics and terrain slopes compared to other methods. Spatial sampling also produced more accurate DEMs than random sampling but was less accurate than octree sampling. The results can be attributed to differences in how the sampling methods represent terrain geometry and retain microtopographic detail. Octree sampling recursively subdivides the point cloud based on density distributions, closely conforming to complex microtopography. In contrast, random sampling disregards underlying densities, reducing accuracy in rough terrain. The findings guide optimal sampling and interpolation methods of airborne lidar point clouds for generating DEMs for similar complex mountainous terrains. [ABSTRACT FROM AUTHOR]

Roof plane segmentation from airborne light detection and ranging (LiDAR) point clouds is an important technology for three-dimensional (3D) building model reconstruction. One of the key issues of plane segmentation is how to design powerful features that can exactly distinguish adjacent planar patches. The quality of point feature directly determines the accuracy of roof plane segmentation. Most of existing approaches use handcrafted features, such as point-to-plane distance, normal vector, etc., to extract roof planes. However, the abilities of these features are relatively low, especially in boundary areas. To solve this problem, we propose a boundary-aware point clustering approach in Euclidean and embedding spaces constructed by a multi-task deep network for roof plane segmentation. We design a three-branch multi-task network to predict semantic labels, point offsets and extract deep embedding features. In the first branch, we classify the input data as non-roof, boundary and plane points. In the second branch, we predict point offsets for shifting each point towards its respective instance center. In the third branch, we constrain that points of the same plane instance should have the similar embeddings. We aim to ensure that points of the same plane instance are close as much as possible in both Euclidean and embedding spaces. However, although deep network has strong feature representative ability, it is still hard to accurately distinguish points near the plane instance boundary. Therefore, we first robustly group plane points into many clusters in Euclidean and embedding spaces to find candidate planes. Then, we assign the rest boundary points to their closest clusters to generate the final complete roof planes. In this way, we can effectively reduce the influence of unreliable boundary points. In addition, to train the network and evaluate the performance of our approach, we prepare a synthetic dataset and two real datasets. The experiments conducted on synthetic and real datasets show that the proposed approach significantly outperforms the existing state-of-the-art approaches in both qualitative evaluation and quantitative metrics. To facilitate future research, we will make datasets and source code of our approach publicly available at https://github.com/Li-Li-Whu/DeepRoofPlane. [ABSTRACT FROM AUTHOR]

This study introduced a novel approach to 3D image segmentation utilizing a neural network framework applied to 2D depth map imagery, with Z axis values visualized through color gradation. This research involved comprehensive data collection from mechanically harvested wild blueberries to populate 3D and red–green–blue (RGB) images of filled totes through time-of-flight and RGB cameras, respectively. Advanced neural network models from the YOLOv8 and Detectron2 frameworks were assessed for their segmentation capabilities. Notably, the YOLOv8 models, particularly YOLOv8n-seg, demonstrated superior processing efficiency, with an average time of 18.10 ms, significantly faster than the Detectron2 models, which exceeded 57 ms, while maintaining high performance with a mean intersection over union (IoU) of 0.944 and a Matthew's correlation coefficient (MCC) of 0.957. A qualitative comparison of segmentation masks indicated that the YOLO models produced smoother and more accurate object boundaries, whereas Detectron2 showed jagged edges and under-segmentation. Statistical analyses, including ANOVA and Tukey's HSD test (α = 0.05), confirmed the superior segmentation performance of models on depth maps over RGB images (p < 0.001). This study concludes by recommending the YOLOv8n-seg model for real-time 3D segmentation in precision agriculture, providing insights that can enhance volume estimation, yield prediction, and resource management practices. [ABSTRACT FROM AUTHOR]

