Laser Scanning Point Clouds Research Articles

A Feature-Level Point Cloud Fusion Method for Timber Volume of Forest Stands Estimation

Accurate diameter at breast height (DBH) and tree height (H) information can be acquired through terrestrial laser scanning (TLS) and airborne LiDAR scanner (ALS) point cloud, respectively. To utilize these two features simultaneously but avoid the difficulties of point cloud fusion, such as technical complexity and time-consuming and laborious efforts, a feature-level point cloud fusion method (FFATTe) is proposed in this paper. Firstly, the TLS and ALS point cloud data in a plot are georeferenced by differential global navigation and positioning system (DGNSS) technology. Secondly, point cloud processing and feature extraction are performed for the georeferenced TLS and ALS to form feature datasets, respectively. Thirdly, the feature-level fusion of LiDAR data from different data sources is realized through spatial join according to the tree trunk location obtained from TLS and ALS, that is, the tally can be implemented at a plot. Finally, the individual tree parameters are optimized based on the tally results and fed into the binary volume model to estimate the total volume (TVS) in a large area (whole study area). The results show that the georeferenced ALS and TLS point cloud data using DGNSS RTK/PPK technology can achieve coarse registration (mean distance ≈ 40 cm), which meets the accuracy requirements for feature-level point cloud fusion. By feature-level fusion of the two point cloud data, the tally can be achieved quickly and accurately in the plot. The proposed FFATTe method achieves high accuracy (with error of 3.09%) due to its advantages of combining different LiDAR data from different sources in a simple way, and it has strong operability when acquiring TVS over large areas.

DGCN-ED: dynamic graph convolutional networks with encoder–decoder structure and its application for airborne LiDAR point classification

ABSTRACT Achieving high-precision automatic classification in real-world applications for airborne laser scanning point clouds is a challenging task duo to their unstructured nature, uneven density distribution, high redundancy, incompleteness and scene complexity. Graph Convolutional Neural Networks can process scattered point clouds directly without regularization, which avoids the loss of depth information，and has recently become a topic of increased interest. Therefore, this study proposed an extension of the Graph-Unet network architecture named DGCN-ED for airborne LiDAR point classification, which uses a Graph Convolutional Neural Network as a representation to describe complex object relationships and an encoder-decoder architecture to capture the multi-scaled point features and describe objects in the high-level feature space. The two-layer dynamic update Graph Convolutional Neural Network is designed to expand the effective range of nodes and enhance the representation ability of learned pointwise features. The effectiveness of the proposed method is evaluated by an experiment on the ISPRS Vaihingen 3D semantic labelling benchmark dataset. Moreover, experiments on the IEEE 2019 Data Fusion Contest Dataset were conducted to demonstrate the generalization abilities of the proposed method. The results show that our method achieved on average 2.8% higher overall accuracy than existing methods, with an overall accuracy of 98.0% and an average F1 score of 0.797.

Consumer‐grade UAV solid‐state LiDAR accurately quantifies topography in a vegetated fluvial environment

AbstractUnoccupied aerial vehicles (UAVs) with passive optical sensors have become popular for reconstructing topography using Structure from Motion (SfM) photogrammetry. Advances in UAV payloads and the advent of solid‐state LiDAR have enabled consumer‐grade active remote sensing equipment to become more widely available, potentially providing opportunities to overcome some challenges associated with SfM photogrammetry, such as vegetation penetration and shadowing, that can occur when processing UAV‐acquired images. We evaluate the application of a DJI Zenmuse L1 solid‐state LiDAR sensor on a Matrice 300 RTK UAV to generate digital elevation models (DEMs). To assess flying height (60–80 m) and speed parameters (5–10 ms−1) on accuracy, four point clouds were acquired at a test site. These point clouds were used to develop a processing workflow to georeference, filter and classify the point clouds to produce a raster DEM product. A dense control network showed that there was no significant difference in georeferencing from differing flying height or speed. Building on the test results, a 3 km reach of the River Feshie was surveyed, collecting over 755 million UAV LiDAR points. The Multiscale Curvature Classification algorithm was found to be the most suitable classifier of ground topography. GNSS check points showed a mean vertical residual of −0.015 m on unvegetated gravel bars. Multiscale Model to Model Cloud Comparison (M3C2) residuals compared UAV LiDAR and Terrestrial Laser Scanner point clouds for seven sample sites demonstrating a close match with marginally zero residuals. Solid‐state LiDAR was effective at penetrating sparse canopy‐type vegetation but was less penetrable through dense ground‐hugging vegetation (e.g. heather, thick grass). Whilst UAV solid‐state LiDAR needs to be supplemented with bathymetric mapping to produce wet–dry DEMs, by itself, it offers advantages to comparable geomatics technologies for kilometre‐scale surveys. Ten best practice recommendations will assist users of UAV solid‐state LiDAR to produce bare earth DEMs.

