Article
Construction & Building Technology
Manohar Yadav, Parvej Khan, Ajai Kumar Singh, Bharat Lohani
Summary: A hybrid ground filtering method was proposed for processing ground points in mobile laser scanning data. After testing and validation, the method showed good performance in various challenging roadway environments. The method is straightforward, computationally efficient, and has the potential for wider application in industry.
AUTOMATION IN CONSTRUCTION
(2021)
Article
Geochemistry & Geophysics
Jintao Li, Hangbin Wu, Xinjiang Ma, Yanyi Li, Han Yue, Chun Liu
Summary: This study focuses on the detection of tree areas blocking traffic signs and simulating branch pruning using MLS point clouds. Four indicators are proposed to quantify the pruning effect, and the method is proven effective in different scenarios. This research is important in automatically identifying street tree branches to be maintained and formulating pruning schemes that balance pruning cost, tree ecological benefits, and cultural benefits.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Computer Science, Software Engineering
Safa Bouguezzi, Hana Ben Fredj, Hassene Faiedh, Chokri Souani
Summary: An improved TSR algorithm inspired by the classical LeNet-5 model is proposed in this work, using the novel activation function SigmaH and convolutional block attention module to enhance recognition accuracy and training efficiency.
Article
Computer Science, Information Systems
Zheyuan Zhang, Jianying Zheng, Yanyun Tao, Yang Xiao, Shumei Yu, Sultan Asiri, Jiacheng Li, Tieshan Li
Summary: This paper focuses on point cloud registration in a complex traffic environment and proposes a 3D registration method based on traffic signs and prior knowledge of traffic scenes. By using traffic signs with reflective films as reference targets, the point cloud data from roadside LiDARs can be registered. The method includes vertical registration and horizontal registration, and has been verified in actual scenarios.
Article
Remote Sensing
Shida Wang, Hangbin Wu, Han Yue, Lianbi Yao, Chun Liu, Haili Sun
Summary: One of the essential works of the tunnel maintenance department is to inspect and maintain the electricity transmission system (ETS). This study proposes a method based on the mobile laser system (MLS) to automatically extract ETS. The method extracts power transmission lines and supporting fixtures from MLS data through two stages of processing. Experimental results validate the effectiveness of the proposed method.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Chemistry, Analytical
Aimad El Issaoui, Ziyi Feng, Matti Lehtomaki, Eric Hyyppa, Hannu Hyyppa, Harri Kaartinen, Antero Kukko, Juha Hyyppa
Summary: This study explored the use of the Roamer-R4DW mobile laser scanning system for road rut depth measurement, developing an automatic algorithm to analyze rut depths and verifying the results against reference pavement plots. The study demonstrated that terrestrial laser scanning data can serve as a suitable reference for MLS-based rutting studies, with MLS-derived rut depths showing adequate precision for operational measurements. The data obtained from this study, covering both pavement and surrounding areas, has the potential for various road environment applications, including high-definition mapping, autonomous navigation, and digitalization of street environments.
Article
Environmental Sciences
Zhipeng Chen, Qingquan Li, Jiayuan Li, Dejin Zhang, Jianwei Yu, Yu Yin, Shiwang Lv, Anbang Liang
Summary: This study proposes a mobile laser scanning (MLS) point cloud registration method based on an inertial trajectory error model, which is proved to be effective and reliable through experiments.
Article
Computer Science, Information Systems
Feng Han, Tao Liang, Jiping Ren, Yuan Li
Summary: This paper presents a novel method to extract rails from MLS point cloud data. It preprocesses the point cloud and defines multi-scale features to extract rail point clouds using a classifier. The method is validated on existing lines and shows high adaptability and accuracy.
