Figure - available from: Environmental Earth Sciences
This content is subject to copyright. Terms and conditions apply.
Source publication
In order to realize low-cost, fast, and accurate fracture identification in lining quality inspection and underground engineering stability evaluation, we propose an intelligent fracture identification method, which can achieve depth extraction of fracture information from rock images. Firstly, the fracture identification model combined with the re...
Similar publications
Most slope collapse accidents are indicated by certain signs before their occurrence, and unnecessary losses can be avoided by predicting slope deformation. However, the early warning signs of slope deformation are often misjudged. It is necessary to establish a method to determine the appropriate early warning signs in sliding thresholds. Here, to...
Citations
... As far as reservoirs with high density and low permeability are concerned, fractures are important percolation channels that have non-negligible impacts on the connectivity (porosity), migration characteristics (permeability), spatial heterogeneity, and anisotropy of the total effective reservoir capacity (Pervago et al. 2018;Tang et al. 2021). In geotechnical engineering and geological disaster prediction and control measures, fractures are considered to be key factors affecting rock stability (Azarafza et al.2021a;Pan et al. 2023). For example, shear fractures in the surrounding rock of roadways and stopes may seriously affect the deformation and stability of the rock mass in those areas (Chen et al. 2021). ...
... As a result, the influencing effects of pores and low-density minerals can be avoided. Deep learning methods have been widely applied to image-based fractures extraction processes, including macroscopic rock outcrops (Azarafza et al. 2019;Pan et al. 2023) and microscopic CT scanning (Lu et al. 2022;Pham et al. 2023;Roslin et al. 2023). Chen et al. (2021) automated rock tunnel fracture segmentation using a CNN-based model named FraSegNet. ...
Image-based automatic fracture extraction methods have many practical applications in geological and engineering. Fracture identification and quantitative characterization require the means of interpreting and statistically analyzing image data. In comparison to traditional digital image processing methods, supervised semantic segmentation methods based on Convolutional Neural Networks (CNN) offer distinct advantages in extracting fractures from CT images. The study analyzes the characteristics of fracture areas in CT images and compares the results obtained through traditional threshold segmentation methods with those achieved using deep learning techniques. An integrated approach combining interactive image segmentation was proposed in this study to extract the fractures, in which deep learning methods were to determine the fracture areas, as well as morphological operation and threshold segmentation methods were adopted to extract the fractures from the small target areas. The implementation of this method significantly enhances the efficiency of extraction results compared to manual fracture extractions. This study selected CT images of igneous rock with a resolution of 0.7 um as the research object. DeepLab V3 + and UNet3 + network models in the PaddleSeg framework were used for the deep learning process.
The information of fractures in rock mass is an essential indicator to evaluate the quality of rock mass. It is precise and efficient to obtain information of fractures by computer vision (CV)-based fracture detection technologies. However, rock masses in practical engineering always exhibit complex fractures with messy edges, shadows, and uneven rock surfaces. These factors negatively impact the effectiveness of fracture detection algorithms. To address such issues, this paper proposes a confidence score-based rock fracture detection algorithm that can effectively identify fractures in complex situations. The image with fractures is traversed to acquire candidate points at first. We proposed a method to evaluate the confidence scores of these candidate points according to the values of Hessian matrix eigenvalues, the degree of symmetry, and the gray-scale value. The candidate points are further filtered through threshold restriction and non-maximum suppression. Furthermore, the complete centerlines of the fractures are obtained by the connection of the candidate points and the connection of discontinuous centerlines. Finally, pseudo-fractures and noise are eliminated to get complete fractures. Experimental results show that the newly proposed algorithm can accurately identify fractures in complex situations. The algorithm avoids interference and noise, which achieves more precise results than other conventional fracture detection algorithms, is an effective method to identify fractures in rocky geological engineering and can also serve as a guide for other relevant fracture identification algorithm in rock mass.
Natural fractures are effective storage spaces and important seepage channels for oil and gas reservoirs. Accurately identifying natural fractures in reservoirs is crucial for the exploration and development of oil and gas resources. This article combines conventional and imaging logging data and uses machine learning to automatically identify natural fractures in reservoirs. The fracture labels of conventional logging come from imaging logging. Conventional logging data is decomposed through multi-scale wavelet to extract components that reflect fracture information, and further build the original data set. The AdaBoost model is trained based on a modified dataset of balanced samples for automatic fractures recognition in logging. The research results indicate that the approximate component and high-frequency component reflect the fluctuation of the formation and noise information respectively, and have little impact on the reservoir fractures identification; The medium frequency component can reflect the characteristic information of fractures and can be used for model training; After hyper-parameter optimization, the AdaBoost model has high accuracy and generalization ability, and can still accurately identify the types and distribution of natural fractures from the actual unbalanced logging data. This research has important guiding significance for the accurate characterization and construction of reservoir.
