Point Cloud Processing Research Articles

Terrain Traversability via Sensed Data for Robots Operating Inside Heterogeneous, Highly Unstructured Spaces

This paper presents a comprehensive approach to evaluating the ability of multi-legged robots to traverse confined and geometrically complex unstructured environments. The proposed approach utilizes advanced point cloud processing techniques integrating voxel-filtered cloud, boundary and mesh generation, and dynamic traversability analysis to enhance the robot’s terrain perception and navigation. The proposed framework was validated through rigorous simulation and experimental testing with humanoid robots, showcasing the potential of the proposed approach for use in applications/environments characterized by complex environmental features (navigation inside collapsed buildings). The results demonstrate that the proposed framework provides the robot with an enhanced capability to perceive and interpret its environment and adapt to dynamic environment changes. This paper contributes to the advancement of robotic navigation and path-planning systems by providing a scalable and efficient framework for environment analysis. The integration of various point cloud processing techniques into a single architecture not only improves computational efficiency but also enhances the robot’s interaction with its environment, making it more capable of operating in complex, hazardous, unstructured settings.

Research on Digital Terrain Construction Based on IMU and LiDAR Fusion Perception

To address the shortcomings of light detection and ranging (LiDAR) sensors in extracting road surface elevation information in front of a vehicle, a scheme for digital terrain construction based on the fusion of an Inertial Measurement Unit (IMU) and LiDAR perception is proposed. First, two sets of sensor coordinate systems were configured, and the parameters of LiDAR and IMU were calibrated. Then, a terrain construction system based on the fusion perception of IMU and LiDAR was established, and improvements were made to the state estimation and mapping architecture. Terrain construction experiments were conducted in an academic setting. Finally, based on the output information from the terrain construction system, a moving average-like algorithm was designed to process point cloud data and extract the road surface elevation information at the vehicle’s trajectory position. By comparing the extraction effects of four different sliding window widths, the 4 cm width sliding window, which yielded the best results, was ultimately selected, making the extracted road surface elevation information more accurate and effective.

Point cloud classification and part segmentation of steel structure elements

AbstractThe classification and part segmentation of point clouds have gained significant attention in the field of artificial intelligence (AI), especially in the construction industry. However, addressing the dataset directly in AI models remains a challenge, as most existing methods are not well-suited for processing point cloud data. PointNet has emerged as an AI architecture algorithm. It transforms individual points independently to learn local and global features. This research aims to develop a comprehensive framework for classification and part segmentation for point clouds of steel structure elements. The framework enhances the accuracy of the PointNet algorithm, and it consists of three stages: (1) dataset creation; (2) model classification; and (3) part segmentation. First, the dataset creation procedure encompasses modeling steel columns, beams, and braces using Autodesk Revit software. For the classification dataset, a dataset comprising 580 columns and 920 beams is obtained. In the part segmentation dataset, five categories of steel braced frame elements are generated, yielding a total of 21,870 elements for braced frame structures. Several point cloud experiments have been applied, including adjusting the number of points in the point cloud, altering the batch size, and fine-tuning the number of epochs. These experimental settings were systematically investigated to identify the optimal combination that yields the highest (AI) model accuracy. PointNet model achieved 100% accuracy across all classification experiments, while part segmentation experiments reached up to 97.10% accuracy, with a mean intersection over union (MIOU) of 93.70%. The comprehensive analysis of the point cloud dataset is applied on an actual case study to demonstrate the practical features of the proposed research.

Synergistic image and point cloud processing of UAV data for urban flood modeling: point cloud smart thinning and curb mapping

Abstract. We propose an integrated approach for automatic point cloud thinning and curb mapping in Uncrewed Aerial Vehicle - Structure from Motion (UAV-SfM) point clouds to enhance hydrological modeling in flood-prone urban areas. UAV flights were conducted to generate an initial orthoimage, which was used to train a convolutional neural network (CNN) segmentation model. The trained model was then applied to the UAV images to produce two binary mask sets: one for vegetation and one for streets and sidewalks. These masks were incorporated during photogrammetric 3D reconstruction to estimate camera geometry and generate a dense point cloud. Our results show that vegetation masks did not improve camera geometry estimation. However, by applying UAV masks, we achieved a 15% reduction in total processing time and decreased the number of points by a factor of 2.7. This targeted approach enabled curb detection by focusing on expected curb locations. Curb candidate points were proposed using geometric characteristics of the point cloud, including normal values, linearity, and verticality. Our rule-based method effectively mapped even subtle curb features, providing a rapid, cost-effective solution for large-area curb mapping. Further, we explored the potential of random forest for curb mapping, with promising results. Our approach can support urban flood modeling efforts and strengthen urban resilience for flood-prone communities.

