Drilling fracture plane feature extraction method, system, equipment and medium
By using morphological processing and cluster analysis of borehole imaging unfolded images, combined with DBSCAN clustering and two-stage robust fitting, the problem of misjudgment of fractures and fissures in borehole imaging technology was solved, achieving high-precision fracture feature extraction and parameter calculation, and supporting three-dimensional reconstruction and risk assessment.
Patent Information
- Application Number
- CN202511877966.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-12
- Publication Date
- 2026-03-20
AI Technical Summary
Existing borehole imaging technology is difficult to accurately handle fractures and fissures caused by image quality or geological reasons. It often misclassifies the same physical fissure as multiple independent objects, affecting the accurate fitting of subsequent geometric parameters. Furthermore, traditional methods are highly subjective and inefficient in evaluating joint roughness, making it difficult to quantify uncertainties and limiting the application of 3D reconstruction.
Morphological processing and cluster analysis of borehole imaging unfolded images are used to connect fracture regions through morphological closing operations. Combined with DBSCAN clustering and two-stage robust fitting, outliers are removed, sinusoidal curve fitting is performed, joint roughness coefficients are calculated, and fracture instantiation and parameter extraction are realized.
It improves the accuracy and stability of fracture geometry parameters, reduces the impact of noise points, realizes automated and intelligent fracture feature extraction, provides a high-precision data foundation, and provides reliable support for 3D reconstruction and risk assessment.
Smart Images
Figure CN121708320A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mining engineering technology, and in particular to a method, system, equipment and medium for extracting planar features of borehole fractures. Background Technology
[0002] The safe and efficient mining of mineral resources relies heavily on the accurate detection of the geological structure within the surrounding rock of the tunnel. The fracture network in the rock mass is a key factor controlling the stability of the surrounding rock, water conduction, and gas migration. Precisely quantifying its geometric morphology is crucial for early warning and prevention of geological disasters.
[0003] Borehole imaging technology has gradually replaced traditional core logging and become the primary means of obtaining information about the internal structure of rock masses. This technology acquires panoramic data of the borehole wall using optical or acoustic probes inside the borehole and unfolds it into a two-dimensional planar image. Due to the spatial geometric relationship between the borehole and the fracture surface, a single fracture typically appears as a "sine curve" in the image.
[0004] However, existing borehole imaging methods often rely on simple connected component analysis in practical applications, which makes it difficult to handle fractures caused by image quality or geological reasons. They often misclassify the same physical fracture as multiple independent objects, thus affecting the accurate fitting of subsequent geometric parameters (such as amplitude, period, and phase). Summary of the Invention
[0005] The purpose of this invention is to address the shortcomings of the prior art by providing a method, system, device, and medium for extracting planar features of borehole fractures, thereby solving the problems in the prior art.
[0006] The present invention specifically provides the following technical solution: A method for extracting planar features of borehole fractures includes the following steps: The borehole imaging unfolded image was acquired, and the cracks were identified at the pixel level in the borehole imaging unfolded image to obtain a binarized crack image; Morphological closing operations are performed on the binarized crack image to connect the fracture regions and extract connected components; Obtain the axial coordinates of the centroid of each connected component, and use the axial coordinates as feature vectors to aggregate connected components whose spatial positions are within a threshold, as crack instances; Using all pixels on the outer contour of the crack instance as observations, a sine curve is initially fitted to obtain an initial curve. The distance from all pixels in each connected region to the initial curve is determined. Based on a preset distance threshold or effective pixel ratio, connected regions with distances deviating from the threshold or pixels lower than the effective pixel ratio are removed as abnormal blocks. Using the remaining pixels after removing outliers, a sine curve fitting is performed again to finally extract the amplitude, period, phase, and centerline position parameters of the crack instance.
[0007] Preferably, morphological closing operations are performed on the binarized crack image to connect the fracture regions and extract connected components, specifically: A 5×5 pixel rectangular structural element is selected for dilation operation to fill the fracture gaps in the binarized fracture image, and an erosion operation is performed to restore the original shape of the fracture, resulting in a fracture image with connected fracture regions. Using the 8-neighborhood connectivity criterion, the fracture image of the fractured region is divided into multiple independent fracture segments. The centroid coordinates, area, and geometric properties of the circumscribed rectangle of each fracture segment are recorded as the extracted connected components.
