Rock core fracture connection method based on multi-constraint dynamic fusion

Abstract

The invention discloses a rock core fracture connection method based on multi-constraint dynamic fusion, which belongs to the field of digital rock core analysis, and finally realizes fracture breakpoint connection from fracture information original CT (Computed Tomography) image data through fracture feature enhancement and extraction, fracture geometric feature analysis and multi-modal constraint evaluation. And a set of complete and logically coherent fracture connection process is formed. According to the scheme, each breakpoint is evaluated through indexes such as spatial proximity and spatial colinearity, fracture breakpoints caused by CT image gray threshold segmentation can be effectively identified, fracture areas are effectively integrated, and continuity and form consistency of fractures in a rock core are remarkably improved; moreover, after fracture breakpoints are connected, the connectivity of the overall fracture network in the rock core is enhanced, a basis can be provided for follow-up evaluation of the permeability and stability of the rock core and a reservoir, and the method has wide popularization and practical application value.

Description

Technical Field

[0001] This invention belongs to the field of digital core analysis, specifically relating to a core fracture connection method based on multi-constraint dynamic fusion. Background Technology

[0002] As the primary pathway for fluid migration in geological reservoirs, fracture networks directly influence the enrichment of resources such as natural gas hydrates due to their spatial connectivity, geometric morphology, and dynamic evolution. Therefore, accurately identifying and analyzing the structural characteristics of fracture networks is of significant guiding importance for the safe and efficient development of natural gas hydrates.

[0003] Due to geological structure and environmental influences, reservoir fracture structures often exhibit complex morphology and strong spatial heterogeneity. Traditional methods such as thin section analysis and electron microscopy struggle to accurately characterize the three-dimensional topology of fractures, leading to significant errors in the prediction of parameters such as reservoir permeability. X-ray computed tomography (CT), a non-invasive, high-resolution imaging technique, is widely used in digital core modeling. It enables three-dimensional visualization and reconstruction of fracture networks, effectively acquiring microstructural information within the core and providing a crucial data foundation for fracture identification and analysis.

[0004] For example, the invention patent with publication number [CN118506104A] proposes a method for identifying and predicting rock fractures. Based on CT scan images of siltstone sample slices, it identifies rock fractures using a large SAM model and uses a convolutional neural network to train a GCN model to predict fracture propagation paths. The invention patent with publication number [CN112085825A] proposes a fractal quantification method for three-dimensional rock fractures. Through three-dimensional reconstruction of rock CT scan images, it obtains data such as the number of connected fractures, fracture volume, and average fracture volume. It then uses a head-and-tail segmentation method to iteratively segment the fracture dataset, calculates the image fractal dimension, and achieves fractal quantification of rock fractures.

[0005] While the aforementioned methods provide quantitative morphological analysis of core fractures, limitations in the spatial resolution of CT instruments and the volume effect of scanned images inevitably lead to misidentification of small fractures with low development or narrow widths as two or more independent fractures at narrow locations. Furthermore, the presence of inclusions (such as mineral crystal grains or clay filling) within the fractures can also cause similar misidentifications in CT scans, thus disrupting the integrity and connectivity of the identified fractures and hindering a true reflection of the fracture distribution patterns in the core. Existing methods do not address this issue.

[0006] Therefore, there is an urgent need to propose a method for connecting the breakpoints of originally intact fractures in CT images, so as to achieve more accurate quantitative analysis of fracture network morphology and thus meet the needs of studying the migration law of geological reservoir fluids. Summary of the Invention

[0007] This invention provides a core fracture connection method based on multi-constraint dynamic fusion, which integrates interactive processing methods and morphological operation mechanisms. It extracts fracture orientation features through principal component analysis and combines geometric collinearity verification and spatial proximity screening to achieve effective connection of damaged fracture networks identified by CT scans.