Semantic Enrichment of Architectural Heritage Point Clouds Using Artificial Intelligence: The Palacio de Sástago in Zaragoza, Spain

In the current landscape dominated by Artificial Intelligence, the integration of Machine Learning and Deep Learning within the realm of Cultural Heritage, particularly within architectural contexts, is paramount for the efficient processing and interpretation of point clouds. These advanced methods facilitate automated segmentation and classification, significantly improving both the clarity and practical use of data acquired from laser scanning and photogrammetry. The present study investigates the Palacio de Sástago—a prominent Renaissance palace in Zaragoza, Spain—and introduces a cutting-edge modus operandi for the automated recognition of architectural elements within the palace’s inner courtyard. Employing the well-established Random Forest algorithm, implemented in a Python environment, the framework begins with a comprehensive evaluation of the geometric features identified in the LiDAR point cloud. This process employs the Mean Decrease in Impurity metric to evaluate the relevance of each variable. To boost the accuracy and efficiency of the final classifications, the features are refined post-assessment, enhancing both the training phase and the algorithm’s later evaluation. The research’s findings demonstrate significant potential, supporting advancements in CAD systems and HBIM that will enable more precise, automated modelling of architectural elements, thereby enhancing the accuracy of digital reconstructions and improving conservation planning for heritage sites.

Calculating the Optimal Point Cloud Density for Airborne LiDAR Landslide Investigation: An Adaptive Approach

Ensuring that ground point density after raw point cloud processing meets the accuracy requirements for subsequent DEM construction represents a challenge for field operators during airborne LiDAR data acquisition. In this study, we propose a method to quantify DEM quality by combining the RMSE of elevation and terrain complexity, analyzing the DEM quality error curves constructed with different point cloud densities by a discrete difference peak-seeking method, to determine the optimal ground point density, and then constructing an ICP-NN algorithm for predicting the collected point cloud density. After analysis of DEM quality at eight point cloud dilution levels, the optimal ground point cloud densities were determined to be 2.43 pts/m2 (0.2 m resolution), 2.08 pts/m2 (1 m and 0.5 m resolution), and 1.84 pts/m2 (2 m resolution). Using the obtained optimal ground point densities, survey area slopes, canopy density, and elevation differences as eigenvalues, the ICP-NN model can be used to directly predict the collected point cloud density intervals in other regions, and the model has interval lengths ranging from 36 to 70.33 pts/m2 at 5 CLs. This method solves the problem of determining point cloud density in landslide surveys using airborne LiDAR and provides direct guidance for practical applications.

4th EuroSDR Workshop on Point Cloud Processing

4th EuroSDR Workshop on Point Cloud Processing

A Survey of Label-Efficient Deep Learning for 3D Point Clouds.

In the past decade, deep neural networks have achieved significant progress in point cloud learning. However, collecting large-scale precisely-annotated point clouds is extremely laborious and expensive, which hinders the scalability of existing point cloud datasets and poses a bottleneck for efficient exploration of point cloud data in various tasks and applications. Label-efficient learning offers a promising solution by enabling effective deep network training with much-reduced annotation efforts. This paper presents the first comprehensive survey of label-efficient learning of point clouds. We address three critical questions in this emerging research field: i) the importance and urgency of label-efficient learning in point cloud processing, ii) the subfields it encompasses, and iii) the progress achieved in this area. To this end, we propose a taxonomy that organizes label-efficient learning methods based on the data prerequisites provided by different types of labels. We categorize four typical label-efficient learning approaches that significantly reduce point cloud annotation efforts: data augmentation, domain transfer learning, weakly-supervised learning, and pretrained foundation models. For each approach, we outline the problem setup and provide an extensive literature review that showcases relevant progress and challenges. Finally, we share our views on the current research challenges and potential future directions.

