3D Point Cloud Data Research Articles

Application of UAV 3D Point Cloud Data in Highway Slope Disaster Investigation

The safety and stability of highway slopes have always been critical concerns in highway management and maintenance. However, mountain road slopes, characterized by their small scale, large quantity, and complex geological conditions, pose challenges for effective deformation detection, putting significant pressure on highway management and maintenance. To address the difficulty in identifying hidden hazards of highway slopes in mountainous areas, this study proposes the use of unmanned aerial vehicle (UAV) closed-range photogrammetry technology for slope inspection. The M3C2 (Multiscale Model-to-Model Cloud Comparison) algorithm is employed to accurately calculate differences in multi-phase three-dimensional point cloud data, enabling the detection of overall slope deformation with centimeter-level precision. Field verification conducted on a highway slope in Qingtian County, Zhejiang Province, demonstrated the applicability of UAV 3D point cloud data in investigating highway slope hazards, thereby enhancing inspection efficiency and quality and providing technical support for highway management and maintenance.

A slope management method by monitoring soil volumetric water content and its validation through numerical analysis

Landslides induced by heavy rainfall have become increasingly common in recent years. As the groundwater level rises, the safety factor of slopes tends to decrease. However, traditional monitoring methods, such as employing water-level (pressure) gauges in boreholes, require significant effort. Thus, this study proposes a slope management approach based on monitoring the volumetric water content of the ground surface. The monitored slope, approximately 20 m in height and 30 m in length, consists of topsoil mixed with grass roots and sand with a slight silt admixture, as revealed by two borehole investigations. Soil moisture meters were strategically installed at three locations along the slope (upper, middle, and lower sections) to gauge the volumetric water content. Observations indicated a sensitive increase in ground water content following rainfall. Furthermore, a three-dimensional Finite Element Method (FEM) analysis was employed to explore the correlation between rainfall (intensity and duration) and groundwater content. The FEM analysis also examined the impact of rainfall-induced changes in slope water content on the overall slope safety factor, utilizing 3D point cloud data to model the monitored slope. Qualitative agreement was observed between the monitoring and analysis results, highlighting a decrease in slope safety factor as slope water content increased. These findings underscore the potential of conveniently assessing slope safety factors based on the water content of a slope’s ground surface.

A Multi-Level Robust Positioning Method for Three-Dimensional Ground Penetrating Radar (3D GPR) Road Underground Imaging in Dense Urban Areas

Three-Dimensional Ground Penetrating Radar (3D GPR) detects subsurface targets non-destructively, rapidly, and continuously. The complex environment around urban roads affects the positioning accuracy of 3D GPR. The positioning accuracy directly affects the data quality, as inaccurate positioning can lead to distortion and misalignment of 3D GPR data. This paper proposed a multi-level robust positioning method to improve the positioning accuracy of 3D GPR in dense urban areas in order to obtain more accurate underground data. In environments with good GNSS signals, fast and high-precision positioning can be achieved based on GNSS data using differential GNSS technology; in scenes with weak GNSS signals, high-precision positioning of subsurface data can be achieved by using GNSS and IMU as well as using GNSS/INS tightly coupled solution technology; in scenes with no GNSS signals, SLAM technology is used for positioning based on INS data and 3D point cloud data. In summary, this method ensures a positioning accuracy of 3D GPR better than 10 cm and high-quality 3D images of underground urban roads in any environment. This provides data support for urban road underground structure surveys and has broad application prospects in underground disease detection and prevention.

Research on Deformation Monitoring of Roadway Surrounding Rock Based on Mobile 3D Laser Scanning Technology

