Article
Environmental Sciences
Deniz Kumlu, Kubra Tas, Isin Erer
Summary: This study proposes two methods based on deep networks for missing data recovery, one using the PEN-Net architecture and the other a regularization based data recovery method. These methods outperform conventional methods in handling challenging pixel-wise and column-wise missing cases, and they are able to deal with extreme column-wise missing data cases where traditional methods fail completely.
Article
Environmental Sciences
Fanruo Li, Feng Yang, Rui Yan, Xu Qiao, Hongjia Xing, Yijin Li
Summary: This study proposes an abnormal region detection algorithm for underground radar based on visual attention mechanism, which enhances and screens abnormal regions quickly through steps such as suppressing background noise and enhancing brightness and directional characteristics. The effectiveness of the algorithm has been verified by actual tests.
Article
Geochemistry & Geophysics
Zi Xian Leong, Tieyuan Zhu
Summary: This paper introduces a deep learning-based velocity inversion method GPRNet for ground penetrating radar (GPR), which accurately inverts electromagnetic velocity models from GPR data for exploration of Earth and planets.
JOURNAL OF GEOPHYSICAL RESEARCH-SOLID EARTH
(2021)
Article
Environmental Sciences
Iliyana D. Dobreva, Henry A. Ruiz-Guzman, Ilse Barrios-Perez, Tyler Adams, Brody L. Teare, Paxton Payton, Mark E. Everett, Mark D. Burow, Dirk B. Hays
Summary: This study successfully quantified peanut yield for different market types using a novel GPR platform and data analysis methods, developing a noninvasive high-throughput peanut phenotyping and yield-monitoring methodology. By systematically searching thresholding range and analyzing relevant features, good correlations were achieved with peanut yield, especially when using multiple linear regression models.
Article
Geochemistry & Geophysics
Pengyu Zhang, Liang Shen, Tailai Wen, Xiaotao Huang, Qin Xin
Summary: In this study, a method for improving the detection of hyperbola in ground-penetrating radar (GPR) images is proposed. The method utilizes phase symmetry to enhance feature resolution and robustness, and constructs a handcrafted feature descriptor based on phase symmetry to extract structural information. Experimental results demonstrate that the proposed method outperforms other methods on different datasets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Federico Lombardi, Hugh D. Griffiths, Maurizio Lualdi, Alessio Balleri
Summary: One of the main limitations of using ground-penetrating radar (GPR) for landmine detection is the interference from clutter, which raises the detection threshold of the system. Characterizing the internal structure of a target may provide key information for developing algorithms to differentiate between landmines and clutter. Through numerical assessment and experimental validation, it has been found that it is possible to identify and characterize the scattering components in the GPR signature of a landmine.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2021)
Article
Geochemistry & Geophysics
Budiman P. A. Rohman, Masahiko Nishimoto, Kohichi Ogata
Summary: This article proposes a lightweight deep learning method for reconstructing missing traces in ground-penetrating radar data. The method simplifies the model structure while maintaining high performance, enabling accurate reconstruction with improved speed and computation. The method shows promising potential for applications such as reinforced concrete inspection.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Jian Huang, Xi Yang, Feng Zhou, Xiaofeng Li, Bin Zhou, Song Lu, Sergey Ivashov, Iraklis Giannakis, Fannian Kong, Evert Slob
Summary: An improved self-supervised learning algorithm SA-DenseCL is proposed and combined with Mask R-CNN for tunnel lining inspection using GPR. The experimental results show that the proposed method outperforms the conventional methods in reinforcement bar identification, void detection, and secondary lining thickness estimation. Therefore, the improved self-supervised learning-based framework can enhance the detection and estimation accuracy in GPR tunnel lining inspection.
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING
(2023)
Article
Environmental Sciences
Hui Wang, Shan Ouyang, Qinghua Liu, Kefei Liao, Lijun Zhou
Summary: Accurately estimating the relative permittivity of buried targets is crucial for reconstructing geological structures. This study proposes a method based on a deep neural network to effectively identify the dielectric properties of buried targets and improve the accuracy of estimating the relative permittivity.
Article
Engineering, Electrical & Electronic
Yintao Ji, Fengkai Zhang, Jing Wang, Zhengfang Wang, Peng Jiang, Hanchi Liu, Qingmei Sui
Summary: A Deep Neural Network based inversion network was proposed for reconstructing the relative permittivity of geo-structures from GPR B-scans. The network utilizes time dimension compression operation and global feature encoder to process data and extract global information.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Multidisciplinary
Rongxiang Gao, Hongqing Zhu, Qi Liao, Baolin Qu, Lintao Hu, Haoran Wang
Summary: This paper proposes a deep learning-based method using ground-penetrating radar (GPR) for recognizing coal fire, which improves the accuracy and speed of delineating coal fire areas. By scanning a self-built coal fire physical model with GPR and comparing the results with GPR images, the spatial evolution law of coal fire areas and the signal characteristics of coal fire in radar images, including combustion cavity, combustion surface, and underground combustion collapse surface, are summarized. The results show that YOLOv5l achieves the highest detection accuracy among different algorithms, meeting the need for coal fire detection. This method lays the foundation for detecting the combustion range in coal fire areas.
Article
Computer Science, Information Systems
Feng Lin, Meng Sun, Shiyu Mao, Bin Wang
Summary: A deep neural network (DNN)-based time delay estimation method is proposed in this paper to estimate the time delays of backscattered echoes. This method is more robust to noise and does not require decorrelation procedures for coherent backscattered echoes, compared to conventional subspace-based and compressive sensing-based methods. Simulation results demonstrate the efficiency of this method in terms of signal-to-noise ratio (SNR) and GPR resolution.