PPK PROCESSING TO BOOST UAS ACCURACY IN CADASTRAL MAPPING

Abstract. Unmanned Aerial Systems (UAS) are increasingly used in different applications, including 3D urban modelling, cadastral mapping, urban planning, GIS information system and other fields because of their advantages. As a consequence, UAS equipment is constantly developed to provide more accurate results in a more reliable mode. This paper aims to evaluate the performances of a low-cost UAS system, namely DJI Phantom 4 Pro v2 equipped with a TeoKIT GNSS PPK (post-processing kinematic) module for cadastral mapping purposes. Two fights (oblique and nadir) over a residential area at 60 m height were performed and some 100 ground points were used to derive RMSE accuracies. Comparison between GNSS-aided with PPK processing and indirect georeferencing processes are performed. Given a mobile laser scanner (MLS) point cloud as ground truth, comparison with UAS point clouds and manually digitized features are also performed and reported.

A novel geometric feature-based wood-leaf separation method for large and crown-heavy tropical trees using handheld laser scanning point cloud

ABSTRACT As the prerequisite of non-destructively measuring structural parameters and leaf distribution, wood-leaf separation plays an essential role in forest inventories. However, fewer studies focus on the separation and measurement of large tropical trees with heavy crowns and complex branch structures. This study proposed a novel method to automatically separate the leaf and wood points of large tropical trees based on geometric features. Instead of identifying all wood points using the same rules, we used different methods to separate small and large wood components, respectively. The identification of small wood components was implemented mainly by the differences in point density and linear distribution pattern between leaf and wood points, while the identification of large wood components was implemented through the comprehensive analysis of verticality, linearity, anisotropy, as well as point density. To improve the separation accuracy and implementation effectiveness, the segment-wise and point-wise methods were combined in this study. The robustness and generalization of the proposed method were tested using two datasets, i.e. twenty large tropical trees with heavy crowns and twenty-four general tropical trees without heavy crowns. The separation results verified that the proposed method could achieve good separation of wood and leaf points of large crown-heavy tropical trees with the accuracy of up to 91.5%. The highest separation accuracy of general tropical trees was about 95.03%. The examination of the tropical trees without heavy crowns demonstrated that the proposed method has promising robustness and generalization ability. In addition, to fill the gap in the large tropical tree point clouds, an open-source dataset library was built for the wood-leaf separation research, including manually labelled 20 large crown-heavy tropical trees with different types of branch structures and basic structural parameters of each tree (tree height, crown spread, and diameter at the breast height (DBH)).

Soft Segmentation of Terrestrial Laser Scanning Point Cloud of Forests

As the three-dimensional (3D) laser scanner is widely used for forest inventory, analyzing and processing point cloud data captured with a 3D laser scanner have become an important research topic in recent years. The extraction of single trees from point cloud data is essential for further investigation at the individual tree level, such as counting trees and trunk analysis, and many developments related to this topic have been published. However, constructing an accurate and automated method to obtain the tree crown silhouette from the point cloud data is challenging because the tree crowns often overlap between adjacent trees. A soft segmentation method that uses K-Nearest Neighbor (KNN) and contour shape constraints at the overlap region is proposed to solve this task. Experimental results show that the visual effect of the tree crown shape and the precision of point cloud segmentation have improved. It is concluded that the proposed method works well for tree crown segmentation and silhouette reconstruction from the terrestrial laser scanning point cloud data of the forest.