Aiming at the problems of low data acquisition efficiency, unintuitive data form, low relative accuracy, and inability to continuously monitor the whole cross-section in traditional roadway rock deformation monitoring methods, a roadway rock deformation monitoring method based on mobile 3D laser scanning technology is proposed. Firstly, the mobile 3D laser scanning technology is used to obtain the real 3D point cloud data of the roadway surrounding rock; secondly, the multi-scale dimensional features are used to classify the 3D point cloud data, filter the invalid point cloud data, and form the accurate point cloud data of the roadway surrounding rock surface; then the 3D laser scanning is carried out on a regular basis for the deformation-prone areas of the roadway, and the Iterative Closest Point Algorithm (ICP algorithm) is used to align the overlapping part of the point cloud to achieve the point cloud splicing and point cloud stitching of the two scans. The point cloud splicing and point cloud overlapping of the two scans are achieved; finally, the multi-scale model-to-model point cloud comparison algorithm (M3C2 algorithm) is used to compare the point cloud data of the two periods of the roadway, and the appropriate parameters are selected to indicate the displacement change of the surface of the roadway rock with the size of the color value, so as to realize the monitoring of the deformation of the roadway rock in the three-dimensional full cross-section of the roadway. The method is applied to the deformation monitoring of the surrounding rock in the return roadway of 21407 working face in a coal mine, and the results show that: the mobile laser scanning technology can be used to obtain the complete point cloud data of the roadway, and the deformation area of the surrounding rock is successfully identified after the classification of the scanned point clouds in different time periods, the alignment of the point clouds, and the comparison of the point clouds, and the deformation condition of the surrounding rock of the roadway is visually reflected in the three-dimensional full-section deformation condition of the surrounding rock of the roadway. This method is suitable for monitoring the overall de-formation of the surrounding rock of the roadway under complex environmental conditions, and the de-formation of the surrounding rock is identified by different color areas.

VAE-GAN3D: Leveraging image-based semantics for 3D zero-shot recognition

The current state of 3D zero-shot recognition falls short in performance when compared to its counterpart in 2D images. A major challenge is the absence of a robust feature extractor for 3D point cloud data. To overcome this challenge, a large dataset needs to be collected, processed, and fed into a deep-learning model capable of generating distinguishable features after training. To this end, we propose VAE-GAN3D, a model that uses a combination of a variational autoencoder (VAE) and a generative adversarial network (GAN) to supplement the small amount of data available for training with synthetic examples of classes. This allows for the underlying data distribution of large and complex data to be learned by the deep VAE network. When combined with a GAN, this network generates synthetic features that exhibit consistency in both seen and unseen classes. Furthermore, we notice that for tasks involving a small domain of classes, the existing text features do not contribute significantly to zero-shot learning. Therefore, we introduce image representation-based semantic features of classes, which improve the performance of zero-shot recognition for 3D objects. We assess the proposed model using three different datasets and present a technique for partitioning the RGB-D object dataset, which contains real-world objects, into seen and unseen classes. Our approach shows promising results in addressing the challenges of 3D zero-shot recognition and presents a novel solution for improving the accuracy of 3D point cloud recognition.

ACCURACY ASSESSMENT OF A SLAM-ACQUIRED POINT CLOUD DATA USING A VARIETY OF CLASSIFICATION APPLICATIONS

Abstract. Laser scanning techniques, such as Simultaneous Localization and Mapping (SLAM), produce three-dimensional data representing the real world, which may provide significant information for Building Information Models (BIM). These processes produce 3D point clouds, which require classification before being used in various applications such as structural assessments. However, most widely available software applications for classifying 3D point clouds are proprietary, giving an incomplete depiction of how the data is manipulated and processed. Thus, this research aims to assess the accuracy of the different classification applications in classifying the 3D point cloud data and perform a comparative analysis of the results. Precision, Recall, F1-score, and Accuracy are the evaluation metrics used to assess the classified 3D point cloud data. Results for Precision and Recall show that some of the applications can classify a particular class, the Ground and Building classes. However, the overall performance of the classification method, which is evaluated through the F1-score, produced low values. Results for the F1-score demonstrate that these low values indicate low overall reliability of the classification results despite high values for Accuracy. Based on the conducted experiments, further research is suggested to investigate the effect of increasing dataset size and equalizing class sizes used in classification.

