Method for quantitatively characterizing coal rock borehole fracture parameters

Abstract

The application discloses a method for quantitatively characterizing coal rock borehole fissure parameters, and comprises the following steps: 1, obtaining a fissure binary image of a coal rock borehole; 2, post-processing the fissure binary image to obtain a post-processed fissure binary image; 3, extracting a fissure single-pixel skeleton from the post-processed fissure binary image by using a Zhang-Suen thinning algorithm to obtain a fissure skeleton image; 4, pruning the fissure skeleton on the fissure skeleton image to obtain a pruned fissure skeleton image; 5, performing a broken connection processing on the pruned fissure skeleton image to obtain a final fissure skeleton image; and 6, obtaining fissure parameters based on the final fissure skeleton image and the post-processed fissure binary image. The application firstly performs pseudo-branch pruning processing, and then matches and connects broken skeleton endpoints belonging to different skeleton connected components to obtain a final fissure skeleton and obtain fissure parameters, so that the accuracy of the fissure skeleton and the structure parameters is improved.

Description

Technical Field

[0001] This invention belongs to the field of borehole fracture detection technology, specifically relating to a method for quantitative characterization of borehole fracture parameters in coal and rock. Background Technology

[0002] With the continuous increase in coal mining depth, deep mines generally experience high ground stress, high disturbance, and complex geological environments. Affected by mining operations at the working face, primary fractures, joints, and weak surfaces within the coal and rock mass are prone to expansion and connection, further inducing engineering problems such as roadway deformation, roof falls, spalling, and localized dynamic disasters. Borehole imaging technology can directly acquire images of the borehole walls within the coal and rock mass, providing a direct data source for detecting surrounding rock fractures. Furthermore, by obtaining binary images of fractures from the borehole wall images, fracture structure parameters can be extracted, forming a crucial foundation for evaluating roadway surrounding rock stability and deploying disaster prevention measures.

[0003] Traditional characterization methods rely primarily on empirical observation, making it difficult to perform refined and batch quantitative characterization of fracture structure parameters. With the development of image recognition technology, existing techniques for identifying and extracting parameters from borehole fracture images have formed a basic technical route consisting of image acquisition, image preprocessing, fracture region extraction, and parameter statistics. These typically involve statistically analyzing fracture length, area, and fracture ratio based on binary fracture regions; and converting fracture regions into fracture skeletons to characterize fracture propagation paths, branching morphology, and spatial distribution. However, the fracture skeletons are often not effectively corrected, leading to misidentified and discontinuous fractures, thus affecting the accuracy of subsequent skeleton extraction and structural parameter calculations. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for quantitative characterization of fracture parameters in coal and rock boreholes, which addresses the shortcomings of the prior art. The method involves performing pseudo-branch pruning on the fracture skeleton extracted from the binary fracture image, then matching and connecting the endpoints of the fracture skeleton belonging to different skeleton connected components to obtain the final fracture skeleton, and finally obtaining the fracture parameters based on the final fracture skeleton, thereby improving the accuracy of the fracture skeleton and structural parameters.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for quantitative characterization of fracture parameters in coal and rock boreholes, the method comprising the following steps: Step 1: Obtain a binary image of the fractures in the coal and rock borehole; where the pixel value of the fracture area is 1 and the pixel value of the background area is 0. Step 2: Post-processing of the binary image of the crack to obtain a post-processed binary image of the crack; Step 3: Use the Zhang-Suen thinning algorithm to extract the single-pixel skeleton of the crack from the post-processed binary crack image to obtain the crack skeleton image; Step 4: Prune the fracture skeleton in the fracture skeleton image to obtain the pruned fracture skeleton image; Step 5: Perform disconnection and connectivity processing on the pruned fracture skeleton image to obtain the final fracture skeleton image; Step 6: Obtain the fracture parameters based on the final fracture skeleton image and the post-processed binary fracture image.

[0006] The above-mentioned method for quantitative characterization of coal and rock borehole fracture parameters, further, the specific process of step two is as follows: Step 201: Mark the connected components in the binary image of the crack to obtain multiple candidate connected components of the crack; Step 202: Obtain the number of pixels in each candidate connected region of the crack, and obtain the area of ​​each candidate connected region of the crack; Step 203: Arrange the areas of the K candidate connected regions of the fracture in ascending order, according to G. j =A j+1 / A j This yields the ratio of the j-th adjacent areas; where A j+1 Let A be the area of ​​the (j+1)th candidate connected region of the fracture. j Let A be the area of ​​the j-th candidate connected region of the fracture, and A j+1 Greater than A j , j is a positive integer, and the value of j ranges from 1 to (K-1); K is the number of candidate connected components of the fracture; Step 204: Obtain the largest adjacent area ratio from multiple adjacent area ratios, and then take 1.1 times the area of ​​the previous candidate connected region of the fracture corresponding to the largest adjacent area ratio as the connected region area threshold As; Step 205: Compare the area Ai of the i-th fracture candidate connected region with the connected region area threshold As. If Ai is less than As, the fracture candidate connected region is determined as an isolated noise point and deleted. Otherwise, the fracture candidate connected region is retained as a fracture connected region. Step 206: Repeat step 205 multiple times to complete the judgment of K candidate connected components of the crack and obtain the binary image of the crack after removing the small connected components. Step 207: On the binary image of the fracture after the small connected regions are removed, fill the holes in the fracture connected regions with 1 to obtain the filled binary image of the fracture. Step 208: Use morphological closing operation to perform local connectivity correction on the filled crack binary image to obtain the post-processed crack binary image.