Learning object shapes from a single image is challenging due to variations in scene content, geometric structures, and environmental factors, which create significant disparities between 2D image features and their corresponding 3D representations, hindering the effective training of deep learning models. Existing learning-based approaches can be divided into two-stage and single-stage methods, each with limitations. Two-stage methods often rely on generating intermediate proposals by searching for similar structures across the entire dataset, a process that is computationally expensive due to the large search space and high-dimensional feature-matching requirements, further limiting flexibility to predefined object categories. In contrast, single-stage methods directly reconstruct 3D shapes from images without intermediate steps, but they struggle to capture complex object geometries due to high feature loss between image features and 3D shapes and limit their ability to represent intricate details. To address these challenges, this paper introduces SS3DNet-AF, a single-stage, single-view 3D reconstruction network with an attention-based fusion (AF) mechanism to enhance focus on relevant image features, effectively capturing geometric details and generalizing across diverse object categories. The proposed method is quantitatively evaluated using the ShapeNet dataset, demonstrating its effectiveness in achieving accurate 3D reconstructions while overcoming the computational challenges associated with traditional approaches. [ABSTRACT FROM AUTHOR]

Wood–leaf separation from forest LiDAR point clouds is a challenging task due to the complex and irregular structures of tree canopies. Traditional machine vision and deep learning methods often struggle to accurately distinguish between fine branches and leaves. This challenge arises primarily from the lack of suitable features and the limitations of existing position encodings in capturing the unique and intricate characteristics of forest point clouds. In this work, we propose an innovative approach that integrates Local Surface Features (LSF) and a Position Encoding (PosE) module within the Point Transformer (PT) network to address these challenges. We began by preprocessing point clouds and applying a machine vision technique, supplemented by manual correction, to create wood–leaf-separated datasets of forest point clouds for training. Next, we introduced Point Feature Histogram (PFH) to construct LSF for each point network input, while utilizing Fast PFH (FPFH) to enhance computational efficiency. Subsequently, we designed a PosE module within PT, leveraging trigonometric dimensionality expansion and Random Fourier Feature-based Transformation (RFFT) for nuanced feature analysis. This design significantly enhances the representational richness and precision of forest point clouds. Afterward, the segmented branch point cloud was used to model tree skeletons automatically, while the leaves were incorporated to complete the digital twin. Our enhanced network, tested on three different types of forests, achieved up to 96.23% in accuracy and 91.51% in mean intersection over union (mIoU) in wood–leaf separation, outperforming the original PT by approximately 5%. This study not only expands the limits of forest point cloud research but also demonstrates significant improvements in the reconstruction results, particularly in capturing the intricate structures of twigs, which paves the way for more accurate forest resource surveys and advanced digital twin construction. [ABSTRACT FROM AUTHOR]

Perception systems for assisted driving and autonomy enable the identification and classification of objects through a concentration of sensors installed in vehicles, including Radio Detection and Ranging (RADAR), camera, Light Detection and Ranging (LIDAR), ultrasound, and HD maps. These sensors ensure a reliable and robust navigation system. Radar, in particular, operates with electromagnetic waves and remains effective under a variety of weather conditions. It uses point cloud technology to map the objects in front of you, making it easy to group these points to associate them with real-world objects. Numerous clustering algorithms have been developed and can be integrated into radar systems to identify, investigate, and track objects. In this study, we evaluate several clustering algorithms to determine their suitability for application in automotive radar systems. Our analysis covered a variety of current methods, the mathematical process of these methods, and presented a comparison table between these algorithms, including Hierarchical Clustering, Affinity Propagation Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH), Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Mini-Batch K-Means, K-Means Mean Shift, OPTICS, Spectral Clustering, and Gaussian Mixture. We have found that K-Means, Mean Shift, and DBSCAN are particularly suitable for these applications, based on performance indicators that assess suitability and efficiency. However, DBSCAN shows better performance compared to others. Furthermore, our findings highlight that the choice of radar significantly impacts the effectiveness of these object recognition methods. [ABSTRACT FROM AUTHOR]