[0008] This invention is achieved using the following technical solution: a core fracture connection method based on multi-constraint dynamic fusion. Starting from the original CT image data of fracture information, the method proceeds through fracture feature enhancement and extraction, fracture geometric feature analysis, multi-modal constraint evaluation, and finally fracture breakpoint connection, forming a complete and logically coherent fracture connection process. Specifically, it includes the following steps: Step A, Crack Feature Enhancement and Extraction: HSV color space conversion combined with morphological closing operation optimization strategy is used to achieve fine segmentation of crack regions in CT images, so as to obtain the precise shape and location of cracks; Step B, Geometric analysis of crack features: Based on the spatial continuity measurement of contour pixels, the crack length is selected; Principal component analysis (PCA) is used to determine the principal direction vector of the crack, and the slope of the crack is calculated based on the vector to quickly compare the directional similarity of two cracks.

[0009] Step C, Multimodal Constraint Evaluation: For two cracks with similar slopes, further screen candidate crack pairs that are spatially close. If the Euclidean distance between any two endpoints of the two cracks is less than a threshold, they are considered to be spatially close and may need to be connected. Then, through collinearity judgment, determine whether the crack pair needs to be connected to ensure that the connected cracks are geometrically consistent.

[0010] Step D, Crack Breakpoint Connection: Determine the spatial proximity and geometric collinearity of the base cracks. Optimize the distance between crack endpoints to calculate a connection path with the minimum error in geometric space, and then connect the crack breakpoints.

[0011] Compared with the prior art, the advantages and positive effects of the present invention are as follows: This scheme analyzes the geometric features of fractures and evaluates each breakpoint in the fracture network using indices such as spatial proximity and spatial collinearity. It can effectively identify non-real fracture breakpoints caused by human factors such as grayscale thresholding in CT images, and connects these non-real fracture breakpoints in the image, effectively integrating the fractured area and significantly improving the continuity and morphological consistency of fractures inside the core. In addition, after connecting the fracture breakpoints, the overall connectivity of the fracture network inside the core is enhanced, which can provide a basis for subsequent evaluation of the permeability and stability of the core and reservoir. Attached Figure Description

[0012] Figure 1 This is a flowchart of the fracture connection method according to an embodiment of the present invention; Figure 2 The image shows a comparison of connectivity differences before and after crack extraction in an embodiment of the present invention. (a) is the original grayscale image, and (b) is the binarized image after crack extraction. The circles are marked crack breakpoints. Figure 3 This is a schematic diagram of the two-dimensional effect of the core sample before and after fracture connection in an embodiment of the present invention; Figure 4 These are three-dimensional images of the rock core sample before and after the fracture connection in an embodiment of the present invention. (a) and (c) are images of the fracture point before connection, and (b) and (d) are images of the fracture point after connection. Figure 5 These are three-dimensional images of the core sample before and after the fracture connection in an embodiment of the present invention. (a) and (c) are images of the fracture point before connection, and (b) and (d) are images of the fracture point after connection. Detailed Implementation

[0013] To better understand the above-described objects, features, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. Many specific details are set forth in the following description to provide a thorough understanding of the present invention; however, the present invention may be practiced in other ways than those described herein, and therefore, the present invention is not limited to the specific embodiments disclosed below.

[0014] A core fracture connection method based on multi-constraint dynamic fusion, combined with Figure 1 As shown, it includes the following steps: Step A: Crack Feature Enhancement and Extraction; First, the cracks are separated from the background to provide a clear input image for subsequent geometric attribute analysis.

[0015] (1) The input image is converted from the RGB color space to the HSV color space to avoid the interference of brightness fluctuations in the RGB space on crack extraction. In the HSV color space, H (hue) represents the color type, S (saturation) represents the color purity, and V (brightness) represents the brightness of the color. The HSV color space can better separate color information, thereby improving the accuracy of crack extraction. Through color space conversion, the algorithm can more effectively identify the target crack region.

