Mobile LiDAR Data Research Articles

Instance-Based Clustering of Road Markings with Wear and Occlusion from Mobile Lidar Data

Instance-Based Clustering of Road Markings with Wear and Occlusion from Mobile Lidar Data

Powerline extraction from aerial and mobile LiDAR data using deep learning

Powerline extraction from aerial and mobile LiDAR data using deep learning

Combining Cylindrical Voxel and Mask R-CNN for Automatic Detection of Water Leakages in Shield Tunnel Point Clouds

Water leakages can affect the safety and durability of shield tunnels, so rapid and accurate identification and diagnosis are urgently needed. However, current leakage detection methods are mostly based on mobile LiDAR data, making it challenging to detect leakage damage in both mobile and terrestrial LiDAR data simultaneously, and the detection results are not intuitive. Therefore, an integrated cylindrical voxel and Mask R-CNN method for water leakage inspection is presented in this paper. This method includes the following three steps: (1) a 3D cylindrical-voxel data organization structure is constructed to transform the tunnel point cloud from disordered to ordered and achieve the projection of a 3D point cloud to a 2D image; (2) automated leakage segmentation and localization is carried out via Mask R-CNN; (3) the segmentation results of water leakage are mapped back to the 3D point cloud based on a cylindrical-voxel structure of shield tunnel point cloud, achieving the expression of water leakage disease in 3D space. The proposed approach can efficiently detect water leakage and leakage not only in mobile laser point cloud data but also in ground laser point cloud data, especially in processing its curved parts. Additionally, it achieves the visualization of water leakage in shield tunnels in 3D space, making the water leakage results more intuitive. Experimental validation is conducted based on the MLS and TLS point cloud data collected in Nanjing and Suzhou, respectively. Compared with the current commonly used detection method, which combines cylindrical projection and Mask R-CNN, the proposed method can achieve water leakage detection and 3D visualization in different tunnel scenarios, and the accuracy of water leakage detection of the method in this paper has improved by nearly 10%.

Computer vision based first floor elevation estimation from mobile LiDAR data

First Floor Elevation (FFE) of a house is crucial information for flood management and for accurately assessing the flood exposure risk of a property. However, the lack of reliable FFE data on a large geographic scale significantly limits efforts to mitigate flood risk, such as decision on elevating a property. The traditional method of collecting elevation data of a house relies on time-consuming and labor-intensive on-site inspections conducted by licensed surveyors or engineers. In this paper, we propose an automated and scalable method for extracting FFE from mobile LiDAR point cloud data. The fine-tuned yolov5 model is employed to detect doors, windows, and garage doors on the intensity-based projection of the point cloud, achieving an mAP@0.5:0.95 of 0.689. Subsequently, FFE is estimated using detected objects. We evaluated the Median Absolute Error (MAE) metric for the estimated FFE in Manville, Ventnor, and Longport, which resulted in values of 0.2 ft, 0.27 ft, and 0.24 ft, respectively. The availability of FFE data has the potential to provide valuable guidance for setting flood insurance premiums and facilitating benefit-cost analyses of buyout programs targeting residential buildings with a high flood risk.

Feature-Centric Approach for Learning-Based Prediction of Pavement Marking Retroreflectivity from Mobile LiDAR Data

Given the crucial importance of pavement marking retroreflectivity in ensuring visibility for road safety, this research investigates the correlation between pavement marking reflectivity and LiDAR data. Empirical data were collected from eight road sections using both a handheld retroreflectometer and a mobile LiDAR. The approach proposed focuses on extracting important features from pavement marking regions of the LiDAR point cloud. A comprehensive feature extraction and feature selection process was employed. In addition, a well-rounded selection of learning algorithms was evaluated. A rigorous hold-out evaluation was incorporated, ensuring that the reported performance metrics were robustly generalizable. The best performing model was able to achieve an R2 of 0.824 on unseen data. The findings of this study illuminate the potential for leveraging relatively inexpensive mobile LiDAR sensors in combination with machine learning techniques in conducting efficient pavement marking assessments, not only to detect completely degraded markings, but to accurately estimate retroreflective properties.