[0007] The above-mentioned method for quantitative characterization of coal and rock borehole fracture parameters, further, the specific process of step four is as follows: Step 401: For any skeleton pixel p in the fracture skeleton image, calculate the sum of the pixel values ​​in its 8-neighborhood and n.p If n p If =1, then the skeleton pixel p is determined to be a skeleton endpoint; Step 402: Repeat step 401 multiple times until all skeleton pixels of the fracture skeleton image are judged, and multiple skeleton endpoints are obtained. Step 403: If the number of skeleton endpoints belonging to the same connected region is greater than 2, then obtain the skeleton length of any two skeleton endpoints. If the two skeleton endpoints belong to the same connected region and the skeleton length is the largest, then the skeleton between the two skeleton endpoints is the main fracture. Step 404: Starting from any remaining skeleton endpoint d, determine the sum of the pixel values ​​in its 8-neighborhood and n. d If n d =1, then the pixels with a value of 1 in its 8 neighborhood are recorded as tracking pixels; Step 405: Using the tracked pixel as the starting point, perform the judgment multiple times according to the method in step 404 until the sum of the pixel values ​​in the 8-neighborhood of the next tracked pixel is greater than or equal to 3. Then, the next tracked pixel is considered as the end point, and the skeleton branch is recorded as the d-th skeleton candidate branch of the main fracture; where d is a positive integer. Step 406: Following the methods in steps 404 and 405, obtain all candidate skeleton branches of the main fracture; Step 407: Following the method in step 403, obtain the skeleton length L between the start and end points of the d-th skeleton candidate branch. d ; Step 408: On the post-processed binary image of the crack, obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the background pixel, and denote it as D. i ; where, the nearest Euclidean distance from the skeleton endpoints to the background pixels in the d-th skeleton candidate branch is denoted as D. f ; Step 409, according to The average crack width W of the d-th skeleton candidate branch is obtained. d ;according to The aspect ratio R of the d-th candidate skeleton branch is obtained. d Where I is the total number of skeleton pixels on the d-th skeleton candidate branch, i is a positive integer, and 1≤i≤I; ε is a minimal constant. Step 4010: Obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the image boundary and denote it as d. i ,if And L d ≤Ls,R dIf ≤Rs, then the candidate branch of the skeleton is determined to be a pseudo branch; otherwise, the d-th candidate branch of the skeleton is a branch crack retention; where d0 is the minimum boundary protection distance, λ is the local crack width adjustment coefficient, δ is the distance compensation amount, Ls is the maximum length threshold of the end branch, and Rs is the branch length-to-width ratio threshold. Step 4011: Keep the pixel values ​​of the pixels in the neighborhood of the end point 8 on the pseudo-branch unchanged, and set the pixel values ​​of other pixels from 1 to zero to complete the pruning of the pseudo-branch. Step 4012: Following the methods in steps 407 to 4010, complete the judgment of all candidate skeleton branches to obtain the pruned slit skeleton image.

[0008] The above-mentioned method for quantitative characterization of fracture parameters in coal and rock boreholes further includes step 403, in which the skeleton lengths of any two skeleton endpoints are obtained. The specific process is as follows: Step 4031: Define the two skeleton endpoints as the left and right endpoints of the skeleton, respectively; Step 4032: Starting from the left endpoint of the skeleton, designate the skeleton pixel adjacent to the left endpoint as the first skeleton pixel and the skeleton pixel adjacent to the first skeleton pixel as the second skeleton pixel. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint of the skeleton, and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then add to the skeleton length. If the first skeleton pixel is diagonally adjacent to the left endpoint of the skeleton, then add the skeleton length. The initial value of the skeleton length is set to 0. Step 4033: Following the method in step 4032, continue until the right end of the skeleton is reached to obtain the skeleton length between the two skeleton endpoints.

[0009] The above-mentioned method for quantitative characterization of coal and rock borehole fracture parameters, further, the specific process of step five is as follows: Step 501: Mark the connected regions of the pruned skeletal image to obtain multiple skeleton connected components; Step 502: Following the method in step 401, determine the skeleton pixels in each skeleton connected component to obtain multiple skeleton endpoints of each skeleton connected component. Step 503: Obtain the minimum Euclidean distance between two skeleton connected components, and denote the minimum Euclidean distance as d. ab And the two corresponding skeleton endpoints are e a and e b and d ab Compare with the distance threshold ds; if 0 <d ab ≤ds, two skeleton endpoints e a and e b As candidate endpoint pairs; Step 504, from skeleton endpoint e a The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. a From the skeleton endpoint e b The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. b ; where ua and ub are both directions pointing from the endpoints of the skeleton into the skeleton; Step 505, according to Obtain the skeleton endpoint e a deviation angle θ a and skeleton endpoint e b deviation angle θ b ; where v ab For the skeleton endpoint e a to skeleton endpoint e b The unit vector between; where θ a and θ b The value range is 0 to 180°; Step 506, if θ a ≤θs, and θ b ≤θs, according to The overall score Q is obtained. ab Where θs is the deviation angle threshold; Step 507: Following the methods of steps 503 to 506, complete the judgment of multiple skeleton endpoints between multiple skeleton connected components and obtain the comprehensive score of multiple candidate endpoint pairs. Step 508: Sort the comprehensive scores corresponding to multiple candidate endpoints from high to low, select the top N candidate endpoint pairs, and use the Bresenham digital line algorithm to connect the candidate endpoint pairs to form a connection path; set the pixel value of the pixel point on the connection path to 1 to obtain the final crack skeleton image; where N is a positive integer and N is less than the number of candidate endpoint pairs.