Accurate semantic segmentation in transmission corridor scenes is crucial for the maintenance and inspection of power infrastructure, facilitating the timely detection of potential hazards. In this study, we propose SA-KPConv, an advanced segmentation model specifically designed for transmission corridor scenarios. Traditional approaches, including Random Forest and point-based deep learning models such as PointNet++, demonstrate limitations in segmenting critical infrastructure components, particularly power lines and towers, primarily due to their inadequate capacity to capture complex spatial relationships and local geometric details. Our model effectively addresses these challenges by integrating a spatial attention module with kernel point convolution, enhancing both global context and local feature extraction. Experiments demonstrate that SA-KPConv outperforms state-of-the-art methods, achieving a mean Intersection over Union (mIoU) of 89.62%, particularly excelling in challenging terrains such as mountainous areas. Ablation studies further validate the significance of our model's components in enhancing overall performance and effectively addressing class imbalance. This study presents a robust solution for semantic segmentation, with considerable potential for monitoring and maintaining power infrastructure. [ABSTRACT FROM AUTHOR]

Three raster-based (RB) and one point cloud-based (PCB) algorithms were tested to segment individual Aleppo pine trees and extract their tree height (H) and crown diameter (CD) using two types of point clouds generated from two different techniques: (1) Low-Density (≈1.5 points/m2) Airborne Laser Scanning (LD-ALS) and (2) photogrammetry based on high-resolution unmanned aerial vehicle (UAV) images. Through intensive experiments, it was concluded that the tested RB algorithms performed best in the case of UAV point clouds (F1-score > 80.57%, H Pearson's r > 0.97, and CD Pearson´s r > 0.73), while the PCB algorithm yielded the best results when working with LD-ALS point clouds (F1-score = 89.51%, H Pearson´s r = 0.94, and CD Pearson´s r = 0.57). The best set of algorithm parameters was applied to all plots, i.e., it was not optimized for each plot, in order to develop an automatic pipeline for mapping large areas of Mediterranean forests. In this case, tree detection and height estimation showed good results for both UAV and LD-ALS (F1-score > 85% and >76%, and H Pearson´s r > 0.96 and >0.93, respectively). However, very poor results were found when estimating crown diameter (CD Pearson´s r around 0.20 for both approaches). [ABSTRACT FROM AUTHOR]

To bridge the gap in as-built Building Information Model (BIM) creation between the architectural, engineering, and construction (AEC) community and the computer vision community, this paper presents an automated Scan-to-BIM framework for modeling both structural and nonstructural building components using a low-cost scanning data. The state-of-the-art instance-level semantic segmentation algorithm, SoftGroup, is adopted to classify individual building components. Detected wall segments are projected onto a two-dimensional (2D) XY grid, and an interest point detection algorithm, SuperPoint, is used to extract wall corner points. Subsequently, a series of refinement steps is proposed to generate the wall boundary. With optimized parameters, an intersection-over-union of 82.56% was achieved when tested on the benchmark Stanford Three-Dimensional (3D) Indoor Scene Data Set. Our results demonstrated the usability of the proposed wall boundary extraction to the incomplete and complex indoor scan data compared to an existing as-built modeling method. Instance-level segments and the refined wall boundary were combined to generate as-built BIM via parametric modeling. [ABSTRACT FROM AUTHOR]

In transportation infrastructure systems, feature images and spatial characteristics are generally utilized as complementary elements derived from point clouds for road edge extraction, but the involvement of one or more hyperparameters in each makes the extraction complicated. This study proposes an autotuning hybrid method with Bayesian optimization for road edge extraction in highway systems. The hybrid method combines the strengths of 2D feature images and 3D spatial characteristics while also automatically tuning the hyperparameter combination using Bayesian optimization. The hyperparameters encompass high and low pixel gradient thresholds, neighborhood radius, and normal vector threshold. Later, the point cloud dataset of national highways in Henan Province, China, is taken as the case study to evaluate the performance of the proposed method against three benchmark methods in two typical road scenarios: straight and curved edges. Experimental results show that the proposed method outperforms the benchmarks in detection quality and accuracy. It can serve as a decision-making tool to complement traditional manual road surveying, enabling efficient and automated road edge extraction in highway systems. [ABSTRACT FROM AUTHOR]