[0016] (2) In the HSV color space, threshold segmentation is performed by setting the color range to extract the crack region and generate a binary mask. Color threshold segmentation is a key step in crack extraction. By adjusting the color range, the target crack can be effectively separated from the background, which can adapt to the crack extraction needs in different scenarios.

[0017] (3) Extract the crack profile from the binary mask, filter out excessively short cracks, avoid misjudging pores or noise as tiny cracks, and obtain the precise shape and location of the crack. Profile extraction is a prerequisite for crack attribute calculation and can provide accurate geometric information for subsequent processing.

[0018] Step B: Analysis of crack geometric features; Quantitative morphological analysis of the extracted cracks is performed to obtain geometric parameters such as crack length and direction, providing data support for subsequent evaluation of crack break points.

[0019] (1) Calculation of crack length The crack length is calculated by accumulating the distances between contour points. Crack length filtering is based on the spatial continuity measurement of contour pixels, and its core is to quantify the linear extension of the crack in the two-dimensional plane. This parameter is achieved through geometric topological analysis, that is, by accumulating the Euclidean distances between adjacent points: (1) Where L is the crack length, n is the number of contour points, (x i y i (x) represents the coordinates of the i-th contour point. i+1 y i+1 Let be the coordinates of the (i+1)th contour point. For example, if the crack contour consists of points P0 → P1 → P2 → ... → Pn, then the total length is the sum of the lengths of the line segments between all adjacent points.

[0020] (2) Determine the principal direction vector of the fracture Principal component analysis (PCA) is a dimensionality reduction method in statistics that projects data onto a new coordinate system through orthogonal transformation, maximizing the variance of the data along the first principal component direction. Subsequent principal components then explain the remaining variance in turn. PCA is used to determine the principal direction vector of the fracture, and the slope is calculated based on this vector.

[0021] The set of crack profile points is most widely distributed in space along its extension direction. PCA analyzes the data covariance matrix to find the direction with the largest variance (i.e., the principal direction), which is taken as the extension direction of the crack. For the crack profile point set {pi}, PCA is used to calculate its principal direction vector v and centroid μ (average coordinates). The principal direction vector v represents the main extension direction of the crack, and the largest eigenvector is selected as the principal direction vector of the crack. The calculation process is as follows: Formula for calculating the center of mass: (2) Represent each point as a deviation vector relative to the centroid: (3) By constructing the covariance matrix Vectorization reflects the distribution of data in various directions: (4) in, The centroid of the set of points representing the fracture profile reflects the geometric center of the fracture. Represents the x-axis coordinate of the centroid. The y-axis coordinate represents the centroid, and N represents the total number of crack profile points. Represents the two-dimensional coordinates of the i-th contour point This represents the deviation vector of the i-th contour point relative to the centroid, used to eliminate the influence of the centroid's position. This represents the deviation between the x-coordinate of the i-th contour point and the x-coordinate of the centroid. Var(x) represents the deviation between the y-coordinate of the i-th contour point and the y-coordinate of the centroid. Var(x) represents the variance of the x-coordinate (the degree of dispersion of the data in the x-axis direction). Var(y) represents the variance of the y-coordinate (the degree of dispersion of the data in the y-axis direction). Cov(x, y) represents the covariance of x and y (the coherence of the data in the x and y directions).

[0022] For covariance matrix Perform eigenvalue decomposition to obtain a set of eigenvalues. And the corresponding eigenvector V. The eigenvalue represents the variance along the direction of the corresponding eigenvector. (5) The eigenvector V main =[dx, dy], It represents the quadratic form of vector v on the covariance matrix, reflecting the direction of the maximum variance of the data.