3D MODELING OF ROAD INFRASTRUCTURES ACCORDING TO CITYGML 3.0 AND ITS CITYJSON ENCODING

Abstract. This study proposes a systematic approach for standardized 3D modeling of road infrastructures based on CityGML 3.0 and its CityJSON encoding. The approach involves generating a LoD2 3D model of the road infrastructure based on a semi-automatic extraction of linear features from mobile mapping LiDAR data. This enables geometric and semantic modeling of the roads. A codification system is proposed to assign predefined codes to each linear feature, allowing each section and intersection and each road surface to be modeled separately. The resulting model is converted to a CityJSON file, stored in a document-oriented database and visualized through a web application. The proposed approach provides a cost-effective alternative to traditional manual modeling methods while maintaining a high level of accuracy and consistency. It also considers the validation of data schema and geometric primitives to ensure that any non-conformity with CityJSON schemas, and any topological and/or geometric errors can be detected and then corrected. This is important since schema changes in new versions of CityJSON can result in compatibility issues, while geometric and topological errors can affect the accuracy of 3D models and ultimately lead to inaccurate simulation outcomes or analysis results.

Automatically Refining Degraded and Occluded Lane Marking Information to Enhance the quality of High Definition Maps for Autonomous Vehicles

Autonomous vehicle technology has been advancing over the past decade with many challenges and obstacles. One obstacle to achieving fully autonomous driving is the feasibility of efficiently producing robust high definition (HD) maps of existing road infrastructure in a sustainable manner. HD maps are digital twins of road infrastructure that include 3-D representation of road features critical to a vehicle’s ability to navigate a road and maintain a consistent trajectory. This includes lane markings, barriers, and road edges. HD maps are often produced by first surveying roads using remote sensing technology. The collected data are then segmented using computer vision and machine learning algorithms to produce an HD map of the features of concern. One common challenge that is specific to extracting lane marking information is that the markings are occasionally occluded or degraded by high traffic volumes. This results in failure to detect lane markings and the existence of significant gaps in the extracted information. These gaps are manually filled in by quality control staff in a tedious and time-consuming process. To help overcome such challenges, this paper proposes a novel algorithm through which Kalman filtering is combined with Bézier curve fitting to automatically refine lane marking information. The proposed algorithm is tested on multiple roads where mobile lidar data were collected and segmentation was first applied using deep learning technology. The proposed algorithm was then used to refine the lane marking information which helped close all the gaps and recreate missing portions of the extracted lane markings.

Automatic levee surface extraction from mobile LiDAR data using directional equalization and projection clustering

Automatic levee surface extraction from mobile LiDAR data using directional equalization and projection clustering

Alongshore Variable Accretional and Erosional Coastal Foredune Dynamics at Event to Interannual Timescales

AbstractNatural and constructed dunes are increasingly being utilized to buffer flooding impacts from storms and rising sea levels along sandy coastlines. However, a lack of data at the appropriate spatial and temporal resolution often precludes isolating the total magnitude, specific timing, and alongshore variability of volumetric coastal dune changes resulting from both wave and wind‐driven processes. Here mobile terrestrial lidar data from 46 collections along a 6.5 km stretch of sandy beach and dunes in Duck, NC, USA are used to assess the magnitude and drivers of alongshore variable dune responses between 2012 and 2020. Despite numerous major storm events which impacted the dunes over this time period, the dunes grew volumetrically both in nourished and un‐nourished sections of the study site. Growth of the dunes was seasonally and spatially variable with the largest total growth recorded in the Fall, coinciding with a period of frequent windy storms but also the highest total water levels, and along the regions with the lowest gradient βbeach. The vertical accretion patterns of these wind‐blown sediments are shown to vary depending on the dune management style, specifically, if dunes were not managed, sand fenced, or artificially constructed. Dune erosion also occurred episodically within the study period, with the steepest sloped beach sections likely to be the most impacted during individual storm events.