[0010] The method for quantitative characterization of coal and rock borehole fracture parameters described above, further, includes the following specific steps in step six: Step 601: Mark the connected components of the post-processed binary image of the crack to obtain the connected components of the crack region; Step 602: Count the number of pixels in the connected domains of all fracture regions to obtain the fracture area; obtain the fracture rate based on the ratio of the fracture area to the total number of pixels in the post-processed binary fracture image. Step 603: Based on the final fracture skeleton image, obtain the skeleton length of each main fracture skeleton and branch fracture skeleton according to the skeleton length method in step 403. Step 604: Following the method in step 409, obtain the average fracture width and aspect ratio of the main fracture and the branch fracture.

[0011] Compared with the prior art, the present invention has the following advantages: 1. This invention solves the problem of misidentified cracks and discontinuous cracks, thereby performing pruning and connectivity processing to improve the accuracy of crack skeleton and structural parameters.

[0012] 2. In the post-processing of binary images of fractures, this invention first uses a connected region area thresholding method to remove isolated noise and falsely detected coal and rock texture areas; then, it repairs the internal background areas of the fracture that are not connected to the image boundary by filling holes, only filling the background completely surrounded by the foreground. By filling the internal hole areas with fracture foreground pixels, it can repair the internal gaps of the fracture caused by weak boundaries, local shadows, uneven gray levels, or incomplete recognition, making the fracture area more complete; and it uses morphological closing operations to bridge the gaps in the fracture segments and smooth the fracture edge contours, thereby suppressing noise while maintaining the slender and continuous structure of the coal and rock borehole fractures as much as possible.

[0013] 3. This invention employs the Zhang-Suen thinning algorithm, which, while preserving the fracture topology as much as possible, removes redundant pixels from the fracture region boundaries layer by layer, ultimately compressing the fracture region into a single-pixel skeleton image. This skeleton essentially retains the overall fracture orientation, branching relationships, and connectivity, effectively reflecting the spatial extension characteristics of the fracture. Compared to directly estimating the length of the fracture region, length calculation based on the skeleton more accurately reflects the actual fracture extension path.

[0014] 4. In this invention, the crack skeleton obtained by the Zhang-Suen thinning algorithm is already a single pixel wide. However, due to crack boundary protrusions, local noise, and thinning errors, short pseudo-branches may still exist in the crack skeleton. Simultaneously, due to incomplete crack recognition, the originally continuous crack skeleton may exhibit short-distance breaks. Therefore, this invention first prunes the initial skeleton to obtain a pruned crack skeleton image, and then connects the broken skeletons to obtain the final crack skeleton.

[0015] 5. The present invention sets the skeleton length L. d and aspect ratio R d Frame length L d Used to determine whether the candidate branches of the skeleton are short, R d Used to determine whether the branch is relatively short relative to the local width of the fracture it is in. When L d and R d When both values ​​are relatively small, it indicates that the branch is not only short in absolute length, but its length is also close to the width of the fracture it is located on, making it more likely to be a short burr formed by fracture boundary protrusion or refinement error. To prevent misjudging real fracture branches as refinement burrs, boundary-related branches are protected.

[0016] 6. This invention considers a comprehensive scoring system. The first item is the distance score between the skeleton endpoints, and the second and third items are the directional continuity scores between the two skeleton endpoints, respectively. The smaller the distance between the skeleton endpoints and the smaller the directional deviation angle, the larger the comprehensive score Qab, indicating that the candidate endpoint pair is more likely to belong to the same crack. Therefore, the first N candidate endpoint pairs can be selected for connection according to actual requirements; the number of N can also be adjusted to check the connection effect, so as to meet the distribution of the binary image of the crack and reduce post-processing errors.

[0017] In summary, this invention performs pseudo-branch pruning on the fracture skeleton extracted from the binary fracture image, then matches and connects the endpoints of the fracture skeleton belonging to different skeleton connected components to obtain the final fracture skeleton, and completes the acquisition of fracture parameters based on the final fracture skeleton, thereby improving the accuracy of the fracture skeleton and structural parameters.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention.

[0020] Figure 2 This is a schematic diagram illustrating the transformation from the original coal and rock borehole image to the final fracture skeleton image of the present invention. Detailed Implementation

[0021] like Figure 1 The method for quantitative characterization of fracture parameters in coal and rock boreholes, as shown, includes the following steps: Step 1: Obtain a binary image of the fractures in the coal and rock borehole; where the pixel value of the fracture area is 1 and the pixel value of the background area is 0. Step 2: Post-processing of the binary image of the crack to obtain a post-processed binary image of the crack; Step 3: Use the Zhang-Suen thinning algorithm to extract the single-pixel skeleton of the crack from the post-processed binary crack image to obtain the crack skeleton image; Step 4: Prune the fracture skeleton in the fracture skeleton image to obtain the pruned fracture skeleton image; Step 5: Perform disconnection and connectivity processing on the pruned fracture skeleton image to obtain the final fracture skeleton image; Step 6: Obtain the fracture parameters based on the final fracture skeleton image and the post-processed binary fracture image.