3D Point Cloud Dataset of Heavy Construction Equipment

Object recognition algorithms and datasets based on point cloud data have been mainly designed for autonomous vehicles. When applied to the construction industry, they face challenges due to the origin of point cloud data from large earthwork sites, resulting in high volumes of data and density. This research prioritized the development of 3D point cloud datasets specifically for heavy construction equipment, including dump trucks, rollers, graders, excavators, and dozers; all of which are extensively used in earthwork sites. The aim was to enhance the efficiency and productivity of machine learning (ML) and deep learning (DL) research that relies on 3D point cloud data in the construction industry. Notably, unlike conventional approaches to acquiring point cloud data using UAVs (Unmanned Aerial Vehicles) and UGVs (Unmanned Ground Vehicles), the datasets for the five types of heavy construction equipment established in this research were generated using 3D-scanned diecast models of heavy construction equipment to create point cloud data.

RETROFIT: Real-Time Control of Time-Dependent 3D Point Cloud Profiles

Abstract In modern manufacturing processes, ensuring the precision of 3D profiles of products is crucial. Nonetheless, achieving this accuracy is challenging due to the complex interactions between process inputs and the data structure of the 3D profile data. Our solution, a 3D profile-based control framework, addresses this challenge by actively adapting and controlling the manufacturing process to enhance 3D shape accuracy. 3D profile scans represent the ultimate measure of desired part quality. Therefore, utilizing them as the system responses for control purposes yields the most direct and effective feedback. We leverage recent advancements from Koopman operator theory to create an effective model-based control strategy. Initially, we estimate the process model by exploring the relationship between 3D profiles and heterogeneous process inputs. Then, we formulate an online model predictive control law. Challenges include dealing with unstructured, high-dimensional 3D point cloud data, capturing spatial and temporal structures, and integrating heterogeneous, high-dimensional process input data into the control model. To overcome these challenges, we introduce RETROFIT, a solution designed for the real-time control of time-dependent 3D point cloud profiles. Unlike traditional models, RETROFIT is not bound by linear assumptions and can handle unstructured 3D point cloud data directly. We demonstrate its effectiveness through a wire arc additive manufacturing case study, highlighting its potential to enhance 3D profile accuracy in manufacturing processes.

Automated Quality Inspection of Formwork Systems Using 3D Point Cloud Data

Ensuring that formwork systems are properly installed is essential for construction safety and quality. They have to comply with specific design requirements and meet strict tolerances regarding the installation of the different members. The current method of quality control during installation mostly relies on manual measuring tools and inspections heavily reliant on the human factor, which could lead to inconsistencies and inaccurate results. This study proposes a way to automate the inspection process and presents a framework within which to measure the spacing of the different members of the formwork system using 3D point cloud data. 3D point cloud data are preprocessed, processed, and analyzed with various techniques, including filtering, downsampling, transforming, fitting, and clustering. The novelty is not only in the integration of the different techniques used but also in the detection and measurement of key members in the formwork system with limited human intervention. The proposed framework was tested on a real construction site. Five cases were investigated to compare the proposed approach to the manual and traditional one. The results indicate that this approach is a promising solution and could potentially be an effective alternative to manual inspections for quality control during the installation of formwork systems.

Error Analysis of Stonex X300 Laser Scanner Close-range Measurements

This research reports an error analysis of close-range measurements from a Stonex X300 laser scanner in order to address range uncertainty behavior based on indoor experiments under fixed environmental conditions. The analysis includes procedures for estimating the precision and accuracy of the observational errors estimated from the Stonex X300 observations and conducted at intervals of 5 m within a range of 5 to 30 m. The laser 3D point cloud data of the individual scans is analyzed following a roughness analysis prior to the implementation of a Levenberg–Marquardt iterative closest points (LM-ICP) registration. This leads to identifying the level of roughness that was encountered due to the range-finder’s limitations in close-ranging as well as measurements that were obtained from extreme incident angle signals. The measurements were processed using a statistical outlier removal (SOR) filter to reduce the noise impact toward a smoother data set. The geometric differences and the RMSE values in the 3D coordinate directions were computed and analyzed, which showed the potential of the Stonex X300 measurements in close-ranging following a careful statistical analysis. It was found that the error differences in the vertical direction had a consistent behavior when the range increased, whereas the errors in the horizontal direction varied. However, it is more common to produce errors in the vertical direction as compared to the horizontal one.