[0008] Preferably, the initial curve is obtained by using all pixels of the outer contour of the crack instance as observation values and performing preliminary sine curve fitting, specifically as follows: A preliminary sine curve fitting was performed on the clustered fracture instances using the nonlinear least squares method; the specific expression is as follows: ; in, For amplitude, For a period of time, For phase, The centerline position, y This is the output value.
[0009] Preferably, the step of performing pixel-level crack identification on the borehole imaging unfolded image to obtain a binarized crack image specifically involves: The borehole imaging unfolded image is filtered; The filtered borehole image unfolded image is binarized, and morphological opening is performed using a 3×3 pixel rectangular structuring element to obtain a binarized fracture image.
[0010] Preferred options also include: Based on the refined and fitted sine curve, the central contour line of the crack is derived, and the central contour line is sampled with equal arc length to obtain discrete coordinate points. The joint roughness coefficient is calculated based on the discrete coordinate points.
[0011] Preferably, based on the refined and fitted sine curve, the center contour line of the crack is derived, and the center contour line is sampled with equal arc lengths to obtain discrete coordinate points, specifically: For the x-coordinate of each data point, obtain the corresponding y-value on the central contour line; Based on the coordinates of each data point, an equal arc length sampling strategy is used to calculate the total arc length of the center contour line and determine the target number of sampling points. The total arc length is divided into several segments, and the corresponding discrete coordinate points on the contour line are calculated by linear interpolation.
[0012] Preferably, the joint roughness coefficient is calculated based on the discrete coordinate points, specifically as follows: For discontinuous contour lines, a segmented calculation and length-weighted averaging method is used. The line is segmented at the discontinuities, and the root mean square slope parameter is calculated for each continuous segment. The specific expression is as follows: ; in, Indicates the first The root mean square slope parameter of the segment. Indicates the first The arc length of the segment, Indicates the first Data points, Indicates the first Section coordinate, Indicates the first Section value; Based on the root mean square slope parameter, using Calculate the JRC value of each segment, and then perform a weighted average based on the arc length of each segment to obtain the joint roughness coefficient of the overall fracture.
[0013] This invention provides a borehole fracture planar feature extraction system, comprising: The image acquisition module is used to acquire borehole imaging unfolded images and perform pixel-level crack recognition on the borehole imaging unfolded images to obtain binarized crack images. The connected component extraction module is used to perform morphological closing operations on the binarized crack image to connect the fracture regions and extract connected components. The aggregation module is used to obtain the axial coordinates of the centroid of each connected domain, and use the axial coordinates as feature vectors to aggregate connected domains whose spatial positions are within a threshold, as crack instances. The stage processing module is used to perform preliminary fitting of a sine curve with all pixels of the outer contour of the crack instance as observation values to obtain an initial curve, and to determine the distance of all pixels in each connected region to the initial curve. Based on a preset distance threshold or effective pixel ratio, connected regions with distance deviations greater than the threshold or pixels lower than the effective pixel ratio are discarded as abnormal blocks. The output module is used to perform sine curve fitting again using the remaining pixels after removing outliers, and finally extract the amplitude, period, phase and centerline position parameters of the crack instance.
[0014] The present invention provides a computer device, including a memory and a processor. The memory stores a program, and when the program is executed by the processor, the processor performs the steps of the above-described method for extracting planar features of borehole fractures.
[0015] The present invention provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for extracting planar features of borehole fractures.