Article
Engineering, Electrical & Electronic
Youyuan Li, Chun Liu, Hangbin Wu, Yuanfan Qi
Summary: Mobile laser scanning (MLS) has gained attention in urban applications due to its ability to provide spatial information for urban roads. However, accurate positioning is often hindered in complex urban areas, resulting in degraded point cloud data quality. This article proposes a baseline-based approach to extract intrinsic characteristics from raw data for 3D correction, without requiring additional reference information. The proposed approach, called self-constrained baseline correction model, locates problematic data using abnormal feature information and improves accuracy consistency through nonrigid correction. Experimental results demonstrate significant improvement in data quality.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Jintao Li, Hangbin Wu, Xiaolong Cheng, Yuanhang Kong, Xufei Wang, Yanyi Li, Chun Liu
Summary: This article proposes a new method to extract street trees individually from mobile laser scanning point clouds. The method removes the ground and buildings through data preprocessing, further removes artificial poles that may overlap with street tree crowns through supervoxels region growing, and selects regions of interest (ROI) including street trees and understory vegetation. Then, the main branch part of each tree is separated from the ROI by nonphotosynthetic components clustering, and the remaining photosynthetic components in the ROI are segmented individually based on the individual clustering results.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Shuang Liu, Haili Sun, Zhenxin Zhang, Yuqi Li, Ruofei Zhong, Jincheng Li, Siyun Chen
Summary: In this paper, a novel and efficient deep learning model is proposed for extracting multiscale and discriminative features of water leakages. By integrating a new residual network module (Res2Net) with a cascade structure, the proposed method expands the size of receptive field and improves the feature extraction performance. Experimental results demonstrate the advantages of the proposed method in extracting water leakage features.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Qingyang Xu, Xuefeng Guan, Jun Cao, Yanli Ma, Huayi Wu
Summary: Efficient point cloud visualization is crucial for practical applications. In this paper, a novel neural rendering framework called MPR-GAN is proposed for rendering MLS point clouds. The framework utilizes perspective projection and CGAN-based rendering model to achieve high-quality and real-time rendering. Experimental results show that MPR-GAN outperforms other baseline frameworks in terms of rendering performance.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Proceedings Paper
Automation & Control Systems
Nur Nabilah Abu Mangshor, Nor Syahirah Saharuddin, Shafaf Ibrahim, Ahmad Firdaus Ahmad Fadzil, Khyrina Airin Fariza Abu Samah
Summary: This study developed a speed limit sign recognition method for TSR system using image processing technique with the SSD algorithm to identify the inter-class similarity among different speed limit signs. The model was trained using the GTSD dataset and tested with real-time images of standard Malaysian speed limit signs, achieving an average accuracy of over 92.4% for detecting and recognizing the speed limit signs.
2021 11TH IEEE INTERNATIONAL CONFERENCE ON CONTROL SYSTEM, COMPUTING AND ENGINEERING (ICCSCE 2021)
(2021)
Article
Geochemistry & Geophysics
Siyun Chen, Zhenxin Zhang, Hao Ma, Liqiang Zhang, Ruofei Zhong
Summary: Accurate and automatic detection of road surface elements is crucial for many applications. We propose a content-adaptive hierarchical deep learning model to detect arbitrary-oriented road surface elements from mobile laser scanning point clouds. Experimental results demonstrate that our method maintains robust detection performance in cases of unbalanced category numbers or overlapping road surface elements, improves the recognition effects of small targets, and accurately predicts boundary offsets for each target.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Remote Sensing
Xinxiang Zhu, Craig L. Glennie, Benjamin A. Brooks
Summary: This study proposes an automated change detection strategy using geometric primitives to quantify off-fault deformation in the near field. Based on point cloud data and a deep neural network, the research reveals ground deformation after the South Napa earthquake and summarizes the relationship between deformation and off-fault distances.
JOURNAL OF APPLIED GEODESY
(2022)
Article
Geochemistry & Geophysics
Haojia Lin, Zhipeng Luo, Wen Li, Yiping Chen, Cheng Wang, Jonathan Li
Summary: This paper introduces the research progress of deep learning for 3-D point cloud perception and proposes an adaptive pyramid context fusion (APCF) module to capture contextual information from the point cloud. A multiscale context-aware network called APCF-Net is proposed by applying APCF to the PointConv architecture. The experiments demonstrate that APCF-Net outperforms other methods on 3-D object classification and semantic segmentation tasks.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Zhengchuan Sha, Yiping Chen, Yangbin Lin, Cheng Wang, Jose Marcato, Jonathan Li
Summary: This paper proposes a novel supervoxel segmentation algorithm framework for enhancing road boundaries from 3-D point clouds. The method achieves good performance in road scenes by segmenting the point clouds using seed points and adjusting the centroids of supervoxels.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Geochemistry & Geophysics
Zhipeng Luo, Ziyue Zhang, Wen Li, Yiping Chen, Cheng Wang, Abdul Awal Md Nurunnabi, Jonathan Li
Summary: This study proposes a deep learning framework for individual tree detection in complex forests using UAV laser scanning point clouds. The framework consists of two stages: ground filtering and tree detection. The ground filtering stage utilizes a modified graph convolution network with a local topological information layer, while the tree detection stage employs a top-down slice module and a special multichannel representation. Experimental results show that the framework achieves excellent performance.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Engineering, Civil
Kun Zhao, Lingfei Ma, Yu Meng, Li Liu, Junbo Wang, Jose Marcato, Wesley Nunes Goncalves, Jonathan Li