Automated Derivation of Vine Objects and Ecosystem Structures Using UAS-Based Data Acquisition, 3D Point Cloud Analysis, and OBIA

This study delves into the analysis of a vineyard in Carinthia, Austria, focusing on the automated derivation of ecosystem structures of individual vine parameters, including vine heights, leaf area index (LAI), leaf surface area (LSA), and the geographic positioning of single plants. For the derivation of these parameters, intricate segmentation processes and nuanced UAS-based data acquisition techniques are necessary. The detection of single vines was based on 3D point cloud data, generated at a phenological stage in which the plants were in the absence of foliage. The mean distance from derived vine locations to reference measurements taken with a GNSS device was 10.7 cm, with a root mean square error (RMSE) of 1.07. Vine height derivation from a normalized digital surface model (nDSM) using photogrammetric data showcased a strong correlation (R2 = 0.83) with real-world measurements. Vines underwent automated classification through an object-based image analysis (OBIA) framework. This process enabled the computation of ecosystem structures at the individual plant level post-segmentation. Consequently, it delivered comprehensive canopy characteristics rapidly, surpassing the speed of manual measurements. With the use of uncrewed aerial systems (UAS) equipped with optical sensors, dense 3D point clouds were computed for the derivation of canopy-related ecosystem structures of vines. While LAI and LSA computations await validation, they underscore the technical feasibility of obtaining precise geometric and morphological datasets from UAS-collected data paired with 3D point cloud analysis and object-based image analysis.

Dynamic Downsampling Algorithm for 3D Point Cloud Map Based on Voxel Filtering

In response to the challenge of handling large-scale 3D point cloud data, downsampling is a common approach, yet it often leads to the problem of feature loss. We present a dynamic downsampling algorithm for 3D point cloud maps based on an improved voxel filtering approach. The algorithm consists of two modules, namely, dynamic downsampling and point cloud edge extraction. The former adapts voxel downsampling according to the features of the point cloud, while the latter preserves edge information within the 3D point cloud map. Comparative experiments with voxel downsampling, grid downsampling, clustering-based downsampling, random downsampling, uniform downsampling, and farthest-point downsampling were conducted. The proposed algorithm exhibited favorable downsampling simplification results, with a processing time of 0.01289 s and a simplification rate of 91.89%. Additionally, it demonstrated faster downsampling speed and showcased improved overall performance. This enhancement not only benefits productivity but also highlights the system’s efficiency and effectiveness.

Estimation of Weight and Body Measurement Model for Pigs Based on Back Point Cloud Data.

Pig farming is a crucial sector in global animal husbandry. The weight and body dimension data of pigs reflect their growth and development status, serving as vital metrics for assessing their progress. Presently, pig weight and body dimensions are predominantly measured manually, which poses challenges such as difficulties in herding, stress responses in pigs, and the control of zoonotic diseases. To address these issues, this study proposes a non-contact weight estimation and body measurement model based on point cloud data from pig backs. A depth camera was installed above a weighbridge to acquire 3D point cloud data from 258 Yorkshire-Landrace crossbred sows. We selected 200 Yorkshire-Landrace sows as the research subjects and applied point cloud filtering and denoising techniques to their three-dimensional point cloud data. Subsequently, a K-means clustering segmentation algorithm was employed to extract the point cloud corresponding to the pigs' backs. A convolutional neural network with a multi-head attention was established for pig weight prediction and added RGB information as an additional feature. During the data processing process, we also measured the back body size information of the pigs. During the model evaluation, 58 Yorkshire-Landrace sows were specifically selected for experimental assessment. Compared to manual measurements, the weight estimation exhibited an average absolute error of 11.552 kg, average relative error of 4.812%, and root mean square error of 11.181 kg. Specifically, for the MACNN, incorporating RGB information as an additional feature resulted in a decrease of 2.469 kg in the RMSE, a decrease of 0.8% in the MAPE, and a decrease of 1.032 kg in the MAE. Measurements of shoulder width, abdominal width, and hip width yielded corresponding average relative errors of 3.144%, 3.798%, and 3.820%. In conclusion, a convolutional neural network with a multi-head attention was established for pig weight prediction, and incorporating RGB information as an additional feature method demonstrated accuracy and reliability for weight estimation and body dimension measurement.