[0016] Compared with the prior art, the present invention has the following significant advantages: This invention obtains a binarized fracture image by identifying the borehole imaging unfolded diagram, performs morphological closing operations to extract connected components, and obtains the centroid coordinates of the connected components. Connected components whose spatial location is within a threshold are aggregated as fracture instances. It can intelligently aggregate fracture segments based on their natural distribution density along the borehole axis, effectively distinguishing spatially adjacent segments belonging to different fractures. This solves the problem of traditional connected component analysis misclassifying fractures as multiple independent objects, providing a correct data foundation for subsequent parameter extraction. Using all pixels of the outer contour of the fracture instance as observation values, a sine curve is initially fitted. The distance from all pixels in each connected component to the initial curve is calculated. Based on a preset threshold or effective pixel ratio, connected components deviating from the threshold are removed as outliers. The process is then repeated to obtain the final parameters. The innovative two-stage robust fitting mechanism (initial fitting - outlier removal - refined fitting) effectively resists the influence of noise points and local geometric anomalies in the data, avoids parameter deviations caused by direct fitting, and significantly improves the accuracy and stability of extracting key geometric parameters such as amplitude and period. Attached Figure Description
[0017] Figure 1 A flowchart of a borehole fracture planar feature extraction method provided by the present invention; Figure 2 Original images for specific embodiments provided by the present invention; Figure 3 Binarized images provided for specific embodiments of the present invention; Figure 4 The image after connected component analysis is provided as a specific embodiment of the present invention; Figure 5 The image provided here is a DBSCAN clustering result from a specific embodiment of the present invention. Figure 6 The image provided is a preliminary sine fitting result of a specific embodiment of the present invention; Figure 7 The image provided here is an example of an anomaly block detection image from a specific embodiment of the present invention. Figure 8The image provided is a result of quadratic sine fitting in a specific embodiment of the present invention; Figure 9 Extracting the contour centerline image for a specific embodiment of the present invention; Figure 10 The image provided is a detrended contour line image according to a specific embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0019] Several technical bottlenecks still exist in the existing solutions: First, borehole image quality is easily affected by interference. The complex texture of the rock itself, coupled with drilling scratches, mud contamination, and image stitching artifacts, leads to a high degree of similarity between the fracture and the background in terms of grayscale and texture. Traditional image processing or thresholding methods have weak anti-interference capabilities, and the extracted fracture pixels often contain broken, noisy, and discontinuous regions. Second, the process of instantiating a complete fracture from discrete fracture pixels is difficult. Existing methods mostly rely on simple connected component analysis, which is difficult to handle fractures caused by image quality or geological reasons. They often misclassify the same physical fracture as multiple independent objects, thus affecting the accurate fitting of subsequent geometric parameters (such as amplitude, period, and phase). Third, robust sine curve fitting of the identified fractures remains challenging. Conventional least squares methods are easily affected by noise points and local geometric anomalies, causing the fitting results to deviate from the true shape and lack robustness. In addition, the joint roughness coefficient (JRC), as a key indicator for evaluating the mechanical behavior of fracture surfaces, relies on manual comparison with standard contour lines in traditional methods, which is highly subjective and inefficient. Existing discrete coordinate-based calculation methods often employ equal-interval sampling, which fails to accurately reflect the contour undulation characteristics and is sensitive to sampling density, resulting in poor stability. Finally, when reconstructing 3D fractures based on multi-bore information, there is inherent uncertainty in inferring the 3D plane from a one-dimensional curve. Traditional deterministic models struggle to quantify this uncertainty, limiting the assessment of fracture network connectivity probabilities and hindering the in-depth application of this technology in risk assessment and borehole scheme optimization.
[0020] Based on this, the purpose of the present invention is to overcome the defects existing in the prior art and provide an extraction method that can accurately identify and characterize borehole fractures. This method effectively solves the technical bottlenecks of traditional methods in fracture instantiation, parameter extraction and roughness evaluation by systematically integrating image processing, cluster analysis and robust fitting techniques.
[0021] like Figure 1 As shown, this embodiment of the method for extracting planar features of borehole fractures includes the following steps: Step S1: Acquire the borehole imaging unfolded image and perform pixel-level crack identification on the borehole imaging unfolded image to obtain a binarized crack image.
[0022] Drill hole imaging unfolded diagram as follows Figure 2 As shown, the obtained binarized crack image is as follows: Figure 3 As shown in the image. White represents the background, and black represents the cracks.
[0023] Image preprocessing was performed on the borehole imaging unfolded image. A Frangi filter based on the Hessian matrix was used for image enhancement. This filter effectively highlights linear fracture structures in the image by calculating the second derivative features of the image, while suppressing interference factors such as rock texture, drilling scratches, and image stitching artifacts. After filtering, Otsu's thresholding algorithm was used for binarization, followed by morphological opening operations using 3×3 pixel rectangular structuring elements to eliminate isolated noise points, ultimately obtaining a high-quality binarized fracture image. This preprocessing significantly improved the contrast between the fracture and the background, providing a reliable input data foundation for subsequent analysis.
[0024] Step S2: Perform morphological closing operations on the binarized crack image to connect the fracture regions and extract connected components.