Inventory of close-to-nature forest stands using terrestrial mobile laser scanning

• Hand-held mobile LiDAR data of a close-to-nature forest in the leaf-on season. • Fully automatic processing using modified mean-shift and RANSAC algorithms. • 96.5 % detection and 3.5 cm DBH RMSE for trees with DBHs from 4 to 80 cm. • A comparison of different levels of low-intensity filtering is performed. • A walking trajectory consisting of concentric circles is recommended. In this work, we evaluated mobile laser scanning (MLS) technology for the acquisition of individual tree data in a close-to-natural forest structure. The data was collected during the leaf-on period in a single-tree selection forest stand. Individual tree positions and diameters at breast height (DBHs) were acquired in an automatic process. MLS data was collected on 1000 m 2 circular inventory plots. We compared three trajectories consisting of the plot perimeter and 1), two perpendicular lines, 2) an additional concentric circle, and 3) four parallel lines. We compared two algorithms for tree segmentation: 1) a density-based approach and 2) a modified mean-shift algorithm. The diameters were estimated using a modified random sample consensus (RANSAC) algorithm. We tested a series of intensity thresholds for filtering returns from green vegetation. We achieved the best results with an intensity threshold of 0.7 quantile of point intensities, and mean-shift segmentation, resulting in the correct identification of trees representing 96.5 % of basal area and an overestimation of 6.8 % of the total basal area. The algorithm omitted mainly small trees and trees at close distances. False detections mainly comprised unvalidated detections of real trees that were not field-measured as their diameter did not exceed the registration limit, or were caused by point structures representing leaves and understory vegetation. Diameters were estimated with a mean error of 0.03 cm and a root mean square error of 3.5 cm. A joinpoint regression model demonstrated that for small trees (<9 cm) the diameters were generally overestimated. Diameters above 12 cm were underestimated consistently by 1 cm. The trajectory comprising two concentric circles was the most efficient.

Network-Level Guardrail Extraction Based on 3D Local Features from Mobile LiDAR Sensor

Guardrails are critical boundary infrastructures protecting against road departures and traffic collisions. The presence and condition information of in-service guardrails are essential for transportation agencies to perform necessary repair or replacement operations on time. Unfortunately, most current practices still rely on manual field surveys or windshield inspections that can be time-consuming, labor-intensive, and subjective. This study proposes an automated, network-level guardrail detection and tracking model based on 3D local features captured in mobile LiDAR data. The 3D local features, including corrugation, vertical profile, connectivity, and continuity of the guardrails, are introduced to extract guardrail status through four key sequential and corresponding steps, including Difference of Normals (DoN)-based segmentation, vertical profile-based filtering, guardrail-associated point re-population, and guardrail tracking. The proposed method is evaluated in two sections on State Route 113 and State Route 9 in Massachusetts. It shows promising performance with high precision rates of 95.6% and 95.5% and excellent length covering rates of 97.9% and 100.0%, respectively. The proposed method will provide a reliable and efficient means for transportation agencies to inspect and evaluate their critical guardrail infrastructure on a network level.

Semantic segmentation of bridge components and road infrastructure from mobile LiDAR data

Emerging mobile LiDAR mapping systems exhibit great potential as an alternative for mapping urban environments. Such systems can acquire high-quality, dense point clouds that capture detailed information over an area of interest through efficient field surveys. However, automatically recognizing and semantically segmenting different components from the point clouds with efficiency and high accuracy remains a challenge. Towards this end, this study proposes a semantic segmentation framework to simultaneously classify bridge components and road infrastructure using mobile LiDAR point clouds while providing the following contributions: 1) a deep learning approach exploiting graph convolutions is adopted for point cloud semantic segmentation; 2) cross-labeling and transfer learning techniques are developed to reduce the need for manual annotation; and 3) geometric quality control strategies are proposed to refine the semantic segmentation results. The proposed framework is evaluated using data from two mobile mapping systems along an interstate highway with 27 highway bridges. With the help of the proposed cross-labeling and transfer learning strategies, the deep learning model achieves an overall accuracy of 84% using limited training data. Moreover, the effectiveness of the proposed framework is verified through test covering approximately 42 miles along the interstate highway, where substantial improvement after quality control can be observed.