The aim of this paper is to propose a workflow for the real-time visualization of virtual environments that supports diagnostic tasks in heritage buildings. The approach integrates data from terrestrial laser scanning (3D point clouds and meshes), along with panoramic and thermal images, into a unified virtual model. Additionally, the methodology incorporates several post-processing stages designed to enhance the user experience in visualizing both the building and its associated damage. The methodology was tested on the Medieval Templar Church of Vera Cruz in Segovia, utilizing a combination of visible and infrared data, along with manually prepared damage maps. The project results demonstrate that the use of a hybrid digital model—combining 3D point clouds, polygonal meshes, and panoramic images—is highly effective for real-time rendering, providing detailed visualization while maintaining adaptability for mobile devices with limited computational power. [ABSTRACT FROM AUTHOR]

This paper presents a study on aircraft engine identification using real-time 3D LiDAR point cloud segmentation technology, a key element for the development of automated docking systems in airport boarding facilities, known as jetbridges. To achieve this, 3D LiDAR sensors utilizing a spinning method were employed to gather surrounding environmental 3D point cloud data. The raw 3D environmental data were then filtered using the 3D RANSAC technique, excluding ground data and irrelevant apron areas. Segmentation was subsequently conducted based on the filtered data, focusing on aircraft sections. For the segmented aircraft engine parts, the centroid of the grouped data was computed to determine the 3D position of the aircraft engine. Additionally, PointNet was applied to identify aircraft engines from the segmented data. Dynamic tests were conducted in various weather and environmental conditions, evaluating the detection performance across different jetbridge movement speeds and object-to-object distances. The study achieved a mean intersection over union (mIoU) of 81.25% in detecting aircraft engines, despite experiencing challenging conditions such as low-frequency vibrations and changes in the field of view during jetbridge maneuvers. This research provides a strong foundation for enhancing the robustness of jetbridge autonomous docking systems by reducing the sensor noise and distortion in real-time applications. Our future research will focus on optimizing sensor configurations, especially in environments where sea fog, snow, and rain are frequent, by combining RGB image data with 3D LiDAR information. The ultimate goal is to further improve the system's reliability and efficiency, not only in jetbridge operations but also in broader autonomous vehicle and robotics applications, where precision and reliability are critical. The methodologies and findings of this study hold the potential to significantly advance the development of autonomous technologies across various industrial sectors. [ABSTRACT FROM AUTHOR]

Low-frequency Ultra-WideBand (UWB) radar offers significant advantages in the field of human activity recognition owing to its excellent penetration and resolution. To address the issues of high computational complexity and extensive network parameters in existing action recognition algorithms, this study proposes an efficient and lightweight human activity recognition method using UWB radar based on spatiotemporal point clouds. First, four-dimensional motion data of the human body are collected using UWB radar. A discrete sampling method is then employed to convert the radar images into point cloud representations. Because human activity recognition is a classification problem on time series, this paper combines the PointNet++ network with the Transformer network to propose a lightweight spatiotemporal network. By extracting and analyzing the spatiotemporal features of four-dimensional point clouds, end-to-end human activity recognition is achieved. During the model training process, a multithreshold fusion method is proposed for point cloud data to further enhance the model’s generalization and recognition capabilities. The proposed method is then validated using a public four-dimensional radar imaging dataset and compared with existing methods. The results show that the proposed method achieves a human activity recognition rate of 96.75% while consuming fewer parameters and computational resources, thereby verifying its effectiveness.