[0025] In the morphological processing stage, morphological closing operations are used to connect fractures, a common phenomenon in crack images. Further morphological closing operations (dilation followed by erosion) are performed on the binary image. Dilation fills the fracture gaps, while erosion restores the original shape of the cracks. The number of erosion operations is consistent with the number of dilation operations, with dilation occurring 1-2 times. A slightly larger structuring element (e.g., a 5×5 pixel rectangular kernel) is used to connect larger portions of the cracks that have broken due to noise or blurring. This operation achieves effective connection of fractured regions while preserving geometric features.
[0026] The processed binary image is analyzed using a connected component labeling algorithm (such as 8-neighborhood connectivity). This involves dividing the processed binary image into multiple independent fracture segments and labeling all connected pixel regions; each connected component is considered an independent "fracture segment." The geometric properties of each fracture segment are calculated, such as area, centroid coordinates, and circumscribed rectangle. The centroid ordinate (i.e., the coordinate along the borehole axis) of each segment is recorded as a key feature for subsequent clustering. Figure 4 As shown.
[0027] Step S3: Obtain the axial coordinates of the centroid of each connected component, and use the axial coordinates as feature vectors to aggregate connected components whose spatial location is within the threshold, as crack instances.
[0028] In the fracture instance clustering stage, the DBSCAN density clustering algorithm is used. Using the axial coordinates of fracture segments as feature vectors, segments with similar spatial locations are clustered into single fracture instances based on the density reachability principle. Specifically, the fracture instance clustering is based on DBSCAN:
[0029] This step aims to aggregate multiple segments belonging to the same physical fracture. First, the centroid ordinates of all fracture segments are collected to form a feature dataset. Since DBSCAN is sensitive to data scale, the ordinate data is standardized using the StandardScaler scaling technique, resulting in a mean of 0 and a variance of 1. For irregularly shaped connected components, axial keypoint coordinates based on geometric features are used; these axial keypoints include, but are not limited to, center points and connection points.
[0030] The standardized data is then input into the DBSCAN clustering algorithm. Two core parameters of DBSCAN need to be set according to the actual situation: neighborhood radius (… ) and minimum sample size ( ). The parameters can be set based on empirical values of borehole diameter and expected fracture spacing. For example, the initial parameter can be set to the neighborhood radius. (Standardized units); This can be set to 1 or 2, representing the minimum number of fragments required to form a core point. This algorithm does not require a pre-defined number of clusters, can automatically identify clusters of arbitrary shapes, and is naturally robust to noise. After execution, the algorithm groups fragments with achievable density into the same cluster (i.e., the same fracture instance), and marks fragments that cannot be assigned to any cluster as noise (which may be considered for removal or further processing). After clustering, multiple fracture fragments within each cluster are considered to belong to the same real, continuous fracture, such as... Figure 5 As shown. This step successfully solved the problem of instantiating fracture fissures, avoiding the misclassification of the same physical fissure as multiple independent objects.
[0031] Step S4: Using all pixels on the outer contour of the crack instance as observations, perform preliminary fitting of a sine curve to obtain initial curve parameters, and determine the distance from all pixels in each connected region to the initial curve. Based on a preset distance threshold or effective pixel ratio, connected regions with distances deviating from the threshold or pixels lower than the effective pixel ratio are removed as abnormal blocks.
[0032] For geometric parameter extraction, a two-stage robust sine curve fitting method is used: This step is used to accurately extract the geometric parameters of each fracture instance. For each fracture cluster (representing a fracture instance) obtained in S3:
[0033] 1. The first stage involves preliminary fitting of the clustered fracture instances: extracting the coordinates of all pixels in all fracture segments within the cluster. The function model is:
[0034] ; in For amplitude, The period is (usually fixed at the borehole circumference). For phase, The centerline position, y This is the output value. A nonlinear least squares algorithm (such as the Levenberg-Marquardt algorithm) is used to perform initial curve fitting on the pixel set, with initial parameters set as follows: amplitude. Take data points y Half the range of the coordinates, period Fixed as the borehole perimeter, phase Initialized to 0, centerline position Take data points y The mean of the coordinates is used to obtain the initial parameter estimates. ),like Figure 6 As shown.