Quantitative regional observations of gravel and bedrock influence on beach morphologies

Geological features such as underlying bedrock and coarse sediment layers/deposits are known to influence sandy beach morphology, however existing observations are often spatially limited, qualitative, and/or focused on limited geological feature types. Using mobile terrestrial LiDAR data collected twice yearly from 2017 to 2021, this study quantitatively examines the influence of intermittently and/or semi-permanently exposed gravel deposits and bedrock on seasonal subaerial beach elevations along ~40 km of coast in northern San Diego County, California. Both seasonal beach elevations, surface gravel/bedrock exposure, and modeled nearshore wave conditions varied considerably in space and time. Sandy beaches with limited bedrock/gravel exposure generally had relatively large typical winter-erosion and summer-accretion elevation changes, while beaches with frequent bedrock exposure often had relatively small elevation changes and occasionally exhibited localized seasonal elevation change reversals (e.g. winter-accretion). Beaches with surface gravel often had relatively large foreshore elevation changes compared to the upper beach, and exhibited upper beach elevation change reversals (e.g. winter-accretion). The results also suggest gravel and bedrock influences can be interrelated. Localized seasonal elevation change reversals were also observed on beaches with large alongshore variation in radiation stress near submarine canyons. This study highlights the utility of spatially large-scale and concurrent quantitative mapping of beach elevation and multiple geologic feature types.

Change detection in street environments based on mobile laser scanning: A fuzzy spatial reasoning approach

Automated change detection based on urban mobile laser scanning data is the foundation for a whole range of applications such as building model updates, map generation for autonomous driving and natural disaster assessment. The challenge with mobile LiDAR data is that various sources of error, such as localization errors, lead to uncertainties and contradictions in the derived information. This paper presents an approach to automatic change detection using a new category of generic evidence grids that addresses the above problems. Said technique, referred to as fuzzy spatial reasoning, solves common problems of state-of-the-art evidence grids and also provides a method of inference utilizing fuzzy Boolean reasoning. Based on this, logical operations are used to determine changes and combine them with semantic information. A quantitative evaluation based on a hand-annotated version of the TUM-MLS data set shows that the proposed method is able to identify confirmed and changed elements of the environment with F1-scores of 0.93 and 0.89.

Improvement of 3D LiDAR point cloud classification of urban road environment based on random forest classifier

3D road mapping is essential for intelligent transportation system in smart cities. Road environment receives its data from mobile laser scanning (MLS) systems in the format of LiDAR point clouds, which are distinguished with their accuracy and high density. In this paper, a mobile LiDAR data classification method based on machine learning (ML) is presented. First, data subsampling and slicing are applied, followed by cylindrical neighbourhood selection method to determine the neighbourhood of each point. Second, a new LiDAR-based point feature namely Z mod is introduced, and used along with existing fifteen geometric features as input for a ML algorithm. Finally, Random Forest classifier is applied to a part of (Paris-Lille-3D) MLS point clouds belonging to NPM3D Benchmark. The dataset is about 1.5 km long road in Lille, France with more than 98 million points. The use of Z mod improved the accuracy from 90.26% to 95.23% and achieved sufficient results for classes with low points' portion in the dataset. In addition, the Z mod is the third important feature in the classification process among the sixteen features with about 14.63%. Moreover, using the first six important features achieved almost the maximum overall accuracy with about 60% reduction in the processing time.

STORM DRAIN DETECTION AND LOCALISATION ON MOBILE LIDAR DATA USING A PRE-TRAINED RANDLA-NET SEMANTIC SEGMENTATION NETWORK

Abstract. As the expansion of cities and urban areas results in the construction of more impermeable road surfaces, a well designed urban drainage system becomes of greater importance. However, the accurate and up-to-date mapping of storm drains necessary to create accurate drainage models is often lacking. In recent years, mapping of the road infrastructure is increasingly carried out by highly efficient mobile mapping systems but which lack automatic interpretation of the massive amount data. In this paper we present a fully automatic storm drain detection method to extract and locate storm drain inlets in mobile mapping lidar data. The point cloud is first segmented by a pre-trained RandLa-Net model, which although untrained to segment storm drains, is able the segment storm drain clusters in the hardscape class. The results from this class are further processed by enforcing different requirements to only extract and locate storm drain clusters. Our approach is evaluated on a large testing dataset with 171 storm drains and achieves 81.9%, 95.2% and 88.1% for recall, precision and F1-score respectively. The majority of the false positive and false negative detections are due to incorrect point cloud segmentation of the RandLa-Net. In terms of localisation, our approach achieves an RMSE of 5.5 cm on the centre location while the dimensions of the bounding box are on average 23% off compared to the ground truth.