As mining operations extend to greater depths, the risk of deformation in high-stress tunnels increases significantly, posing a substantial threat. This study introduces a novel framework known as “robust mobility deformation detection” (RM2D), designed for real-time tunnel deformation detection. RM2D employs mobile LiDAR scanner to capture real-time point cloud data within the tunnel. This data is then voxelized and analyzed using covariance matrices to create a voxel-based multi-distribution representation of the rugged tunnel surface. Leveraging this representation, we assess deformations and scrutinize results through machine learning models to swiftly pinpoint tunnel deformation locations. Extensive experimental validation confirms the framework’s capacity to successfully detect deformations, including floor heave, side rib spalling, and roof fall, with remarkable accuracy. For deformation levels at 0.15 m, RM2D was able to successfully detect deformations with an area greater than 2 m2. For deformation areas of (3 ± 0.5) m2, RM2D successfully detected deformations of levels at (0.05 ± 0.01) m, and its detection capability meets the standard criteria for mining tunnel deformation detection. When compared to two conventional methods, RM2D demonstrates its real-time deformation detection capability in complex environments and on rough surfaces with precision, all at speeds below 10 km/h. Furthermore, we evaluated the predictive performance using multiple evaluation metrics and provided insights into the decision mechanism of the machine learning employed in our research, thereby offering valuable information for practical engineering applications in tunnel deformation detection.

Remote Sensing (RS) and Geographic Information Science (GIS) techniques are powerful tools for spatial data collection, analysis, management, and digitization within cultural heritage frameworks. Despite their capabilities, challenges remain in automating data semantic classification for conservation purposes. To address this, leveraging airborne Light Detection And Ranging (LiDAR) point clouds, complex spatial analyses, and automated data structuring is crucial for supporting heritage preservation and knowledge processes. In this context, the present contribution investigates the latest Artificial Intelligence (AI) technologies for automating existing LiDAR data structuring, focusing on the case study of Sardinia coastlines. Moreover, the study preliminary addresses automation challenges in the perspective of historical defensive landscapes mapping. Since historical defensive architectures and landscapes are characterized by several challenging complexities—including their association with dark periods in recent history and chronological stratification—their digitization and preservation are highly multidisciplinary issues. This research aims to improve data structuring automation in these large heritage contexts with a multiscale approach by applying Machine Learning (ML) techniques to low-scale 3D Airborne Laser Scanning (ALS) point clouds. The study thus develops a predictive Deep Learning Model (DLM) for the semantic segmentation of sparse point clouds (2), adaptable to large landscape heritage contexts and heterogeneous data scales. Additionally, a preliminary investigation into object-detection methods has been conducted to map specific fortification artifacts efficiently.

In traditional farming, fertilizers are often used without precision, resulting in unnecessary expenses and potential damage to the environment. This study introduces a new method for accurately identifying macronutrient deficiencies in Rhodena lettuce crops. We have developed a four-stage process. First, we gathered two sets of data for lettuce seedlings: one is composed of color images and the other of point clouds. In the second stage, we employed the interactive closest point (ICP) method to align the point clouds and extract 3D morphology features for detecting nitrogen deficiencies using machine learning techniques. Next, we trained and compared multiple detection models to identify potassium deficiencies. Finally, we compared the outcomes with traditional lab tests and expert analysis. Our results show that the decision tree classifier achieved 90.87% accuracy in detecting nitrogen deficiencies, while YOLOv9c attained an mAP of 0.79 for identifying potassium deficiencies. This innovative approach has the potential to transform how we monitor and manage crop nutrition in agriculture.

Automatic 3D reconstruction of indoor scenes remains a challenging task due to the incomplete and noisy nature of scanned data. We propose a semantic-guided method for reconstructing indoor scene based on semantic segmentation of point clouds. Firstly, a Multi-Feature Adaptive Aggregation Network is designed for semantic segmentation, assigning the semantic label for each point. Then, a novel slicing-projection method is proposed to segment and reconstruct the walls. Next, a hierarchical Euclidean Clustering is proposed to separate objects into individual ones. Finally, each object is replaced with the most similar CAD model from the database, utilizing the Rotational Projection Statistics (RoPS) descriptor and the iterative closest point (ICP) algorithm. The selected template models are then deformed and transformed to fit the objects in the scene. Experimental results demonstrate that the proposed method achieves high-quality reconstruction even when faced with defective scanned point clouds.