[0035] 2. The second stage employs an outlier detection and removal algorithm: Based on the initially fitted curve, the average or median vertical distance from all pixels within each fracture segment (connected region) to the fitted curve is calculated. A distance threshold is set (2-3 times the standard deviation of the average distance). If the average distance of a segment exceeds this threshold, the segment is identified as an "outlier block." Simultaneously, the proportion of pixels within each segment whose distance is less than a certain tolerance (e.g., 2 pixels) (the proportion of valid points) is calculated. If the proportion of pixels within a segment whose distance is less than 2 pixels is too low (e.g., below 30%), it is also identified as an outlier block. These outlier blocks are then removed from the data point set of the fracture instance. Figure 7 As shown in the figure, the red lines represent detected anomalies. Finally, a refined fitting is performed using a clean data point set to obtain the final geometric parameters of the fracture. This staged fitting strategy significantly improves the accuracy and robustness of parameter estimation.
[0036] Step S5: Using the remaining pixels after removing the outlier blocks, perform sine curve fitting again to finally extract the amplitude, period, phase and centerline position parameters of the crack instance.
[0037] Refined Fit: Using the clean pixel set remaining after removing outliers, a nonlinear least-squares sine curve fitting is performed again. The parameters obtained from this fitting ( This refers to the final precise quantitative geometric characterization parameters of the fracture instance, such as... Figure 8 As shown.
[0038] To further quantify the surface morphology characteristics of the fractures, a joint roughness coefficient calculation method based on equal arc length sampling was adopted. The central contour line of the fracture was extracted based on the refined fitted sine curve. The specific operation is as follows:
[0039] The original center outline is calculated by each horizontal coordinate. The arithmetic mean of the ordinates of all pixels is obtained for each... x Coordinates (along the borehole development direction), and their corresponding coordinates y The value is a point on the curve, that is... The expression is: ; in, Indicates horizontal position The total number of pixels at that location. Indicates the first indivual The contour line directly reflects the comprehensive roughness characteristics of the fracture at both macroscopic and microscopic scales. Considering that JRC is mainly used to characterize the roughness characteristics of horizontal fractures, the contour line is further de-trended to separate the contributions of undulations at different scales. The processed image is shown below. Figure 9 As shown.
[0040] To eliminate the influence of macroscopic fluctuations on JRC evaluation, the extracted center contour line is removed from the processed image, and the fitted sine function is used. As a trend benchmark, in practice, this sine function is subtracted from the original contour line relative to its overall mean. fluctuation components This preserves the baseline height of the contour while removing large-scale waveform undulations. Detrended contour lines. The calculation process is as follows:
[0041] ; ; in, To fit the sine function in The value at that location, This is the average value of the function across all data points. The resulting detrended profile mainly reflects the local and microscopic roughness of the fracture surface and is suitable for analyzing the influence of secondary roughness on JRC. The resulting detrended profile image is shown below. Figure 10As shown.
[0042] To calculate the JRC, discrete sampling is required from a continuous center profile. This invention employs a constant arc length sampling strategy:
[0043] 1. Calculate the total arc length of the center contour line using a constant arc length sampling strategy. .
[0044] 2. Determine the target number of sampling points (For example (50-200 points are recommended). The total arc length... Divided into equal parts Section, obtained The arc length position corresponding to each sampling point.
[0045] 3. Based on these arc length positions, calculate the corresponding arc lengths on the contour line using linear interpolation. A discrete coordinate point.
[0046] For discontinuous contour lines, a segmented calculation and length-weighted averaging method is used. The contour lines are segmented at the discontinuities, and the local roughness coefficient is calculated for each continuous segment, i.e., the root mean square slope parameter of these discrete points is calculated. The formula is: ; in, For the first in this section The coordinates of each data point For the first The arc length corresponding to each sampling point Indicates the number of data points.
[0047] Finally, using the formula: Calculate the joint roughness coefficient JRC value. If the contour line is discontinuous, calculate the JRC of each continuous segment, and then perform a weighted average according to the arc length of each segment to obtain the overall joint roughness coefficient JRC value of the fracture. This method not only improves the calculation accuracy but also enhances the adaptability to discontinuous contours. The final results are shown in Table 1.