Instance recognition of street trees from urban point clouds using a three-stage neural network

As one of the most important components of urban space, the geometric and semantic properties of road trees are crucial for the assessment and upgrade of urban environments. Accurate positioning of individual trees is a critical procedure for this task; however, traditional manual methods require considerable personnel and material resources. By contrast, individual tree segmentation based on a vehicle-mounted mobile laser scanner (MLS) operating along urban streets can provide a feasible and efficient solution. Recent studies (especially those based on deep learning) on point cloud semantic segmentation and forest tree detection have achieved significant progress; however, they rarely perform instance recognition of urban trees. In addition, extracting individual urban trees from point clouds to balance accuracy and efficiency remains challenging. To this end, we propose an effective pipeline to segment individual trees from urban MLS data. The proposed approach encompasses two main steps: tree-point extraction and individual-tree segmentation. First, a patch-guided semi-supervised semantic segmentation network was designed to separate non-tree points from entire urban scene. The extracted tree points are further divided into instance-level trees using a three-stage segmentation network for the individual tree segmentation. Specifically, a tree-centroid prediction module is proposed to ensure the accurate localization of street trees, even with irregular positions, in complex urban environments. Subsequently, a special confidence-aware attention mechanism is developed to resolve challenging ambiguities from complementary perspectives. Finally, a dynamic shifting-based clustering algorithm is introduced to speed up the segmentation of each tree guided by the predicted centroid and directly transfer the IDs back to the point clouds. The proposed method is evaluated and validated on MLS point clouds collected in two urban areas, robustly producing accurate instance-level roadside trees. The instance-level segmentation results in average precision and recall of (96.62%, 96.17%) and (95.62%, 94.58%), respectively, on these two MLS point cloud datasets. Point-level segmentation also exhibits mean precision, recall, and weighted coverage exceeding 90%. We also verify that the proposed method could achieve a promising result on an airborne laser scanning point cloud dataset without retraining, thereby demonstrating a good generalization ability. To further verify the quality of the segmented tree instances, a reconstruction method is adopted to employ these independent street trees in trunk point cloud modeling. Our contribution may assist in developing intelligent urban green space planning and management.

Canopy and surface fuel estimations using RPAS and ground-based point clouds

Abstract Forest management activities intended to reduce wildfire risk rely on accurate characterizations of the amount and arrangement of canopy and surface fuels. Metrics that describe these fuels are typically estimated with various systems that transform plot-level field data into metrics that can be used within fire behaviour models. Remote sensing data have long been used to estimate these metrics across large spatial scales, but more advanced, high-density point clouds have the potential to estimate these metrics with higher accuracy. This study collected LiDAR and digital aerial photogrammetric (DAP) point clouds from a remotely piloted aerial system (RPAS), as well as mobile laser scanning (MLS) point clouds from a mobile ground-based system, and compared their ability to estimate fuel metrics. This involved the extraction of predictor variables from each point cloud, of which small subsets were used to estimate various fuel metrics. These included six overstory canopy metrics (stand height, canopy cover, tree density, canopy fuel load, canopy bulk density and canopy base height), three diameter at breast height (DBH)–related metrics (stand density index, basal area and quadratic mean diameter) and three surface fuel metrics (total woody debris (TWD), coarse woody debris (CWD) and fine woody debris (FWD)). Overall, canopy metrics were estimated most accurately by the RPAS LiDAR models, although none of the point clouds were able to accurately estimate DBH-related metrics. For the other six canopy metrics, RPAS LiDAR models had an average R2 value of 0.70; DAP – 0.63 and MLS – 0.63. CWD (>7 cm) and TWD loads were estimated most accurately by the MLS models (average R2 values – 0.70), followed by the RPAS LiDAR – 0.38 and DAP – 0.13. None of these models were able to accurately estimate FWD loads (≤7 cm in diameter), with the three types of point clouds having a maximum R2 value of 0.08. Overall, this research shows the relative ability of three types of high-density point clouds to estimate metrics relevant for fire behaviour modeling.

