Characterization of discontinuities in large-scale rock surfaces from 3d point clouds using deep learning

Abstract

Rock mass discontinuity characterization is essential for stability evaluation and discrete fracture network (DFN) modeling, but conventional field measurements are labor-intensive and hazardous, while many point-cloud-based methods still rely on handcrafted descriptors and lack a direct pathway for converting recognition results into engineering parameters. To address these limitations, this study proposes a Rock Mass Discontinuity (RMD) framework for automated engineering-oriented discontinuity characterization from point clouds. The framework is a task-decoupled pipeline with three stages: sparse voxel-based discontinuity-set segmentation, instance-level geometric partitioning, and geometry-constrained parameter extraction. In the segmentation stage, a point – voxel – point feature learning strategy is adopted to identify discontinuity sets from XYZ-only point cloud input. Based on the segmented discontinuity sets, PCA-based plane normal estimation and DBSCAN-based intra-set clustering are further performed to derive key engineering parameters, including dip direction, dip angle, trace length, normal spacing, and DFN-style geometric representations. Tests on a road-cut slope and an underground mine face show improved semantic segmentation over PointNet++ in the road-cut case and feasible application under more fragmented underground conditions. The extracted dominant orientations are broadly consistent with DSE or field references, indicating that RMD offers a practical pathway from point-cloud recognition to engineering discontinuity parameters and preliminary DFN-style representation.

Keywords

rock mass discontinuities / 3D point cloud / semantic segmentation / instance partitioning / geometric parameter extraction

References

[1] Liu C, Bao H, Wang T, et al.. Intelligent characterization of discontinuities and heterogeneity evaluation of potential hazard sources in high-steep rock slope by TLS-UAV technology [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2026, 18(1): 509-527

[2] Arif A, Zhang C, Sajib M H, et al.. Rock slope stability prediction: A review of machine learning techniques [J]. Geotechnical and Geological Engineering, 2025, 43(3): 124

[3] Zhang W, Wang Z, Wang J, et al.. Thermal infrared response evolution of rock mass discontinuities: Insights from large-scale physical model [J]. International Journal of Thermal Sciences, 2026, 219: 110199

[4] Cruden D M. Describing the size of discontinuities [J]. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 1977, 14(3): 133-137

[5] Priest S D, Hudson J A. Estimation of discontinuity spacing and trace length using scanline surveys [J]. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 1981, 18(3): 183-197

[6] Pahl P J. Estimating the mean length of discontinuity traces [J]. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 1981, 18(3): 221-228

[7] Mauldon M. Estimating mean fracture trace length and density from observations in convex windows [J]. Rock Mechanics and Rock Engineering, 1998, 31(4): 201-216

[8] Zhang Q, Wang Q, Chen J, et al.. Estimation of mean trace length by setting scanlines in rectangular sampling window [J]. International Journal of Rock Mechanics and Mining Sciences, 2016, 84: 74-79

[9] Reid T R, Harrison J P. A semi-automated methodology for discontinuity trace detection in digital images of rock mass exposures [J]. International Journal of Rock Mechanics and Mining Sciences, 2000, 37(7): 1073-1089

[10] Slob S, van Knapen B, Hack R, et al.. Method for automated discontinuity analysis of rock slopes with three-dimensional laser scanning [J]. Transportation Research Record: Journal of the Transportation Research Board, 2005, 1913(1): 187-194

[11] Deb D, Hariharan S, Rao U M, et al.. Automatic detection and analysis of discontinuity geometry of rock mass from digital images [J]. Computers & Geosciences, 2008, 34(2): 115-126

[12] Ji Y, Song S, Zhang W, et al.. Automatic identification of rock fractures based on deep learning [J]. Engineering Geology, 2025, 345: 107874

[13] Li M, Chen M, Lu W, et al.. Automatic extraction and quantitative analysis of characteristics from complex fractures on rock surfaces via deep learning [J]. International Journal of Rock Mechanics and Mining Sciences, 2025, 187: 106038

[14] Zhang W, Xu G, Li T, et al.. Intelligent recognition of weak discontinuities on outcrops of hard rock masses [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2026, 18(5): 3742-3759

[15] Ren M, Hu J, Peng D, et al.. Automatic extraction of rock discontinuity orientations from 3D point clouds via an adaptive clustering and surface fitting approach [J]. International Journal of Rock Mechanics and Mining Sciences, 2025, 194: 106246

[16] Wang J, Zheng J, Hu J, et al.. An interactive framework integrating segment anything model and structure-from-motion for three-dimensional discontinuity identification in rock masses [J]. International Journal of Mining Science and Technology, 2025, 35(10): 1695-1711

[17] Guo J, Wu L, Zhang M, et al.. Towards automatic discontinuity trace extraction from rock mass point cloud without triangulation [J]. International Journal of Rock Mechanics and Mining Sciences, 2018, 112: 226-237

[18] Chen J, Huang H, Zhou M, et al.. Towards semi-automatic discontinuity characterization in rock tunnel faces using 3D point clouds [J]. Engineering Geology, 2021, 291: 106232

[19] Yi X, Feng W, Wang D, et al.. An efficient method for extracting and clustering rock mass discontinuities from 3D point clouds [J]. Acta Geotechnica, 2023, 18(7): 3485-3503

[20] Günen M A, Aliyazıcıoğlu Ş. Discontinuities identification from rock outcrop using auto-encoder and point clouds [J]. Bulletin of Engineering Geology and the Environment, 2025, 84(9): 418

[21] Tang N, Wang L, Jiang H, et al.. A new clustering method of rock discontinuity sets based on modified K-means algorithm [J]. Bulletin of Engineering Geology and the Environment, 2023, 82(11): 415