[0022] In this embodiment, the specific process of step two is as follows: Step 201: Mark the connected components in the binary image of the crack to obtain multiple candidate connected components of the crack; Step 202: Obtain the number of pixels in each candidate connected region of the crack, and obtain the area of ​​each candidate connected region of the crack; Step 203: Arrange the areas of the K candidate connected regions of the fracture in ascending order, according to G. j =A j+1 / A j This yields the ratio of the j-th adjacent areas; where A j+1 Let A be the area of ​​the (j+1)th candidate connected region of the fracture. j Let A be the area of ​​the j-th candidate connected region of the fracture, and A j+1 Greater than A j , j is a positive integer, and the value of j ranges from 1 to (K-1); K is the number of candidate connected components of the fracture; Step 204: Obtain the largest adjacent area ratio from multiple adjacent area ratios, and then take 1.1 times the area of ​​the previous candidate connected region of the fracture corresponding to the largest adjacent area ratio as the connected region area threshold As; Step 205: Compare the area Ai of the i-th fracture candidate connected region with the connected region area threshold As. If Ai is less than As, the fracture candidate connected region is determined as an isolated noise point and deleted. Otherwise, the fracture candidate connected region is retained as a fracture connected region. Step 206: Repeat step 205 multiple times to complete the judgment of K candidate connected components of the crack and obtain the binary image of the crack after removing the small connected components. Step 207: On the binary image of the fracture after the small connected regions are removed, fill the holes in the fracture connected regions with 1 to obtain the filled binary image of the fracture. Step 208: Use morphological closing operation to perform local connectivity correction on the filled crack binary image to obtain the post-processed crack binary image.

[0023] In this embodiment, the specific process of step four is as follows: Step 401: For any skeleton pixel p in the fracture skeleton image, calculate the sum of the pixel values ​​in its 8-neighborhood and n. p If n p If =1, then the skeleton pixel p is determined to be a skeleton endpoint; Step 402: Repeat step 401 multiple times until all skeleton pixels of the fracture skeleton image are judged, and multiple skeleton endpoints are obtained. Step 403: If the number of skeleton endpoints belonging to the same connected region is greater than 2, then obtain the skeleton length of any two skeleton endpoints. If the two skeleton endpoints belong to the same connected region and the skeleton length is the largest, then the skeleton between the two skeleton endpoints is the main fracture. Step 404: Starting from any remaining skeleton endpoint d, determine the sum of the pixel values ​​in its 8-neighborhood and n. d If n d =1, then the pixels with a value of 1 in its 8 neighborhood are recorded as tracking pixels; Step 405: Using the tracked pixel as the starting point, perform the judgment multiple times according to the method in step 404 until the sum of the pixel values ​​in the 8-neighborhood of the next tracked pixel is greater than or equal to 3. Then, the next tracked pixel is considered as the end point, and the skeleton branch is recorded as the d-th skeleton candidate branch of the main fracture; where d is a positive integer. Step 406: Following the methods in steps 404 and 405, obtain all candidate skeleton branches of the main fracture; Step 407: Following the method in step 403, obtain the skeleton length L between the start and end points of the d-th skeleton candidate branch. d ; Step 408: On the post-processed binary image of the crack, obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the background pixel, and denote it as D. i ; where, the nearest Euclidean distance from the skeleton endpoints to the background pixels in the d-th skeleton candidate branch is denoted as D. f ; Step 409, according to The average crack width W of the d-th skeleton candidate branch is obtained. d ;according to The aspect ratio R of the d-th candidate skeleton branch is obtained. d Where I is the total number of skeleton pixels on the d-th skeleton candidate branch, i is a positive integer, and 1≤i≤I; ε is a minimal constant. Step 4010: Obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the image boundary and denote it as d. i ,if And L d ≤Ls,R dIf ≤Rs, then the candidate branch of the skeleton is determined to be a pseudo branch; otherwise, the d-th candidate branch of the skeleton is a branch crack retention; where d0 is the minimum boundary protection distance, λ is the local crack width adjustment coefficient, δ is the distance compensation amount, Ls is the maximum length threshold of the end branch, and Rs is the branch length-to-width ratio threshold. Step 4011: Keep the pixel values ​​of the pixels in the neighborhood of the end point 8 on the pseudo-branch unchanged, and set the pixel values ​​of other pixels from 1 to zero to complete the pruning of the pseudo-branch. Step 4012: Following the methods in steps 407 to 4010, complete the judgment of all candidate skeleton branches to obtain the pruned slit skeleton image.

[0024] In this embodiment, the process of obtaining the skeleton length of any two skeleton endpoints in step 403 is as follows: Step 4031: Define the two skeleton endpoints as the left and right endpoints of the skeleton, respectively; Step 4032: Starting from the left endpoint of the skeleton, designate the skeleton pixel adjacent to the left endpoint as the first skeleton pixel and the skeleton pixel adjacent to the first skeleton pixel as the second skeleton pixel. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint of the skeleton, and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then add to the skeleton length. If the first skeleton pixel is diagonally adjacent to the left endpoint of the skeleton, then add the skeleton length. The initial value of the skeleton length is set to 0. Step 4033: Following the method in step 4032, continue until the right end of the skeleton is reached to obtain the skeleton length between the two skeleton endpoints.