D-Net: A Density-Based Convolutional Neural Network for Mobile LiDAR Point Clouds Classification in Urban Areas

The 3D semantic segmentation of a LiDAR point cloud is essential for various complex infrastructure analyses such as roadway monitoring, digital twin, or even smart city development. Different geometric and radiometric descriptors or diverse combinations of point descriptors can extract objects from LiDAR data through classification. However, the irregular structure of the point cloud is a typical descriptor learning problem—how to consider each point and its surroundings in an appropriate structure for descriptor extraction? In recent years, convolutional neural networks (CNNs) have received much attention for automatic segmentation and classification. Previous studies demonstrated deep learning models’ high potential and robust performance for classifying complicated point clouds and permutation invariance. Nevertheless, such algorithms still extract descriptors from independent points without investigating the deep descriptor relationship between the center point and its neighbors. This paper proposes a robust and efficient CNN-based framework named D-Net for automatically classifying a mobile laser scanning (MLS) point cloud in urban areas. Initially, the point cloud is converted into a regular voxelized structure during a preprocessing step. This helps to overcome the challenge of irregularity and inhomogeneity. A density value is assigned to each voxel that describes the point distribution within the voxel’s location. Then, by training the designed CNN classifier, each point will receive the label of its corresponding voxel. The performance of the proposed D-Net method was tested using a point cloud dataset in an urban area. Our results demonstrated a relatively high level of performance with an overall accuracy (OA) of about 98% and precision, recall, and F1 scores of over 92%.

Extraction of Forest Structural Parameters by the Comparison of Structure from Motion (SfM) and Backpack Laser Scanning (BLS) Point Clouds

Forest structural parameters are key indicators for forest growth assessment, and play a critical role in forest resources monitoring and ecosystem management. Terrestrial laser scanning (TLS) can obtain three-dimensional (3D) forest structures with ultra-high precision without destruction, whereas some shortcomings such as non-portability and cost-consuming can limit the quick and broad acquisition of forest structure. Structure from motion (SfM) and backpack laser scanning (BLS) technology have the advantages of low-cost and high-portability while obtaining 3D structure information of forests. In this study, the high-overlapped images and the BLS point cloud, combined with the point cloud registration and individual tree segmentation to extract the forest structural parameters and compared with the TLS for assessing the accuracy and efficiency of low-cost SfM and portable BLS point clouds. Three plots with different forest structural complexity (coniferous, broadleaf and mixed plot) in the northern subtropical forests were selected. Firstly, portable photography camera, BLS and TLS were used to acquire 3D SfM and LiDAR point clouds, and spatial co-registration of different-sourced point cloud datasets were carried out based on the understory markers. Secondly, the point clouds of individual tree trunk and crown were segmented by the comparative shortest-path algorithm (CSP), and then the height and position of individual tree were extracted based on the tree crown point cloud. Thirdly, the trunk diameter at different heights were calculated by point cloud slices using the density-based spatial clustering of applications with noise (DBSCAN) algorithm, and combined with the stem curve of individual tree which was constructed using four Taper equations to estimate the individual tree volume. Finally, the extraction accuracy of forest structural parameters based on SfM and BLS point clouds were verified and comprehensively compared with field-measured and TLS data. The results showed that: (1) the individual tree segmentation based on SfM and BLS point clouds all performed quite well, among which the segmentation accuracy (F) of SfM point cloud was 0.80 and the BLS point cloud was 0.85; and (2) the accuracy of DBH and tree height extraction based on the SfM and BLS point clouds in comparison with the field-measured data were relatively high. The root mean square error (RMSE) of DBH and tree height extraction based on SfM point cloud were 2.15 cm and 4.08 m, and the RMSE of DBH and tree height extraction based on BLS point cloud were 2.06 cm and 1.63 m. This study shows that with the adopted image capture method, terrestrial SfM photogrammetry can be applied quite well in extracting DBH.

A Self-contained Stellar Vault Construction Method. The Vault of the Matthias Oratorio in the Inner City Parish Church of Budapest

In the present study, we analysed the laser-scanned point cloud of the stellar vault of the Inner City Parish Church of Budapest's Matthias oratorio. Our analysis concerned the global geometry of the rib network, as well as the webbing. During the study of the rib system, we reconstructed a probable construction method of the late 15th – early 16th-century vault, which seems to be a self-contained construction system in regard to the rib system's plan, the height of the junction points and the curvature of the individual ribs. We found, that to the latter the widely known "Prinzipalbogen" theory applies, however, the "theory of the longest route", which is often mentioned together with the former, most likely was not used. Mapping the webbing and the ribs allowed us to examine the building technique of the webs too.