[0048] The method of this invention has good scalability and adaptability. By adjusting the algorithm parameters, it can adapt to changes in fracture characteristics under different geological conditions. This method is not only applicable to the coal mining field, but can also be extended to other geotechnical engineering fields such as tunnel engineering and slope stability analysis. With the continuous development of borehole imaging technology, this method is expected to play an important role in a wider range of geological exploration and engineering survey fields. This invention establishes a complete and efficient method for extracting planar features of borehole fractures by systematically integrating image processing, cluster analysis, and robust fitting techniques. This method shows significant advantages in terms of instantiation accuracy, parameter fitting robustness, and roughness evaluation accuracy, providing strong technical support for engineering practice and scientific research in related fields.
[0049] Meanwhile, this invention offers more accurate roughness evaluation: employing an equal arc length sampling strategy, the sampling points are evenly distributed along the actual contour line of the fracture, which more accurately reflects the undulation characteristics of the fracture compared to equal-interval sampling. Combined with a piecewise weighted average method, it can provide a reasonable overall JRC estimate even for discontinuous contours, and the calculation results are more consistent with geomechanical realities. Enhanced automation and intelligence: This invention achieves full automation from fracture pixel identification, instance aggregation, parameter fitting to roughness calculation, reducing reliance on human experience and significantly improving processing efficiency and result consistency, providing a feasible solution for batch processing of large-scale borehole data. Laying a solid foundation for 3D reconstruction: The high-precision, quantitative fracture geometry and morphology parameters output by this invention serve as high-quality input data for 3D probabilistic reconstruction of fracture networks, effectively reducing the uncertainty in inferring 3D spatial morphology from 2D information, and providing reliable support for subsequent connectivity analysis, stability evaluation, and exploration scheme optimization.
[0050] Table 1 Quantitative Analysis Results of Complex Fractures Note: The crack numbers are the crack numbers from top to bottom in the final fitting result.
[0051] This invention proposes a borehole fracture planar feature extraction system, comprising: The system comprises the following modules: an image acquisition module for acquiring borehole imaging unfolded images and performing pixel-level crack identification on these images to obtain a binarized crack image; a connected component extraction module for performing morphological closing operations on the binarized crack image to connect fracture regions and extract connected components; an aggregation module for obtaining the axial coordinates of the centroid or representative point of each connected component and using these axial coordinates as feature vectors to aggregate connected components whose spatial positions are within a threshold value, thus creating crack instances; a stage processing module for using all pixels of the outer contour of the crack instance as observation values to perform preliminary sine curve fitting, obtain initial curve parameters, and determine the distance from all pixels in each connected component to the initial curve. Based on a preset distance threshold or effective pixel ratio, connected components whose distance deviation is greater than the threshold or whose pixel ratio is lower than the effective pixel ratio are removed as outliers; and an output module for using the remaining pixels after removing outliers to perform sine curve fitting again, ultimately extracting the amplitude, period, phase, and centerline position parameters of the crack instance.
[0052] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a program, and when the program is executed by the processor, the processor performs the steps of a method for extracting planar features of borehole fractures.
[0053] According to the disclosed embodiments, the computer device can communicate with one or more external devices (e.g., keyboard, pointing device, Bluetooth communication, etc.) or with any device that enables the computing device to communicate with one or more other computing devices (e.g., router, demodulator, etc.).
[0054] The above description, in conjunction with specific preferred embodiments, provides a more detailed explanation of the present invention. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such deductions or substitutions should be considered to fall within the scope of protection of the present invention.
Claims
1. A method for extracting planar features of borehole fractures, characterized in that, Including the following steps: The borehole imaging unfolded image was acquired, and the cracks were identified at the pixel level in the borehole imaging unfolded image to obtain a binarized crack image; Morphological closing operations are performed on the binarized crack image to connect the fracture regions and extract connected components; Obtain the axial coordinates of the centroid of each connected component, and use the axial coordinates as feature vectors to aggregate connected components whose spatial positions are within a threshold, as crack instances; Using all pixels on the outer contour of the crack instance as observations, a sine curve is initially fitted to obtain an initial curve. The distance from all pixels in each connected region to the initial curve is determined. Based on a preset distance threshold or effective pixel ratio, connected regions with distances deviating from the threshold or pixels lower than the effective pixel ratio are removed as abnormal blocks. Using the remaining pixels after removing outliers, a sine curve fitting is performed again to finally extract the amplitude, period, phase, and centerline position parameters of the crack instance.