Application of Mobile Mapping System for a Modern Topography

Abstract The mobile 3D mapping systems (MMS) and the integration with Geographical Information Systems (GIS) represent an innovative and efficient approach to modern topography. It is recommended and useful for infrastructure analysis and urban planning. This modern technology provides accurate, fast, and detailed data, facilitating well-informed decision-making and contributing to the development of sustainable infrastructure in the 21st century. The purpose of this research was to use a mobile mapping system for land surveying measurement of an urban area, independent of pre-established ground control points. On the other hand, the research is supposed to demonstrate the usefulness of kinematic trajectory, digital imaging, and laser scanning in acquiring, storing, and processing point cloud data. This paper shows an affordable method and a modern scanning instrument used for mobile mapping scanning, to create a georeferenced point cloud (PC). The obtained result has centimetric precision of georeferenced point cloud data, even if the average speed of the car was about 30km/h. The trajectory adjustment with static GNSS measurements increases the accuracy of ground control points (GCPs) up to ±5cm for horizontal positioning and vertical positioning. Although this study was carried out while the system was mounted on a car the result followed the national accuracy requirements for land surveying and cadastre fields. In other words, the mobile mapping system is a faster and cost-effective in time alternative to static laser scanning and it is important to continue to explore and adopt these technological solutions to face the future challenges in the field of urban development and infrastructure.

Development of Automated 3D LiDAR System for Dimensional Quality Inspection of Prefabricated Concrete Elements.

The dimensional quality inspection of prefabricated concrete (PC) elements is crucial for ensuring overall assembly quality and enhancing on-site construction efficiency. However, current practices remain heavily reliant on manual inspection, which results in high operator dependency and low efficiency. Existing Light Detection and Ranging (LiDAR)-based methods also require skilled professionals for scanning and subsequent point cloud processing, thereby presenting technical challenges. This study developed a 3D LiDAR system for the automatic identification and measurement of the dimensional quality of PC elements. The system consists of (1) a hardware system integrated with camera and LiDAR components to acquire 3D point cloud data and (2) a user-friendly graphical user interface (GUI) software system incorporating a series of algorithms for automated point cloud processing using PyQt5. Field experiments comparing the system's measurements with manual measurements on prefabricated bridge columns demonstrated that the system's average measurement error was approximately 5 mm. The developed system can provide a quick, accurate, and automated inspection tool for dimensional quality assessment of PC elements, thereby enhancing on-site construction efficiency.

Design and analysis of terrestrial laser scanner based on a 3-SPR parallel mechanism for improved anti-occlusion scanning

Abstract Laser scanner technology swiftly captures point cloud data of objects and their surrounding environments, proving extensive applications across various sectors. However, it often encounters challenges related to incomplete point clouds due to occlusion from stationary objects. This paper presents a terrestrial laser scanning system based on a 3-SPR (3-Spherical Joint-Active Prismatic Joint-Rotating Joint) parallel mechanism (TLS-PM), specifically designed to enhance scanning coverage during single-station measurements, reduce positioning and workload during multi-station measurements, and mitigate point cloud gaps caused by occlusions.
Initially, a simulation model of the TLS-PM was developed, and both forward and inverse kinematic analysis were performed. Subsequently, the workspace was computed for different spherical joints using this model. An introduction to the TLS-PM’s error and the registration algorithm employed was then provided. Finally, through comparative analysis of simulations and experimental results, the device's measurement accuracy and its capability to resist occlusions were validated. Additionally, the TLS-PM’s anti-occlusion performance was evaluated under various scenarios in a simulated setting. The experimental results demonstrate that, when employing the same conventional point cloud processing algorithms, the TLS-PM significantly improves the background scanning coverage.

Intelligent Inspection Method for Rebar Installation Quality of Reinforced Concrete Slab Based on Point Cloud Processing and Semantic Segmentation

The rebar installation quality significantly impacts the safety and durability of reinforced concrete (RC) structures. Traditional manual inspection is time-consuming, inefficient, and highly subjective. In order to solve this problem, this study uses a depth camera and aims to develop an intelligent inspection method for the rebar installation quality of an RC slab. The Random Sample Consensus (RANSAC) method is used to extract point cloud data for the bottom formwork, the upper and lower rebar lattices, and individual rebars. These data are utilized to measure the concrete cover thickness, the distance between the upper and lower rebar lattices, and the spacing between rebars in the RC slab. This paper introduces the concept of the “diameter calculation region” and combines point cloud semantic information with rebar segmentation mask information through the relationship between pixel coordinates and camera coordinates to measure the nominal diameter of the rebar. The verification results indicate that the maximum deviations for the concrete cover thickness, the distance between the upper and lower rebar lattices, and the spacing of the double-layer bidirectional rebar in the RC slab are 0.41 mm, 1.32 mm, and 5 mm, respectively. The accuracy of the nominal rebar diameter measurement reaches 98.4%, demonstrating high precision and applicability for quality inspection during the actual construction stage. Overall, this study integrates computer vision into traditional civil engineering research, utilizing depth cameras to acquire point cloud data and color results. It replaces inefficient manual inspection methods with an intelligent and efficient approach, addressing the challenge of detecting double-layer reinforcement. This has significant implications for practical engineering applications and the development of intelligent engineering monitoring systems.