[0025] In this embodiment, the specific process of step five is as follows: Step 501: Mark the connected regions of the pruned skeletal image to obtain multiple skeleton connected components; Step 502: Following the method in step 401, determine the skeleton pixels in each skeleton connected component to obtain multiple skeleton endpoints of each skeleton connected component. Step 503: Obtain the minimum Euclidean distance between two skeleton connected components, and denote the minimum Euclidean distance as d.ab And the two corresponding skeleton endpoints are e a and e b and d ab Compare with the distance threshold ds; if 0 <d ab ≤ds, two skeleton endpoints e a and e b As candidate endpoint pairs; Step 504, from skeleton endpoint e a The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. a From the skeleton endpoint e b The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. b ; where ua and ub are both directions pointing from the endpoints of the skeleton into the skeleton; Step 505, according to Obtain the skeleton endpoint e a deviation angle θ a and skeleton endpoint e b deviation angle θ b ; where v ab For the skeleton endpoint e a to skeleton endpoint e b The unit vector between; where θ a and θ b The value range is 0 to 180°; Step 506, if θ a ≤θs, and θ b ≤θs, according to The overall score Q is obtained. ab Where θs is the deviation angle threshold; Step 507: Following the methods of steps 503 to 506, complete the judgment of multiple skeleton endpoints between multiple skeleton connected components and obtain the comprehensive score of multiple candidate endpoint pairs. Step 508: Sort the comprehensive scores corresponding to multiple candidate endpoints from high to low, select the top N candidate endpoint pairs, and use the Bresenham digital line algorithm to connect the candidate endpoint pairs to form a connection path; set the pixel value of the pixel point on the connection path to 1 to obtain the final crack skeleton image; where N is a positive integer and N is less than the number of candidate endpoint pairs.

[0026] In this embodiment, the specific process of step six is ​​as follows: Step 601: Mark the connected components of the post-processed binary image of the crack to obtain the connected components of the crack region; Step 602: Count the number of pixels in the connected domains of all fracture regions to obtain the fracture area; obtain the fracture rate based on the ratio of the fracture area to the total number of pixels in the post-processed binary fracture image. Step 603: Based on the final fracture skeleton image, obtain the skeleton length of each main fracture skeleton and branch fracture skeleton according to the skeleton length method in step 403. Step 604: Following the method in step 409, obtain the average fracture width and aspect ratio of the main fracture and the branch fracture.

[0027] In this embodiment, the binary image of the crack can be output by a deep learning segmentation model (such as DeepLab V3+ model, SAM model, etc.), or it can be obtained by traditional threshold segmentation, edge detection or manual annotation results. The specific method is not limited.

[0028] In this embodiment, a binary image of the fracture is obtained based on traditional threshold segmentation, and the binarization threshold is the average of the maximum and minimum pixel values ​​of the original coal and rock borehole image.

[0029] In this embodiment, the image resolution is 1024×1024 pixels, and the pixel size is set according to the actual calibration results of the imaging device; the pixel size is only used for converting the crack area, skeleton length and crack width from the pixel scale to the actual scale.

[0030] In this embodiment, in step two, isolated noise and false detection areas of coal and rock texture are first removed by using the connected domain area threshold filtering method; then, the internal background area inside the crack that is not connected to the image boundary is repaired by filling holes. Only the background completely surrounded by the foreground is filled. By filling the internal hole area as the crack foreground pixel, the internal gaps inside the crack caused by weak boundaries, local shadows, uneven gray levels or incomplete recognition can be repaired, making the crack area more complete. By using morphological closing operations to close the gaps in fracture segments and smooth the fracture edge contours, the slender and continuous structure of coal and rock borehole fractures can be maintained as much as possible while suppressing noise.

[0031] In this embodiment, the closing operation includes two processes: dilation followed by erosion. Dilation is used to close the gaps in the fracture fragments, while erosion is used to restore the boundary scale of the region and smooth the fracture outline. Based on the fracture discontinuity, image resolution, and boundary smoothing requirements, the morphological structuring element is determined to be a 3×3 elliptical structuring element.

[0032] In this embodiment, the Zhang-Suen thinning algorithm is used in step three. This algorithm, while preserving the topological structure of the crack as much as possible, removes redundant pixels at the boundaries of the crack region layer by layer, ultimately compressing the crack region into a single-pixel skeleton image. This skeleton basically retains the overall direction, branching relationship, and connectivity of the crack, and can effectively reflect the spatial extension characteristics of the crack. Compared with directly estimating the length of the crack region, length calculation based on the skeleton is more consistent with the actual extension path of the crack.

[0033] In this embodiment, steps four and five are included because the crack skeleton obtained by the Zhang-Suen thinning algorithm is already a single pixel wide. However, due to crack boundary protrusions, local noise, and thinning errors, short pseudo-branches may still exist in the crack skeleton. Simultaneously, due to incomplete crack identification, the originally continuous crack skeleton may exhibit short-distance breaks. Therefore, the initial skeleton is first pruned to obtain a pruned crack skeleton image, and then the broken skeleton is connected to obtain the final crack skeleton.

[0034] In this embodiment, in step four, for any skeleton pixel p in the fracture skeleton image, the pixel values ​​within its 8-neighborhood and n are counted. p This allows us to determine the skeleton endpoints and obtain the main fracture skeleton and skeleton candidate branches based on the skeleton endpoints. This facilitates the subsequent identification and pruning of pseudo-branches in the skeleton candidate branches.

[0035] In this embodiment, it should be noted that in step 403, the skeleton length along the effective skeleton adjacency relationship between any two skeleton endpoints is calculated, and the skeleton path between the skeleton endpoints with the longest skeleton length is determined as the main fracture path of the connected component. When there are multiple longest paths with the same skeleton length, the union of all parallel longest paths is taken as the main fracture path region.