An Optimized Workflow for Digital Surface Model Series Generation Based on Historical Aerial Images: Testing and Quality Assessment in the Beach-Dune System of Sa Ràpita-Es Trenc (Mallorca, Spain)

We propose an optimized Structure-from-Motion (SfM) Multi-View Stereopsis (MVS) workflow, based on minimizing different errors and inaccuracies of historical aerial photograph series (1945, 1979, 1984, and 2008 surveys), prior to generation of elevation-calibrated historical Digital Surface Models (hDSM) at 1 m resolution. We applied LiDAR techniques on Airborne Laser Scanning (ALS) point clouds (Spanish PNOA LiDAR flights of 2014 and 2019) for comparison and validation purposes. Implementation of these products in multi-temporal analysis requires quality control due to the diversity of sources and technologies involved. To accomplish this, (i) we used the Mean Absolute Error (MAE) between GNSS-Validation Points and the elevations observed by DSM-ALS to evaluate the elevation accuracy of DSM-ALS generated with the LAScatalog processing engine; (ii) optimization of the SfM sparse clouds in the georeferencing step was evaluated by calculating the Root Mean Square Error (RMSE) between the Check Points extracted from DSM-ALS and the predicted elevations per sparse cloud; (iii) the MVS clouds were evaluated by calculating the MAE between ALS-Validation Points and the predicted elevations per MVS cloud; iv) the accuracy of the resulting historical SfM-MVS DSMs were assessed using the MAE between ALS-Validation Points and the observed elevations per historical DSM; and (v) we implemented a calibration method based on a linear correction to reduce the elevation discrepancies between historical DSMs and the DSM-ALS 2019 reference elevations. This optimized workflow can generate high-resolution (1 m pixel size) hDSMs with reasonable accuracy: MAE in z ranges from 0.41 m (2008 DSM) to 5.21 m (1945 DSM). Overall, hDSMs generated using historical images have great potential for geo-environmental processes monitoring in different ecosystems and, in some cases (i.e., sufficient image overlapping and quality), being an acceptable replacement for LiDAR data when it is not available.

Generating synthetic laser scanning data of forests by combining forest inventory information, a tree point cloud database and an open-source laser scanning simulator

Abstract Airborne laser scanning (ALS) data are routinely used to estimate and map structure-related forest inventory variables. The further development, refinement and evaluation of methods to derive forest inventory variables from ALS data require extensive datasets of forest stand information on an individual tree-level and corresponding ALS data. A cost-efficient method to obtain such datasets is the combination of virtual forest stands with a laser scanning simulator. We present an approach to simulate ALS data of forest stands by combining forest inventory information, a tree point cloud database and the laser scanning simulation framework HELIOS++. ALS data of six 1-ha plots were simulated and compared to real ALS data of these plots. The synthetic 3D representations of the forest stands were composed of real laser scanning point clouds of individual trees that were acquired by an uncrewed aerial vehicle (UAV), and, for comparison, simplified tree models with cylindrical stems and spheroidal crowns. The simulated ALS point clouds of the six plots were compared with the real point clouds based on canopy cover, height distribution of returns and several other point cloud metrics. In addition, the performance of biomass models trained using these synthetic data was evaluated. The comparison revealed that, in general, both the real tree models and the simplified tree models can be used to generate synthetic data. The results differed for the different study sites and depending on whether all returns or only first returns were considered. The measure of canopy cover was better represented by the data of the simplified tree models, whereas the height distribution of the returns was – for most of the study sites – better represented by the real tree model data. Training biomass models with metrics derived from the real tree model data led to an overestimation of biomass, while using metrics of the simplified tree model data resulted in an underestimation of biomass. Still, the accuracy of models trained with simulated data was only slightly lower compared to models trained with real ALS data. Our results suggest that the presented approach can be used to generate ALS data that are sufficiently realistic for many applications. The synthetic data may be used to develop new or refine existing ALS-based forest inventory methods, to systematically investigate the relationship between point cloud metrics and forest inventory variables and to analyse how this relationship is affected by laser scanning acquisition settings and field reference data characteristics.