Advancements in Radar Point Cloud Processing for Macro Human Movements in Healthcare and Assisted Living Domains: A Review

Advancements in Radar Point Cloud Processing for Macro Human Movements in Healthcare and Assisted Living Domains: A Review

Advanced Point Cloud Techniques for Improved 3D Object Detection: A Study on DBSCAN, Attention, and Downsampling

To address the challenges of limited detection precision and insufficient segmentation of small to medium-sized objects in dynamic and complex scenarios, such as the dense intermingling of pedestrians, vehicles, and various obstacles in urban environments, we propose an enhanced methodology. Firstly, we integrated a point cloud processing module utilizing the DBSCAN clustering algorithm to effectively segment and extract critical features from the point cloud data. Secondly, we introduced a fusion attention mechanism that significantly improves the network’s capability to capture both global and local features, thereby enhancing object detection performance in complex environments. Finally, we incorporated a CSPNet downsampling module, which substantially boosts the network’s overall performance and processing speed while reducing computational costs through advanced feature map segmentation and fusion techniques. The proposed method was evaluated using the KITTI dataset. Under moderate difficulty, the BEV mAP for detecting cars, pedestrians, and cyclists achieved 87.74%, 55.07%, and 67.78%, reflecting improvements of 1.64%, 5.84%, and 5.53% over PointPillars. For 3D mAP, the detection accuracy for cars, pedestrians, and cyclists reached 77.90%, 49.22%, and 62.10%, with improvements of 2.91%, 5.69%, and 3.03% compared to PointPillars.

Remote Sensing and GIS in Natural Resource Management: Comparing Tools and Emphasizing the Importance of In-Situ Data

Remote sensing (RS) and Geographic Information Systems (GISs) provide significant opportunities for monitoring and managing natural resources across various temporal, spectral, and spatial resolutions. There is a critical need for natural resource managers to understand the expanding capabilities of image sources, analysis techniques, and in situ validation methods. This article reviews key image analysis tools in natural resource management, highlighting their unique strengths across diverse applications such as agriculture, forestry, water resources, soil management, and natural hazard monitoring. Google Earth Engine (GEE), a cloud-based platform introduced in 2010, stands out for its vast geospatial data catalog and scalability, making it ideal for global-scale analysis and algorithm development. ENVI, known for advanced multi- and hyperspectral image processing, excels in vegetation monitoring, environmental analysis, and feature extraction. ERDAS IMAGINE specializes in radar data analysis and LiDAR processing, offering robust classification and terrain analysis capabilities. Global Mapper is recognized for its versatility, supporting over 300 data formats and excelling in 3D visualization and point cloud processing, especially in UAV applications. eCognition leverages object-based image analysis (OBIA) to enhance classification accuracy by grouping pixels into meaningful objects, making it effective in environmental monitoring and urban planning. Lastly, QGIS integrates these remote sensing tools with powerful spatial analysis functions, supporting decision-making in sustainable resource management. Together, these tools when paired with in situ data provide comprehensive solutions for managing and analyzing natural resources across scales.

A Trajectory planning technique for grinding and polishing aerospace turbine blades’ thermal barrier coatings based on PCL point cloud processing

A Trajectory planning technique for grinding and polishing aerospace turbine blades’ thermal barrier coatings based on PCL point cloud processing

Associative graph convolution network for point cloud analysis

Since point cloud is the raw output of most 3D sensors, its effective analysis is in huge demand in the field of autonomous driving and robotic manipulation. However, directly processing point clouds is challenging because point clouds are a kind of disordered and unstructured geometric data. Recently, numerous graph convolution neural networks are proposed for introducing graph structure to point clouds yet far from perfect. Specially, DGCNN tries to learn local geometric of points in semantic space and recomputes the graph using nearest neighbors in the feature space in each layer. However, it discards all the information of the previous graph after each graph update, which neglects the relations between each dynamic update. To this end, we propose an associative graph convolution neural network (AGCN) which mainly consists of associative graph convolution (AGConv) and two kinds of residual connections. AGConv additionally considers the information from the previous graph when computing the edge function on current local neighborhoods in each layer, and it can precisely and continuously capture the local geometric features on point clouds. Residual connections further explore the semantic relations between layers for effective learning on point clouds. Extensive experiments on several benchmark datasets show that our network achieves competitive classification and segmentation results.