Assessing the Performance of Handheld Laser Scanning for Individual Tree Mapping in an Urban Area

Precise individual tree or sample-based inventories derived from 3D point cloud data of mobile laser scanning can improve our comprehensive understanding of the structure, function, resilience, biodiversity, and ecosystem services of urban forests. This study assessed the performance of a handheld laser scanning system (HLS) for the extraction of tree position, diameter at breast height (DBH), and tree height (H) in an urban area. A total of 2083 trees of 13 species from 34 plots were analyzed. The results showed that the registration of tree positions using ground control points (GCPs) demonstrated high accuracy, with errors consistently below 0.4 m, except for a few instances. The extraction accuracy of DBH for all trees and individual species remained consistently high, with a total root mean square error (RMSE) of 2.06 cm (6.89%) and a bias of 0.62 cm (2.07%). Notably, broad-leaved trees outperformed coniferous trees, with RMSE and bias values of 1.86 cm (6%) and 0.76 cm (2.46%), respectively, compared to 2.54 cm (9.46%) and 0.23 cm (0.84%), respectively. The accuracy of H extraction varied significantly among different species, with R2 values ranging from 0.65 to 0.92. Generally, both DBH and H were underestimated compared to ground measurements. Linear mixed-effects models (LMEs) were applied to evaluate factors affecting the performance of HLS with the plot as a random factor. LME analysis revealed that plant type and terrain significantly influenced the accuracy of DBH and H derived from HLS data, while other fixed factors such as plot area, tree density, and trajectory length showed no significance. With a large sample size, we concluded that the HLS demonstrated sufficient accuracy in extracting individual tree parameters in urban forests.

Leveraging local neighborhood features in 3D unstructured point cloud data for geometry-based automated damage delineation

In a post-earthquake environment, manual inspection is resource-intensive and subjective. This article presents a real-world damage assessment study of the Improved Iterative Bi-Level Thresholding (IIBLT) algorithm. Damaged regions are delineated using IIBLT based on the local point curvature. The methodology is tested using a unique dataset: Mono Temporal Disaster City 3D Dataset (C3DO). C3DO is a point cloud data generated by a 3D reconstruction of a controlled post-earthquake environment, which presents challenges for concrete façade damage assessment algorithms. Results from IIBLT demonstrated a 16% increase in Matthews Correlation Coefficient (MCC) and Intersection over Union (mGIoU), compared to the current state-of-the-art methods. This corresponds to improving accuracy and completeness in defect region identification without the dependence on knowledge of damaged regions. Results from this study highlight the novel contributions of the IIBLT algorithm including the elimination of parameter tuning, and demonstrate the applicability of the novel iterative framework for concrete surface defect delineation at large.

An efficient algorithm with fast convergence rate for sparse graph signal reconstruction

In this paper, we consider the graph signals are sparse in the graph Fourier domain and propose an iterative threshold compressed sensing reconstruction (ITCSR) algorithm to reconstruct sparse graph signals in the graph Fourier domain. The proposed ITCSR algorithm derives from the well-known compressed sensing by considering a threshold for sparsity-promoting reconstruction of the underlying graph signals. The proposed ITCSR algorithm enhances the performance of sparse graph signal reconstruction by introducing a threshold function to determine a suitable threshold. Furthermore, we demonstrate that the suitable parameters for the threshold can be automatically determined by leveraging the sparrow search algorithm. Moreover, we analytically prove the convergence property of the proposed ITCSR algorithm. In the experimental, numerical tests with synthetic as well as 3D point cloud data demonstrate the merits of the proposed ITCSR algorithm relative to the baseline algorithms.

Digital Restoration of Historical Buildings by Integrating 3D PC Reconstruction and GAN Algorithm

Historical architecture is an important carrier of cultural and historical heritage in a country and region, and its protection and restoration work plays a crucial role in the inheritance of cultural heritage. However, the damage and destruction of buildings urgently need to be repaired due to the ancient age of historical buildings and the influence of natural environment and human factors. Therefore, an artificial intelligence repair technology based on three-dimensional (3D) point cloud reconstruction and generative adversarial networks was proposed to improve the precision and efficiency of repair work. Firstly, in-depth research on the principles and algorithms of 3D point cloud data processing and generative adversarial networks should be conducted. Secondly, a digital restoration framework was constructed by combining these two artificial intelligence technologies to achieve precise and efficient restoration of historical buildings through continuous adversarial learning processes. The experimental results showed that the errors in the restoration of palace buildings, defense walls, pagodas, altars, temples, and mausoleums were 0.17, 0.12, 0.13, 0.11, and 0.09, respectively. The technique can significantly reduce the error while maintaining the high precision repair effect. This technology with artificial intelligence as the core has excellent accuracy and stability in the digital restoration. It provides a new technical means for the digital restoration of historical buildings and has important practical significance for the protection of cultural heritage.