[0036] In this embodiment, it should be noted that if the number of skeleton endpoints belonging to the same connected region is equal to 2, then the skeleton between the two skeleton endpoints is a secondary primary fracture, which does not have any skeleton candidate branches and does not require pruning; or a primary fracture whose skeleton length is less than the maximum skeleton length is also a secondary primary fracture.

[0037] In this embodiment, the skeleton length L d The pixel length obtained by accumulating segments along the skeleton path is not a straight line distance; the skeleton length L d Used to determine whether the candidate branches of the skeleton are short, R d Used to determine whether the branch is relatively short relative to the local width of the fracture it is in. When L d and R dWhen both values ​​are relatively small, it indicates that the branch is not only short in absolute length, but its length is also close to the width of the fracture it is located on, making it more likely to be a short burr formed by fracture boundary protrusion or refinement error. To prevent misjudging real fracture branches as refinement burrs, boundary-related branches are protected.

[0038] In this embodiment, it should be noted that if If the candidate skeleton branch is determined to be a boundary-related branch, it will be retained to avoid accidentally deleting real cracks that may continue to extend beyond the current image range.

[0039] In this embodiment, the two skeleton endpoints e a and e b Termini belonging to different skeleton connected components, and endpoints within the same skeleton connected component, do not enter the candidate endpoint pair in step 503, so as to avoid forming closed paths within the same skeleton. To prevent misconnecting skeleton connected components that are close to each other but have different directions, several skeleton pixels are traced inward from the two candidate connection endpoints along their respective skeleton connected components. Principal component analysis is performed on the coordinates of the skeleton pixels, and the direction of the first principal component is taken as the principal direction of the skeleton pixels near the skeleton endpoints. a with u b All directions are from the endpoints of the skeleton to the interior of the skeleton.

[0040] In this embodiment, the comprehensive score Q ab The first item is the score for the distance between the skeleton endpoints, and the second and third items are the scores for the directional continuity between the two skeleton endpoints, respectively. The smaller the distance between the skeleton endpoints and the smaller the directional deviation angle, the higher the overall score Q. ab The larger the value, the higher the probability that the candidate endpoints belong to the same crack.

[0041] In this embodiment, the skeleton endpoint distance and skeleton endpoint direction deviation angle are both prerequisites that candidate endpoint pairs must meet. The comprehensive score is used to further filter and sort candidate endpoint pairs that meet the prerequisites, but it cannot replace the above-mentioned judgment conditions. Only when a candidate endpoint pair meets all the aforementioned conditions and the comprehensive score meets the requirements will it proceed to the subsequent one-to-one matching connection.

[0042] In this embodiment, in actual use, the comprehensive scores of multiple candidate endpoint pairs are sorted in descending order. If the comprehensive scores are the same, the endpoint pairs with shorter skeleton endpoint distances are selected for connection. In addition, a one-to-one matching method is used to match sequentially. The same skeleton endpoint participates in the connection only once in the same connection process to avoid a skeleton endpoint connecting multiple skeleton segments in different directions at the same time. After the connection is completed, the skeleton endpoints and skeleton connected components are re-identified, and there are no new connectable endpoint pairs. Secondly, the connection stage only adds skeleton pixels, without deleting existing skeleton pixels in the pruned rift skeleton image.

[0043] In this embodiment, it should be noted that the thresholds are not specifically limited and can be adjusted in practice to obtain the best results.

[0044] In this embodiment, the maximum length threshold Ls of the terminal branch is the rounded-up value of the median skeleton length of all skeleton candidate branches; in actual use, it can also be adjusted between the rounded-up values ​​of the minimum skeleton length and the median skeleton length; the aspect ratio threshold Rs of the branch is the rounded-up value of the median aspect ratio of all skeleton candidate branches; in actual use, it can also be adjusted between the rounded-up values ​​of the minimum aspect ratio and the median aspect ratio. In practical applications, the deletion effect can be verified by applying different thresholds. While ensuring that genuine fracture branches are not mistakenly deleted, the threshold that effectively removes false branches with the smallest error should be selected.

[0045] In this embodiment, the minimum constant ε is 10. -6 Used only to prevent W d When the denominator is close to 0, it is a numerical stability parameter, not a fracture physical parameter.

[0046] In this embodiment, the minimum boundary protection distance d0 is 1 to 5 pixels, the local crack width adjustment coefficient λ is 1 to 1.5, and the distance compensation amount δ is 1 to 5 pixels. It should be noted that these values ​​can be adjusted adaptively in actual use. Under the premise of ensuring that the real crack branches are not mistakenly deleted, the threshold that can effectively remove false branches and minimize error should be selected.

[0047] In this embodiment, it should be noted that the distance threshold ds can be any minimum Euclidean distance between two skeleton connected components obtained in step 503. By verifying the effect of different thresholds, the threshold that can connect broken components with the smallest error is selected.

[0048] In this embodiment, it should be noted that when obtaining the minimum Euclidean distance between two skeleton connected components, the Euclidean distance between any two skeleton endpoints between the two skeleton connected components is first obtained, and then the Euclidean distances between the two skeleton endpoints are sorted from largest to smallest to obtain the minimum Euclidean distance.

[0049] In this embodiment, it should be noted that the deviation angle threshold θs is 15° to 30°. In practice, this can be adjusted to obtain the best results.