2. The method for extracting planar features of borehole fractures as described in claim 1, characterized in that, The binarized crack image is subjected to morphological closing operations to connect the fracture regions and extract connected components, specifically: A 5×5 pixel rectangular structural element is selected for dilation operation to fill the fracture gaps in the binarized fracture image, and an erosion operation is performed to restore the original shape of the fracture, resulting in a fracture image with connected fracture regions. Using the 8-neighborhood connectivity criterion, the fracture image of the fractured region is divided into multiple independent fracture segments. The centroid coordinates, area, and geometric properties of the circumscribed rectangle of each fracture segment are recorded as the extracted connected components.
3. The method for extracting planar features of borehole fractures as described in claim 1, characterized in that, The initial curve is obtained by using all pixels of the outer contour of the crack instance as observation values and performing preliminary fitting of a sine curve, specifically as follows: A preliminary sine curve fitting was performed on the clustered fracture instances using the nonlinear least squares method; the specific expression is as follows: ； in, For amplitude, For a period of time, For phase, The centerline position, y This is the output value.
4. The method for extracting planar features of borehole fractures as described in claim 1, characterized in that, The step of performing pixel-level crack identification on the borehole imaging unfolded image to obtain a binarized crack image is as follows: The borehole imaging unfolded image is filtered; The filtered borehole image unfolded image is binarized, and morphological opening is performed using a 3×3 pixel rectangular structuring element to obtain a binarized fracture image.
5. The method for extracting planar features of borehole fractures as described in claim 1, characterized in that, Also includes: Based on the refined and fitted sine curve, the central contour line of the crack is derived, and the central contour line is sampled with equal arc length to obtain discrete coordinate points. The joint roughness coefficient is calculated based on the discrete coordinate points.
6. The method for extracting planar features of borehole fractures as described in claim 5, characterized in that, Based on the refined fitted sine curve, the central contour line of the crack is derived. The central contour line is then sampled with equal arc lengths to obtain discrete coordinate points, specifically: For the x-coordinate of each data point, obtain the corresponding y-value on the central contour line; Based on the coordinates of each data point, an equal arc length sampling strategy is used to calculate the total arc length of the center contour line and determine the target number of sampling points. The total arc length is divided into several segments, and the corresponding discrete coordinate points on the contour line are calculated by linear interpolation.
7. The method for extracting planar features of borehole fractures as described in claim 5, characterized in that, The joint roughness coefficient is calculated based on the discrete coordinate points, specifically as follows: For discontinuous contour lines, a segmented calculation and length-weighted averaging method is used. The line is segmented at the discontinuities, and the root mean square slope parameter is calculated for each continuous segment. The specific expression is as follows: ； in, Indicates the first The root mean square slope parameter of the segment. Indicates the first The arc length of the segment, Indicates the first Data points, Indicates the first Section coordinate, Indicates the first Section value; Based on the root mean square slope parameter, using Calculate the JRC value of each segment, and then perform a weighted average based on the arc length of each segment to obtain the joint roughness coefficient of the overall fracture.
8. A borehole fracture planar feature extraction system, characterized in that, include: The image acquisition module is used to acquire borehole imaging unfolded images and perform pixel-level crack recognition on the borehole imaging unfolded images to obtain binarized crack images. The connected component extraction module is used to perform morphological closing operations on the binarized crack image to connect the fracture regions and extract connected components. The aggregation module is used to obtain the axial coordinates of the centroid of each connected domain, and use the axial coordinates as feature vectors to aggregate connected domains whose spatial positions are within a threshold, as crack instances. The stage processing module is used to perform preliminary fitting of a sine curve with all pixels of the outer contour of the crack instance as observation values to obtain an initial curve, and to determine the distance of all pixels in each connected region to the initial curve. Based on a preset distance threshold or effective pixel ratio, connected regions with distance deviations greater than the threshold or pixels lower than the effective pixel ratio are discarded as abnormal blocks. The output module is used to perform sine curve fitting again using the remaining pixels after removing outliers, and finally extract the amplitude, period, phase and centerline position parameters of the crack instance.
9. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores a program that, when executed by the processor, causes the processor to perform the steps of the borehole fracture plane feature extraction method as described in any one of claims 1 to 7.
10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the borehole fracture plane feature extraction method according to any one of claims 1 to 7.