Summary: This paper presents a multi-level fusion network for 3D vehicle detection from point clouds and images. Extensive experiments show that the proposed network has better detection performance on occluded and distant vehicles, and reduces the false detection of similarly shaped objects.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Civil
Chengming Ye, He Zhao, Lingfei Ma, Han Jiang, Hongfu Li, Ruisheng Wang, Michael A. Chapman, Jose Marcato Junior, Jonathan Li
Summary: This article presents a method to extract lane features from mobile laser scanning data, which shows strong feasibility and robustness, and achieves higher accuracy and robustness compared to most state-of-the-art methods.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Civil
Jianlan Gao, Yiping Chen, Jose Marcato, Cheng Wang, Jonathan Li
Summary: This paper proposes a novel method for rapidly extracting urban road guardrails from MLS point clouds, combining multi-level filtering and a modified DBSCAN clustering. The method is applicable to most types of guardrails and rough slope roads. Experimental results demonstrate that the proposed method outperforms the state-of-the-art method in extracting 3D guardrails from point clouds.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Remote Sensing
Linwei Chen, Bowen Fang, Lei Zhao, Yu Zang, Weiquan Liu, Yiping Chen, Cheng Wang, Jonathan Li
Summary: In this paper, a physics informed neural network (PINN) framework called DeepUrbanDownscale (DUD) is proposed for accurate high-resolution urban surface temperature estimation. The network uses high-precision 3D point clouds and atmospheric physics guidance to achieve accurate temperature estimation at an ultra-high spatial resolution. Experimental results show that the proposed DUD network outperforms traditional methods in predicting urban surface temperatures.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Environmental Sciences
Ye Cheng-ming, Wei Rui-long, Ge Yong-gang, Li Yao, Jose Marcato Junior, Jonathan Li
Summary: The research introduced road factors of aspect to road and road profile to improve the accuracy of landslide susceptibility mapping by considering the influence of landslide movement direction on road. Using the random forest method, landslide susceptibility maps were generated, and the performance of road factors was evaluated using AUC and Gini importance. The results showed that road profile and aspect to road made significant contributions to predicting landslides along the highway compared to primitive factors.
JOURNAL OF MOUNTAIN SCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Mauro dos Santos de Arruda, Lucas Prado Osco, Plabiany Rodrigo Acosta, Diogo Nunes Goncalves, Jose Marcato Junior, Ana Paula Marques Ramos, Edson Takashi Matsubara, Zhipeng Luo, Jonathan Li, Jonathan de Andrade Silva, Wesley Nunes Goncalves
Summary: This paper presents a Convolutional Neural Network (CNN) approach for counting and locating objects in high-density imagery. The proposed method utilizes a feature map enhancement and a multi-sigma refinement of the confidence map, achieving state-of-the-art performance in counting and locating objects.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Civil
Hai Wu, Wenkai Han, Chenglu Wen, Xin Li, Cheng Wang
Summary: This paper introduces a new 3D multi-object tracker that leverages a novel data association scheme and aggregated pairwise cost to robustly, quickly, and accurately track objects in point clouds. Extensive experiments show that the proposed method outperforms existing state-of-the-art methods in terms of tracking accuracy and speed on the KITTI tracking benchmark.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Civil
Lingfei Ma, Ying Li, Jonathan Li, Jose Marcato Junior, Wesley Nunes Goncalves, Michael A. Chapman
Summary: The paper introduces a novel deep learning framework called BoundaryNet, utilizing mobile laser scanning point clouds and high-resolution satellite imagery for road boundary extraction and completion. By employing various algorithms, it achieves more complete and accurate road boundaries, along with calculating the inherent road geometries.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Geochemistry & Geophysics
Xiaoxue Liu, Yiping Chen, Mingqiang Wei, Cheng Wang, Wesley Nunes Goncalves, Jose Marcato, Jonathan Li
Summary: This letter proposes an automatic building instance extraction method based on an improved hybrid task cascade (HTC), which achieved significant improvements in bounding box branch and mask branch compared to the mainstream Mask R-CNN method, especially in terms of AP values in the two branches.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Remote Sensing
Lina Fang, Hao Chen, Huan Luo, Yingya Guo, Jonathon Li
Summary: This paper presents a new intensity-enhanced method for mobile laser scanning point clouds. The method corrects the inconsistent intensity caused by different scanning distances and introduces the dark channel prior to transform intensity information adaptively. Multiple filters are used to remove isolated intensity noises.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Remote Sensing
Haiyan Guan, Xiangda Lei, Yongtao Yu, Haohao Zhao, Daifeng Peng, Jose Marcato Junior, Jonathan Li
Summary: This article introduces an attentive capsule feature pyramid network (ACapsFPN) for accurately extracting road markings from high-resolution UAV images. The ACapsFPN integrates capsule representations and attention mechanisms into the feature pyramid network to improve extraction accuracy. The network is capable of extracting multi-level and multi-scale capsule features and emphasizes informative features through attention modules, providing high-quality and semantically-strong abstractions. Evaluations show the effectiveness and superiority of the ACapsFPN in extracting road markings under complex conditions.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Remote Sensing
Ruilong Wei, Chengming Ye, Tianbo Sui, Yonggang Ge, Yao Li, Jonathan Li
Summary: This study develops a deep learning framework that integrates spatial response features and machine learning classifiers for reliable landslide susceptibility mapping. In practical application, the framework has certain advantages in prediction accuracy, and considering the scale of landslides can improve the performance of the assessment.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)