Road-Side Individual Tree Segmentation from Urban MLS Point Clouds Using Metric Learning

As one of the most important components of urban space, an outdated inventory of road-side trees may misguide managers in the assessment and upgrade of urban environments, potentially affecting urban road quality. Therefore, automatic and accurate instance segmentation of road-side trees from urban point clouds is an important task in urban ecology research. However, previous works show under- or over-segmentation effects for road-side trees due to overlapping, irregular shapes and incompleteness. In this paper, a deep learning framework that combines semantic and instance segmentation is proposed to extract single road-side trees from vehicle-mounted mobile laser scanning (MLS) point clouds. In the semantic segmentation stage, the ground points are filtered to reduce the processing time. Subsequently, a graph-based semantic segmentation network is developed to segment road-side tree points from the raw MLS point clouds. For the individual tree segmentation stage, a novel joint instance and semantic segmentation network is adopted to detect instance-level roadside trees. Two complex Chinese urban point cloud scenes are used to evaluate the individual urban tree segmentation performance of the proposed method. The proposed method accurately extract approximately 90% of the road-side trees and achieve better segmentation results than existing published methods in both two urban MLS point clouds. Living Vegetation Volume (LVV) calculation can benefit from individual tree segmentation. The proposed method provides a promising solution for ecological construction based on the LVV calculation of urban roads.

Optimal bucking of stems from terrestrial laser scanning data to maximize forest value

ABSTRACT An optimal bucking strategy that allocates cutting patterns to forest stands based on the individual characteristics of each stem is critical for maximizing value recovery. However, cutting patterns are usually excluded from bucking algorithms due to the difficulties associated with capturing tree quality features when collecting forest inventory data (e.g. branchiness and tree shape). This paper presents a non-destructive and fully automated methodology for the optimal bucking of stems based on terrestrial laser scanning (TLS) point clouds that aims to maximize the economic value of trees in a forest stand. It is based on the three-dimensional modelling of stems and includes the diameter and curvature of each log. The bucking algorithm also considers several timber products and calculates the most valuable log combination for each tree. The methodology was tested in a Pinus radiata plot with 120 trees, and the results were compared with those obtained with input data that do not take curvature into account: i.e. only diameters from TLS and taper equations. The analysis of the results suggests that not including curvature in the algorithm for optimal bucking results in an overestimation of the commercial value of timber products.

Identification and Quantification of Surface Depressions on Grassy Land Surfaces of Different Topographic Attributes Using High-Resolution Terrestrial Laser Scanning Point Cloud and Triangulated Irregular Network

The objective of this study was to identify and quantify surface depressions on grass-covered land surfaces using a high-resolution terrestrial laser scanning (TLS) point cloud, and a triangulated irregular network (TIN). The entire grassy land surface in the study area was divided into five subwatersheds of different topographic attributes (i.e., depression depth and surface slope). Surface depressions were identified and quantified using a TIN generated from a high-resolution TLS point cloud. The results indicated that microtopography of the grassy land surface was well-characterized within each subwatershed in comparison with field observations. With the terrestrial light detection and ranging (LIDAR) point cloud of 15-mm point spacing and the TIN method, surface depression storage depths of the five subwatersheds ranged from 1.73 to 14.28 mm in the study area. The surface depression storage depth, as expected, increased with the maximum depth of surface depression. It was also found to increase when the land surface slope became milder. A sensitivity analysis indicated that a point cloud with a point spacing of 30 mm was sufficient to obtain an accurate representation of the terrain surface in the study area. This study also indicated the TIN method can represent the ground surface and the surface depression more realistically than the commonly used digital elevation model (DEM) method due to the TIN method’s higher capability of identifying and filtering out surface obstructions such as blades of grass. Moreover, by using the high-resolution TLS technology and the TIN method, our study provides an important and broad range of reference data on the surface depression storage depth commonly needed in application of the Storm Water Management Model (SWMM) and other watershed models.