[0050] In this embodiment, for Figure 2The original coal and rock borehole image shown in image a is further optimized with the following settings: binarization threshold of 177, maximum length threshold Ls of the end branch of 25, Rs of the branch aspect ratio of 4.0, minimum boundary protection distance d0 of 2, local fracture width adjustment coefficient λ of 1.5, distance compensation amount δ of 1, and distance threshold ds of 22.

[0051] In this embodiment, as Figure 2 Image a shows the original image of a coal and rock borehole. After binarization, the image is obtained... Figure 2 The binary image of the crack in b, after processing in step two, yields... Figure 2 c is the post-processed binary image of the crack; For the post-processed binary image of the crack, after processing in step three, the following is obtained: Figure 2 Image of the fracture skeleton of d; from Figure 2 On image d, one primary fracture and two secondary primary fractures can be observed; then, after processing in step four, the fracture skeleton image after pruning is obtained, as shown below. Figure 2 As shown in e; from Figure 2 e shows that the main fracture in the middle is discontinuous. After processing in step five, we obtain... Figure 2 The final fracture skeleton image of f.

[0052] In summary, this invention performs pseudo-branch pruning on the fracture skeleton extracted from the binary fracture image, then matches and connects the endpoints of the fracture skeleton belonging to different skeleton connected components to obtain the final fracture skeleton, and completes the acquisition of fracture parameters based on the final fracture skeleton, thereby improving the accuracy of the fracture skeleton and structural parameters.

[0053] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for quantitative characterization of fracture parameters in coal and rock boreholes, characterized in that, The method includes the following steps: Step 1: Obtain a binary image of the fractures in the coal and rock borehole; where the pixel value of the fracture area is 1 and the pixel value of the background area is 0. Step 2: Post-processing of the binary image of the crack to obtain a post-processed binary image of the crack; Step 3: Use the Zhang-Suen thinning algorithm to extract the single-pixel skeleton of the crack from the post-processed binary crack image to obtain the crack skeleton image; Step 4: Prune the fracture skeleton in the fracture skeleton image to obtain the pruned fracture skeleton image; Step 5: Perform disconnection and connectivity processing on the pruned fracture skeleton image to obtain the final fracture skeleton image; Step 6: Obtain the fracture parameters based on the final fracture skeleton image and the post-processed binary fracture image.

2. The method for quantitative characterization of coal and rock borehole fracture parameters according to claim 1, characterized in that: The specific process of step two is as follows: Step 201: Mark the connected components in the binary image of the crack to obtain multiple candidate connected components of the crack; Step 202: Obtain the number of pixels in each candidate connected region of the crack, and obtain the area of ​​each candidate connected region of the crack; Step 203: Arrange the areas of the K candidate connected regions of the fracture in ascending order, according to G. j =A j+1 / A j This yields the ratio of the j-th adjacent areas; where A j+1 Let A be the area of ​​the (j+1)th candidate connected region of the fracture. j Let A be the area of ​​the j-th candidate connected region of the fracture, and A j+1 Greater than A j , j is a positive integer, and the value of j ranges from 1 to (K-1); K is the number of candidate connected components of the fracture; Step 204: Obtain the largest adjacent area ratio from multiple adjacent area ratios, and then take 1.1 times the area of ​​the previous candidate connected region of the fracture corresponding to the largest adjacent area ratio as the connected region area threshold As; Step 205: Compare the area Ai of the i-th fracture candidate connected region with the connected region area threshold As. If Ai is less than As, the fracture candidate connected region is determined as an isolated noise point and deleted. Otherwise, the fracture candidate connected region is retained as a fracture connected region. Step 206: Repeat step 205 multiple times to complete the judgment of K candidate connected components of the crack and obtain the binary image of the crack after removing the small connected components. Step 207: On the binary image of the fracture after the small connected regions are removed, fill the holes in the fracture connected regions with 1 to obtain the filled binary image of the fracture. Step 208: Use morphological closing operation to perform local connectivity correction on the filled crack binary image to obtain the post-processed crack binary image.