Article
Engineering, Multidisciplinary
Zhengfang Wang, Ming Lei, Jing Wang, Bo Li, Jing Xu, Yuchen Jiang, Qingmei Sui, Yao Li
Summary: This paper proposes an unsupervised deep learning method for translating real ground penetrating radar (GPR) images to simulated ones. The method introduces geometry-consistency constraints to prevent semantic distortion in translation. It was validated using GPR data collected in various scenarios, and the findings demonstrate accurate identification of internal defects in translated GPR images.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2023)
Article
Engineering, Industrial
Yunjie Zhao, Xi Cheng, Taihong Zhang, Lei Wang, Wei Shao, Joe Wiart
Summary: A global-local attention-based feature reconstruction (GLAFR) surrogate model is proposed for uncertainty analysis (UA) in ground penetrating radar (GPR) simulation. The model effectively quantifies the uncertainty of the output by converting uncertain inputs to electric fields instead of using full-wave simulation. It employs global feature scaling (GFS) and local feature reconstruction (LFR) to capture long-term and short-term relationships of features, and a new loss function to accelerate convergence of the model. The validity of the surrogate model is verified by comparing with the UA result from the Monte Carlo method (MCM), and it outperforms existing deep learning methods in terms of prediction quality and Sobol indices evaluation.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Environmental Sciences
Tess X. H. Luo, Wallace W. L. Lai, Zhanzhan Lei
Summary: The 3D ground-penetrating radar (GPR) has been widely used in subsurface surveys and imaging. The quality of the resulting C-scan images depends on the spatial resolution and visualization contrast. However, the measurement normalization of GPR C-scans is arbitrary and there is human bias in interpretation due to different visualization algorithms. Therefore, an objective scheme for mapping GPR signals to visualization contrast should be established.
Article
Engineering, Multidisciplinary
Man-Sung Kang, Namgyu Kim, Jong Jae Lee, Yun-Kyu An
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2020)
Article
Construction & Building Technology
Namgyu Kim, Sehoon Kim, Yun-Kyu An, Jong-Jae Lee
Summary: This study proposes a novel underground object classification method using two-dimensional grid images and deep learning technology, which can better represent the spatial information of underground objects. Experimental results show that the proposed method outperforms conventional methods in classifying underground objects.
INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING
(2021)
Article
Green & Sustainable Science & Technology
Soonkyu Hwang, Yun-Kyu An, Jinyeol Yang, Hoon Sohn
INTERNATIONAL JOURNAL OF PRECISION ENGINEERING AND MANUFACTURING-GREEN TECHNOLOGY
(2020)
Article
Computer Science, Interdisciplinary Applications
Keunyoung Jang, Yun-Kyu An, Byunghyun Kim, Soojin Cho
Summary: This article introduces a deep learning-based automated crack evaluation technique using a ring-type climbing robot for high-rise bridge piers, achieving a precision of 90.92%. The technique utilizes various image processing algorithms to quantify cracks and automatically establish a digital crack map.
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING
(2021)
Article
Engineering, Multidisciplinary
Hyunjin Bae, Keunyoung Jang, Yun-Kyu An
Summary: This article introduces a new deep learning approach, SrcNet, which improves crack detection by enhancing the resolution of raw digital images. Experimental results show a 24% improvement in crack detection compared to using raw images.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2021)
Article
Chemistry, Multidisciplinary
Myung Soo Kang, Yun-Kyu An
APPLIED SCIENCES-BASEL
(2020)
Article
Chemistry, Multidisciplinary
Hyunjun Jung, Seok-Been Im, Yun-Kyu An
APPLIED SCIENCES-BASEL
(2020)
Article
Construction & Building Technology
Eldor Ibragimov, Hyun-Jong Lee, Jong-Jae Lee, Namgyu Kim
Summary: Automatic pavement crack detection is crucial for maintenance evaluation and driving safety. Existing methods are time-consuming and costly. This study proposes a Faster R-CNN-based method for pavement distress detection, which achieves accurate results on real pavement images.
INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING
(2022)
Article
Environmental Sciences
Man-Sung Kang, Yun-Kyu An
Article
Chemistry, Multidisciplinary
Myung Soo Kang, Yun-Kyu An
Summary: This paper proposes a deep learning-based technique for automatically removing background objects from digital images for structural exterior image stitching, achieving a computational cost reduction of 85.7% and generating precise structural exterior maps.
APPLIED SCIENCES-BASEL
(2021)
Article
Engineering, Civil
Namgyu Kim, Jong-Jae Lee
Summary: This study explored the use of Photoluminescence piezospectroscopy (PLPS) for noncontact stress measurement of concrete, revealing that alumina in concrete can serve as a passive stress sensor by PLPS. The research found that spectral detectability increases with increasing alumina concentrations and that compressive stress and spectral shifts have a negative linear relationship.
JOURNAL OF CIVIL STRUCTURAL HEALTH MONITORING
(2021)
Article
Chemistry, Multidisciplinary
Keunyoung Jang, Jong-Woo Kim, Ki-Beom Ju, Yun-Kyu An
Summary: The application of BIM technique in infrastructure lifecycle management has been increasing rapidly, leading to a systematic literature review on recent research and the proposal of an infrastructure BIM platform framework. This platform, consisting of BIM and state-of-the-art techniques, provides a web-based solution for collaborative construction management and lifecycle management methodology after construction.
APPLIED SCIENCES-BASEL
(2021)