RANDOMISED COMPRESSION RATIOS FOR EFFECTIVE LARGE POINT CLOUD PROCESSING USING COMPRESSIVE SENSING

Abstract. Effectively navigating the intricacies of extensive 3D point cloud data in urban environments poses a series of formidable computational challenges. These challenges are primarily attributed to the substantial data volume and density inherent in urban settings, the presence of noise and inconsistencies within the collected data, and the constraints imposed by limited transmission bandwidth, which consequently impact storage requirements. This paper introduces an innovative methodology for handling large point cloud datasets, based on concepts from Sparse Signal Processing (SSP), also known as compressive sensing. The proposed approach integrates well known geometric data manipulation such as the Octree to work hand in hand with SSP, as unified method. Through experimental validation using the Santiago Urban Dataset (SUD), we demonstrate the effectiveness of our method in achieving high data fidelity, as measured by Peak Signal-to-Noise Ratio (PSNR) values reaching approximately 60 dB even at substantial compression ratios. Comparative analysis against traditional methods, including those implemented in the widely used Point Cloud Library (PCL), reveals the superior performance of our proposed methodology. The results underscore the robustness and efficiency of our approach, positioning it as a compelling alternative for compressing extensive 3D point cloud data. This has crucial implications for diverse applications, ranging from city planning to rapid and effective disaster response.

A New Approach for Discontinuity Extraction Based on an Improved Naive Bayes Classifier

An increasing number of methods are being used to extract rock discontinuities from 3D point cloud data of rock surfaces. In this paper, a new method for automatic extraction of rock discontinuity based on an improved Naive Bayes classifier is proposed. The method first uses principal component analysis to find the normal vectors of the points, and then generates a certain number of random point sets around the selected training points for training the classifier. The trained, improved Naive Bayes classifier is based on point normal vectors and is able to automatically remove noise points due to various reasons in conjunction with the knee point algorithm, realizing high-precision extraction of the discontinuity sets. Subsequently, the individual discontinuities are segmented using a hierarchical density-based spatial clustering method with noise application. Finally, the PCA algorithm is used to complete the orientation by plane fitting the individual discontinuities. The method was applied in two cases, Kingston and Colorado, and the reliability and advantages of the new method were verified by comparing the results with those of previous research, and the discussion and analysis determined the optimal values of the relevant parameters in the algorithm.

RPEA: A Residual Path Network with Efficient Attention for 3D pedestrian detection from LiDAR point clouds

Efficiently detecting pedestrians from 3D point cloud data is a significantly challenging perception task in numerous robotic and autonomous driving applications, primarily because of the sparsity of point cloud data representing pedestrian objects and the significant deformations in pedestrian body poses. To address these challenges, we present a Residual Path network with Efficient Attention (RPEA), an end-to-end trainable single-stage 3D pedestrian detection network. We first introduce the Residual Path (ResPath) architecture, which incorporates multiple residual blocks to retain the spatial information lost during downsampling and combines features of different resolution scales. To suppress noise in point clouds while generalizing various pedestrian representations, we propose an efficient Channel Attention module with Average and Maximum pooling strategies (CAAM). Experimental results demonstrate that our RPEA ranks first on the JRDB 3D object detection leaderboard among all solutions, while significantly surpassing the ranks second by up to 5.6% average precision. Additionally, our RPEA achieves real-time pedestrian detection at 39 frame-per-second (fps). Since our method has higher accuracy and faster inference, it can be deployed more effectively in vehicles and mobile robots. The code is publicly available at https://github.com/jinzhengguang/RPEA.