A Robust Image-Based Framework for Borehole Fracture Detection and Quantitative Characterization
IETI Transactions on Data Analysis and Forecasting (iTDAF) · 25.03.2026 · 4 (1)
DOI: 10.3991/itdaf.v4i1.59291 ↗Kurz gefasst
Fachartikel in der Zeitschrift IETI Transactions on Data Analysis and Forecasting (iTDAF), online erschienen am 25.03.2026. Verfasst von Yuhang Gao. Die Arbeit nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Ensuring the safety and intelligence of underground coal mining has become a crucial task in the context of developing new productive forces. Fractures within surrounding rock layers are key factors affecting the stability of underground engineering, yet traditional manual interpretation of borehole images is inefficient and subjective. To address this issue, this study proposes an automated framework for fracture identification and quantitative characterization based on image processing and clustering analysis. First, an improved Canny edge detection algorithm is applied to borehole wall images after grayscale conversion and Gaussian filtering, effectively suppressing geological noise such as natural textures, drilling traces, and mud residues. Morphological dilation and binarization are then used to enhance fracture connectivity and extract clear fracture boundaries. Subsequently, an unsupervised clustering algorithm, Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN), is introduced to automatically classify fracture points and separate multiple intersecting fractures. On this basis, a robust sine-model fitting approach is developed using a combination of random sample consensus (RANSAC) and Levenberg-Marquardt optimization, enabling accurate estimation of geometric parameters including amplitude, period, phase, and offset of each fracture. Experimental results demonstrate that the proposed method achieves high recognition accuracy and strong noise resistance. The detected fractures exhibit excellent agreement with manual observations, with the best-fitting models yielding coefficients of determination (R²) above 0.94. Compared with conventional manual interpretation, the proposed approach significantly improves automation, consistency, and computational efficiency. This research provides a reliable and interpretable framework for intelligent fracture detection and quantitative analysis in underground engineering, offering valuable technical support for the intelligent and safe development of coal mines.
Zitieren
APA
Gao, Y. (2026). A Robust Image-Based Framework for Borehole Fracture Detection and Quantitative Characterization. IETI Transactions on Data Analysis and Forecasting (iTDAF), 4(1). https://doi.org/10.3991/itdaf.v4i1.59291
Vancouver
Gao Y. A Robust Image-Based Framework for Borehole Fracture Detection and Quantitative Characterization. IETI Transactions on Data Analysis and Forecasting (iTDAF). 2026;4(1). doi:10.3991/itdaf.v4i1.59291
BibTeX
@article{Gao2026,
author = {Gao, Yuhang},
title = {A Robust Image-Based Framework for Borehole Fracture Detection and Quantitative Characterization},
journal = {IETI Transactions on Data Analysis and Forecasting (iTDAF)},
year = {2026},
volume = {4},
number = {1},
doi = {10.3991/itdaf.v4i1.59291}
}
EUVIMED