A novel computer vision and point cloud-based approach for accurate structural analysis of a tall irregular timber structure

Wood material has been widely used in critical heritage structures such as pagodas, totem poles, and large-scale sculptures. Conducting rigorous structural analysis is crucial to protect these structures in high-seismic regions. Typically, these types of structures are modeled using an equivalent finite element (FE) model which used simple cylinder with a constant cross section equal to the average of the top and bottom cross section. However, simplified equivalent FE models may not accurately consider the irregularities and complexities of these structures. In this paper, advanced computer vision and point cloud techniques were adopted to accurately and rapidly construct a FE model of a 30-meter irregular timber sculpture. This was achieved using video scans, computer vision-aided 3D reconstruction, point cloud processing, and mesh to solid element conversion. The refined FE model was used to conduct capacity check, mesh sensitivity study, pushover analysis, and response spectrum analysis. The results of the refined FE model were compared to an equivalent FE model. The results show: 1) the proposed numerical modeling methodology for structural analysis can efficiently and accurately measure the dimension of the irregular sculpture up to 98.2 % accuracy; 2) the lateral stiffnesses of the 30-meter irregular sculpture vary significantly (up to 42.6 %) from one direction to the other; 3) the equivalent FE model overestimated the shear and moment capacities by 20.6 % and 13.2 %, respectively; 4) on average, the equivalent FE model overestimated the shear and moment demands by 8.9 % and 5.5 %, respectively. Overall, the proposed application has demonstrated a fast, economical and accurate method to conduct seismic evaluation and design for irregular structures.

MSSA: Multi-Representation Semantics-Augmented Set Abstraction for 3D Object Detection

Accurate recognition and localization of 3D objects is a fundamental research problem in 3D computer vision. Benefiting from transformation-free point cloud processing and flexible receptive fields, point-based methods have become accurate in 3D point cloud modeling, but still fall behind voxel-based competitors in 3D detection. We observe that the set abstraction module, commonly utilized by point-based methods for downsampling points, tends to retain excessive irrelevant background information, thus hindering the effective learning of features for object detection tasks. To address this issue, we propose MSSA, a Multi-representation Semantics-augmented Set Abstraction for 3D object detection. Specifically, we first design a backbone network to encode different representation features of point clouds, which extracts point-wise features through PointNet to preserve fine-grained geometric structure features, and adopts VoxelNet to extract voxel features and BEV features to enhance the semantic features of key points. Second, to efficiently fuse different representation features of keypoints, we propose a Point feature-guided Voxel feature and BEV feature fusion (PVB-Fusion) module to adaptively fuse multi-representation features and remove noise. At last, a novel Multi-representation Semantic-guided Farthest Point Sampling (MS-FPS) algorithm is designed to help set abstraction modules progressively downsample point clouds, thereby improving instance recall and detection performance with more important foreground points. We evaluate MSSA on the widely used KITTI dataset and the more challenging nuScenes dataset. Experimental results show that compared to PointRCNN, our method improves the AP of “moderate” level for three classes of objects by 7.02%, 6.76%, and 5.44%, respectively. Compared to the advanced point-voxel-based method PV-RCNN, our method improves the AP of “moderate” level by 1.23%, 2.84%, and 0.55% for the three classes, respectively.

Massive Point Cloud Processing for Efficient Construction Quality Inspection and Control.

The construction of large-scale civil infrastructures requires massive spatiotemporal data to support the management and control of scheduling, quality control, and safety monitoring. Existing artificial-intelligence-based data processing algorithms rely heavily on experienced engineers to adjust the parameters of data processing, which is inefficient and time-consuming when dealing with huge datasets. Limited studies have compared the performance of different algorithms on a unified dataset. This study proposes a framework and evaluation system for comparing different data processing policies for processing huge spatiotemporal data in construction quality control. The proposed method compares the combination of multiple types of algorithms involved in the processing of massive point cloud data. The performance of data processing strategies is evaluated through this framework, and the optimal point cloud processing strategies are explored based on registration accuracy and data fidelity. Results show that a reasonable choice of combinations of point cloud sampling, filtering, and registration algorithms can significantly improve the efficiency of point cloud data processing and satisfy engineering demands for data accuracy and completeness. The proposed method can be applied to the civil engineering problem of processing a large amount of point cloud data and selecting the optimal processing method.