Quantifying the shape of urban street trees and evaluating its influence on their aesthetic functions based on mobile lidar data

Street trees are important components of an urban green space and understanding and measuring their ecological and cultural services is crucial for assessing the quality of streets and managing urban environments. Currently, most studies mainly focus on evaluating the ecological services of street trees by measuring the amount of greenness, but how to evaluate their aesthetic functions through quantitative measurements of street trees remain unclear. To address this problem, we propose a method to assess the aesthetic functions of street trees by quantifying the shape of greenness inspired by assessments of skyline aesthetics. Using a state-of-the-art mobile mapping system, we collected downtown-wide lidar data and panoramic images in Jinzhou City, Hebei Province, China. We developed a method for extracting the canopy line from the mobile lidar data, and then identified two basic elements, peaks and gaps, from street canopy lines and extracted six indexes (i.e., richness of peaks, evenness of peaks, frequency of peaks, total length of gaps, evenness of gaps and frequency of gaps) to describe the fluctuations and continuities of street canopy lines. We analyzed the abundance and spatial distribution of these indexes together with survey responses on the streets’ aesthetics and found that most of them were significantly correlated with human perception of streets. Compared to indexes of amount of greenness (e.g., green volume and green view index), these shape indexes have stronger influences on the physical aesthetic beauty of street trees. These findings suggest that a comprehensive assessment of the aesthetic function of street trees should consider both shape and amount of greenness. This study provides a new perspective for the assessment of urban green spaces and can assist future urban greening planning and urban landscape management.

Impact of neighbourhood selection methods on mobile LiDAR data classification using machine learning algorithms

Impact of neighbourhood selection methods on mobile LiDAR data classification using machine learning algorithms

Utility‐pole Detection Based on Interwoven Column Generation from Terrestrial Mobile Laser Scanner Data

AbstractMobile lidar scanning is one of the recent technologies that is used to map street scenes rapidly. Among street objects, utility‐poles are more critical to energy companies to monitor regularly through time. This paper presents a novel approach to detect utility‐poles from mobile lidar data in complex city scenes. After removing ground points, the scene is gridded into blocks based on a shared‐partitioning algorithm. Next, an interwoven column generation algorithm is used to create columns. Finally, each of these columns is considered to be a utility‐pole or not. The proposed algorithm is tested on two test areas. The algorithm achieved Completeness, Correctness and Quality of 92.8%, 97.5% and 90.6% in Area 1, and 92.8%, 92.2% and 86.1% in Area 2. The total number of utility‐poles in both areas was 265. The algorithm shows promising results in utility‐pole detection in complex city scenes with attached walls.

Processing Strategy and Comparative Performance of Different Mobile LiDAR System Grades for Bridge Monitoring: A Case Study.

Collecting precise as-built data is essential for tracking construction progress. Three-dimensional models generated from such data capture the as-is conditions of the structures, providing valuable information for monitoring existing infrastructure over time. As-built data can be acquired using a wide range of remote sensing technologies, among which mobile LiDAR is gaining increasing attention due to its ability to collect high-resolution data over a relatively large area in a short time. The quality of mobile LiDAR data depends not only on the grade of onboard LiDAR scanners but also on the accuracy of direct georeferencing information and system calibration. Consequently, millimeter-level accuracy is difficult to achieve. In this study, the performance of mapping-grade and surveying-grade mobile LiDAR systems for bridge monitoring is evaluated against static laser scanners. Field surveys were conducted over a concrete bridge where grinding was required to achieve desired smoothness. A semi-automated, feature-based fine registration strategy is proposed to compensate for the impact of georeferencing and system calibration errors on mobile LiDAR data. Bridge deck thickness is evaluated using surface segments to minimize the impact of inherent noise in the point cloud. The results show that the two grades of mobile LiDAR delivered thickness estimates that are in agreement with those derived from static laser scanning in the 1 cm range. The mobile LiDAR data acquisition took roughly five minutes without having a significant impact on traffic, while the static laser scanning required more than three hours.