[22] Riquelme A J, Abellán A, Tomás R, et al.. A new approach for semi-automatic rock mass joints recognition from 3D point clouds [J]. Computers & Geosciences, 2014, 68: 38-52

[23] Ma J, Luo B, Zhao Y, et al.. Detection and automatic identification of landslide areas from the LiDAR point clouds using improved DBSCAN [J]. Landslides, 2025, 22(11): 3843-3854

[24] Ruan Y, Liu W, Wang T, et al.. Dominant partitioning of discontinuities of rock masses based on DBSCAN algorithm [J]. Applied Sciences, 2023, 13(15): 8917

[25] Wang C, Zou X, Han Z, et al.. The automatic interpretation of structural plane parameters in borehole camera images from drilling engineering [J]. Journal of Petroleum Science and Engineering, 2017, 154: 417-424

[26] Ge Y, Tang H, Xia D, et al.. Automated measurements of discontinuity geometric properties from a 3D-point cloud based on a modified region growing algorithm [J]. Engineering Geology, 2018, 242: 44-54

[27] Daghigh H. Efficient automatic extraction of discontinuities from rock mass 3D point cloud data using unsupervised machine learning and RANSAC [J]. International Journal of Rock Mechanics and Mining Sciences, 2022, 172: 105603

[28] Shao Y, Li P, Jing R, et al.. A machine learning-based method for lithology identification of outcrops using TLS-derived spectral and geometric features [J]. Remote Sensing, 2025, 17(14): 2434

[29] Ge Y, Cao B, Tang H. Rock discontinuities identification from 3D point clouds using artificial neural network [J]. Rock Mechanics and Rock Engineering, 2022, 55(3): 1705-1720

[30] Günen M A. Explainable artificial intelligence for rock discontinuity detection from point cloud with ensemble methods [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2025, 17(12): 7590-7611

[31] Chen J, Huang H, Cohn A G, et al.. Machine learning-based classification of rock discontinuity trace: SMOTE oversampling integrated with GBT ensemble learning [J]. International Journal of Mining Science and Technology, 2022, 32(2): 309-322

[32] Peng X, Lin P, Sun H, et al.. A semi-automatic method to recognize discontinuity trace in 3D point clouds based on stacking learning [J]. Mining, Metallurgy & Exploration, 2025, 42(1): 433-448

[33] Ren M, Li H, Hu J, et al.. A framework for automatic discontinuity trace extraction using multi-scale surface variation index and transfer-learning enhanced artificial neural network [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2026, 18(3): 1794-1810

[34] Ge Y, Wang H, Liu G, et al.. Automated identification of rock discontinuities from 3D point clouds using a convolutional neural network [J]. Rock Mechanics and Rock Engineering, 2025, 58(3): 3683-3700

[35] Ghazaryan A, Kirakosyan A. Deep learning-based segmentation of rock discontinuities from photogrammetry and LiDAR: A geotechnical case study from Armenia [J]. E3S Web of Conferences, 2025, 642: 02015

[36] Sun J, Zhu S, Sun J, et al.. A robust deep learning approach for rock discontinuity identification from large scale 3D point clouds [J]. Scientific Reports, 2026, 16: 1654

[37] Pham C, Kim B C, Shin H S. Deep learning-based identification of rock discontinuities on 3D model of tunnel face [J]. Tunnelling and Underground Space Technology, 2025, 158: 106403

[38] Battulwar R, Emami E, Naghadehi M Z, et al., et al.Bebis G, Yin Z, Kim E, et al., et al.. Automatic extraction of joint orientations in rock mass using PointNet and DBSCAN[C]. Advances in Visual Computing, 2020, Cham, Springer International Publishing718-727

[39] Chen Q, Ge Y, Tang H. Rock discontinuities characterization from large-scale point clouds using a point-based deep learning method [J]. Engineering Geology, 2024, 337: 107585

[40] Lato M, Kemeny J, Harrap R M, et al.. Rock bench: Establishing a common repository and standards for assessing rockmass characteristics using LiDAR and photogrammetry [J]. Computers & Geosciences, 2013, 50: 106-114

[41] Bao Z, Li Q, Wang C Y. Metal source of giant Huize Zn-Pb deposit in SW China: New constraints from in situ Pb isotopic compositions of galena [J]. Ore Geology Reviews, 2017, 91: 824-836

[42] Cloudcompare. 3D point cloud and mesh processing software Open Source Project [Z], 2023

[43] Vu T, Kim K, Luu T M, et al.. SoftGroup for 3D instance segmentation on point clouds [C]. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 18–24, 2022, 2022, New Orleans, LA, USA, IEEE26982707

[44] Kong D, Wu F, Saroglou C. Automatic identification and characterization of discontinuities in rock masses from 3D point clouds [J]. Engineering Geology, 2020, 265: 105442

[45] Chen J, Zhu H, Li X. Automatic extraction of discontinuity orientation from rock mass surface 3D point cloud [J]. Computers & Geosciences, 2016, 95: 18-31

[46] Daghigh H, Tannant D D, Jaberipour M. A computationally efficient approach to automatically extract rock mass discontinuities from 3D point cloud data [J]. International Journal of Rock Mechanics and Mining Sciences, 2023, 172: 105603

RIGHTS & PERMISSIONS

Central South University

Just Accepted

This article has successfully passed peer review and final editorial review, and will soon enter typesetting, proofreading and other publishing processes. The currently displayed version is the accepted final manuscript. The officially published version will be updated with format, DOI and citation information upon launch. We recommend that you pay attention to subsequent journal notifications and preferentially cite the officially published version. Thank you for your support and cooperation.