A semantic segmentation method for vehicle‐borne laser scanning point clouds in motorway scenes

AbstractVehicle‐borne laser scanning (VLS) point cloud semantic segmentation is one of the fundamental issues in motorway target extraction. In this study, a uniform sampling‐based neural network (NN) is constructed based on the popular RandLA‐Net. The uniform sampling can ensure high efficiency and stable spatial coverage. In the proposed NN, a local extra feature encoding structure is designed, and it is combined with local geometric spatial encoding and attention mechanisms to enhance the distinctiveness of depth features and improve the geometric detail preservation ability of the network. The VLS point cloud dataset of motorway scenes collected in this study is labelled for network training and evaluation. Experimental results demonstrate the superiority of the proposed network over the state‐of‐the‐art networks, attaining the highest intersection‐over‐union (IoU) on the majority of categories and a 4.39% improvement in mean IoU on all categories. For the pole‐shaped targets, the improvement in IoU is 9.08% relative to those of the compared methods. The labelled VLS point cloud dataset generated in this work will be made publicly available.

SCL-GCN: Stratified Contrastive Learning Graph Convolution Network for pavement crack detection from mobile LiDAR point clouds

Accurate pavement crack detection is important for routine maintenance of pavements and reduction of possible traffic accidents. Most existing rule- or learning-based point-level approaches cannot achieve high detection accuracy and efficiency owing to the disorderly arrangement, scattered intensities, diverse crack structures, large data volumes, and complex annotation of mobile laser scanning (MLS) point clouds. To address these issues, we developed SCL-GCN, a Stratified Contrastive Learning Graph Convolution Network with a novel dual-branch architecture for MLS-point cloud-based pavement crack detection. First, a multi-scale graph representation construction module was designed based on a stratification strategy. This module creates strengthened spaces for the raw pavement point cloud and its downsampled subset, from which adjacency matrices and initial representations are generated. The stratification strategy samples neighbors densely in the raw point clouds and sparsely in the downsampled subset to form the neighborhood for each point, utilizing long-range contexts to increase the effective receptive field while lowing the extra computation. Next, a graph feature contrastive learning module is proposed to take advantage of stratified features. This module supervises the learning process of the two branches to avoid learning bias caused by an imbalanced data distribution, promoting convergence and improving performance. The experimental results show that the developed SCL-GCN model outperforms state-of-the-art methods. With a training/testing ratio of only 1:6 and an overall training time of less than 70 min, the average precision, recall, and F1-score of the SCL-GCN reached 75.7%, 75.1%, and 75.2%, respectively.

A new approach to detect surface defects from 3D point cloud data with surface normal Gabor filter (SNGF)

Surface defect detection is essential feedback for quality control in manufacturing processes. This paper presents a new defect detection method, Surface Normal Gabor Filter (SNGF), to detect surface defects using laser scanning point cloud data. The SNGF method transfers the 3D point cloud data to surface normal vectors to normalize the surface topology geometry. The surface normal vectors are then converted into complex numbers and processed by a Gabor Filter to extract defect-induced geometric features. The feasibility of SNGF is validated in the case of studies for detecting different types of defects on textured surfaces through simulations and experiments. The experimental results show improved stability in detecting defects with different sizes compared to the conventional Region Growing Segmentation Algorithm (RGSA). In addition, test results show the running time of SNGF methods is up to 138 times shorter than the RGSA when processing point cloud data sets with a range of 19,600∼313,600 data points.

Robust Guardrail Instantiation and Trajectory Optimization of Complex Highways Based on Mobile Laser Scanning Point Clouds

As a basic asset of highways, guardrails are essential objects in the digital modeling of highways. Therefore, generating the vectorial 3D trajectory of a guardrail from mobile laser scanning (MLS) point clouds is required for real digital modeling. However, most methods limit straight-line guardrails without considering the continuity and accuracy of the guardrails in turnoff and bend areas; thus, a completed 3D trajectory of a guardrail is not available. We use RANDLA-Net for extracting guardrails as preprocessing of MLS point clouds. We perform a region growth strategy based on linear constraints to obtain correct instantiations and a forward direction. The improved Douglas– Puke algorithm is used to simplify the center points of guardrail, and the 3D trajectory of every guardrail can be vectorized using cubic spline curve fitting. The proposed approach is validated on two 3-km case data sets that can completely instantiate MLS point clouds with remarkable effects. Quantitative evaluations demonstrate that the proposed guardrail instantiation algorithm achieves an overall precision and recall of 98.80% and 97.5%, respectively. The generated 3D trajectory can provide a high-precision design standard for the 3D modeling of the guardrail and has been applied to a long highway scene.