3. The method for quantitative characterization of coal and rock borehole fracture parameters according to claim 2, characterized in that: The specific process of step four is as follows: Step 401: For any skeleton pixel p in the fracture skeleton image, calculate the sum of the pixel values ​​in its 8-neighborhood, denoted as n. p If n p If =1, then the skeleton pixel p is determined to be a skeleton endpoint; Step 402: Repeat step 401 multiple times until all skeleton pixels of the fracture skeleton image are judged, and multiple skeleton endpoints are obtained. Step 403: If the number of skeleton endpoints belonging to the same connected region is greater than 2, then obtain the skeleton length of any two skeleton endpoints. If the two skeleton endpoints belong to the same connected region and the skeleton length is the largest, then the skeleton between the two skeleton endpoints is the main fracture. Step 404: Starting from any remaining skeleton endpoint d, determine the sum of the pixel values ​​in its 8-neighborhood and n. d If n d =1, then the pixels with a value of 1 in its 8 neighborhood are recorded as tracking pixels; Step 405: Using the tracked pixel as the starting point, perform the judgment multiple times according to the method in step 404 until the sum of the pixel values ​​in the 8-neighborhood of the next tracked pixel is greater than or equal to 3. Then, the next tracked pixel is considered as the end point, and the skeleton branch is recorded as the d-th skeleton candidate branch of the main fracture; where d is a positive integer. Step 406: Following the methods in steps 404 and 405, obtain all candidate skeleton branches of the main fracture; Step 407: Following the method in step 403, obtain the skeleton length L between the start and end points of the d-th skeleton candidate branch. d ; Step 408: On the post-processed binary image of the crack, obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the background pixel, and denote it as D. i ; where, the nearest Euclidean distance from the skeleton endpoints to the background pixels in the d-th skeleton candidate branch is denoted as D. f ; Step 409, according to The average crack width W of the d-th skeleton candidate branch is obtained. d ;according to The aspect ratio R of the d-th candidate skeleton branch is obtained. d Where I is the total number of skeleton pixels on the d-th skeleton candidate branch, i is a positive integer, and 1≤i≤I; ε is a minimal constant. Step 4010: Obtain the nearest Euclidean distance from the i-th skeleton pixel in the d-th skeleton candidate branch to the image boundary and denote it as d. i ,if And L d ≤Ls,R d If ≤Rs, then the candidate branch of the skeleton is determined to be a pseudo branch; otherwise, the d-th candidate branch of the skeleton is a branch crack retention; where d0 is the minimum boundary protection distance, λ is the local crack width adjustment coefficient, δ is the distance compensation amount, Ls is the maximum length threshold of the end branch, and Rs is the branch length-to-width ratio threshold. Step 4011: Keep the pixel values ​​of the pixels in the neighborhood of the end point 8 on the pseudo-branch unchanged, and set the pixel values ​​of other pixels from 1 to zero to complete the pruning of the pseudo-branch. Step 4012: Following the methods in steps 407 to 4010, complete the judgment of all candidate skeleton branches to obtain the pruned slit skeleton image.

4. The method for quantitative characterization of coal and rock borehole fracture parameters according to claim 3, characterized in that: In step 403, the skeleton lengths of any two skeleton endpoints are obtained. The specific process is as follows: Step 4031: Define the two skeleton endpoints as the left and right endpoints of the skeleton, respectively; Step 4032: Starting from the left endpoint of the skeleton, designate the skeleton pixel adjacent to the left endpoint as the first skeleton pixel and the skeleton pixel adjacent to the first skeleton pixel as the second skeleton pixel. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is horizontally adjacent to the left endpoint and the second skeleton pixel is vertically adjacent to the first skeleton pixel, then increase the skeleton length by 2. If the first skeleton pixel is vertically adjacent to the left endpoint of the skeleton, and the second skeleton pixel is horizontally adjacent to the first skeleton pixel, then add to the skeleton length. ; If the first skeleton pixel is diagonally adjacent to the left endpoint of the skeleton, then add to the skeleton length. The initial value of the skeleton length is set to 0. Step 4033: Following the method in step 4032, continue until the right end of the skeleton is reached to obtain the skeleton length between the two skeleton endpoints.

5. The method for quantitative characterization of coal and rock borehole fracture parameters according to claim 3, characterized in that: The specific process of step five is as follows: Step 501: Mark the connected regions of the pruned skeletal image to obtain multiple skeleton connected components; Step 502: Following the method in step 401, determine the skeleton pixels in each skeleton connected component to obtain multiple skeleton endpoints of each skeleton connected component. Step 503: Obtain the minimum Euclidean distance between two skeleton connected components, and denote the minimum Euclidean distance as d. ab And the two corresponding skeleton endpoints are e a and e b and d ab Compare with the distance threshold ds; if 0 <d ab ≤ds, two skeleton endpoints e a and e b As candidate endpoint pairs; Step 504, from skeleton endpoint e a The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. a From the skeleton endpoint e b The coordinates of five skeleton pixels are obtained by tracing along the skeleton connected components, and principal component analysis is performed on these five skeleton pixels to obtain the first principal component direction u. b ; where ua and ub are both directions pointing from the endpoints of the skeleton into the skeleton; Step 505, according to Obtain the skeleton endpoint e a deviation angle θ a and skeleton endpoint e b deviation angle θ b ; where v ab For the skeleton endpoint e a to skeleton endpoint e b The unit vector between; where θ a and θ b The value range is 0 to 180°; Step 506, if θ a ≤θs, and θ b ≤θs, according to The overall score Q is obtained. ab Where θs is the deviation angle threshold; Step 507: Following the methods of steps 503 to 506, complete the judgment of multiple skeleton endpoints between multiple skeleton connected components and obtain the comprehensive score of multiple candidate endpoint pairs. Step 508: Sort the comprehensive scores corresponding to multiple candidate endpoints from high to low, select the top N candidate endpoint pairs, and use the Bresenham digital line algorithm to connect the candidate endpoint pairs to form a connection path; set the pixel value of the pixel point on the connection path to 1 to obtain the final crack skeleton image; where N is a positive integer and N is less than the number of candidate endpoint pairs.

6. The method for quantitative characterization of coal and rock borehole fracture parameters according to claim 3, characterized in that: The specific process of step six is ​​as follows: Step 601: Mark the connected components of the post-processed binary image of the crack to obtain the connected components of the crack region; Step 602: Count the number of pixels in the connected domains of all fracture regions to obtain the fracture area; obtain the fracture rate based on the ratio of the fracture area to the total number of pixels in the post-processed binary fracture image. Step 603: Based on the final fracture skeleton image, obtain the skeleton length of each main fracture skeleton and branch fracture skeleton according to the skeleton length method in step 403; Step 604: Following the method in step 409, obtain the average fracture width and aspect ratio of the main fracture and the branch fracture.

Images