[0023] (3) Determine the slope of the fracture To visually represent the fracture direction, we define the angle θ between the fracture direction and the X-axis, with its tangent value being the slope: (6) Specifically, dx represents the principal direction vector V. main The x-component is extracted from the largest eigenvector calculated by PCA, satisfying dx 2 +dy 2 =1, together with dx, determines the direction. For the principal direction vectors of the two fractures, V1=[d x1 d y1 ] and V2=[d x2 d y2 Calculate the angle difference between the two cracks: (7) in, This indicates the angle between the principal directions of the first fracture. This indicates the included angle of the principal direction of the second fracture.

[0024] By limiting the angular difference between the principal directions of two fractures, it is ensured that only fractures with the same direction are connected, avoiding the connection of fractures with excessively large angular differences. The bidirectionality of the straight direction is considered (e.g., 0° and 180° are equivalent). Two fractures may actually have similar directions, but are mistakenly judged as dissimilar simply because their direction vectors are in opposite directions. The angular difference is corrected to a minimum. (8) If the corrected angle difference is less than the angle threshold, the two cracks are considered to be similar in direction.

[0025] Step C: Multimodal constraint evaluation; evaluate each breakpoint in the rift network, and use indicators such as spatial proximity and spatial collinearity to screen out non-realistic breakpoints caused by problems in the CT images themselves, i.e., the locations where rift connections need to be made.

[0026] (1) Spatial proximity screening If the Euclidean distance between any two endpoints of two cracks is less than a threshold, they are considered spatially close and may need to be connected. For each crack, its principal direction is determined using PCA, and all contour points are projected along the principal direction. The points with the smallest and largest projections are taken as endpoints. For the four endpoints of two cracks (two endpoints for each crack), the Euclidean distance between all endpoint pairs is calculated: (9) The minimum distance among the four endpoints is taken. If this value is less than a set threshold, the fracture pair is considered to satisfy spatial proximity. The determination of this threshold needs to consider the spatial resolution of the CT image, the fracture type, and the natural distribution characteristics of fractures in the core, so as to accurately distinguish between non-real breakpoints that need to be connected and independent fractures that do not need to be connected. If the threshold is too large, it may cause unrelated fractures to be connected incorrectly; if it is too small, it may miss fractures that should actually be connected.

[0027] (2) Determination of spatial collinearity Collinearity is assessed to verify the geometric collinearity of the two cracks, ensuring that the connected cracks form a smooth, straight line extension in space. Specifically, this embodiment determines whether the two cracks are collinear by calculating the straight-line distance from a point to its extension line. A crack is defined as an infinitely extending straight line defined by the principal direction vector v and the centroid μ. If the crack is curved, the local principal direction needs to be recalculated using 8-10 consecutive contour points near the break point, and the extension line redefined to avoid using the overall principal direction, which could cause the straight line to deviate from the local trend of the crack. The perpendicular distance from the nearest endpoint of the other crack to this straight line is calculated. The centroid μ is considered as the origin O(x0, y0), and the principal direction vector v is considered as the unit direction vector v = (v... x v y Define the endpoint P(x) p y p The calculation steps are as follows: (10) Projected length (scalar): (11) Projection vector (parallel components): (12) Vertical vector (vertical component): (13) Vertical distance: (14) The projection vector represents the closest position of point P along the straight line; the perpendicular vector represents the line connecting the projection point to the original point, indicating the distance between the point and the line; the perpendicular distance d quantifies the degree of vertical deviation of point P from the line, and the smaller the value, the better the collinearity.

[0028] The distances from the endpoints of each of the two fractures to the straight line connecting them are calculated. If this distance is less than a set collinearity threshold (determined by combining fracture morphology characteristics and CT image quality to ensure that the geometry of the connected fractures conforms to the natural geological extension patterns), the two fractures are considered collinear. This parameter ensures that the connected fractures extend geometrically naturally. For example, if the endpoints are close but deviate from the main extension line, the connection may result in an abrupt broken line.

[0029] Step D: Connection of fracture breakpoints Based on the above judgments regarding spatial proximity and geometric collinearity of the fractures, a connection path with minimal error in geometric space is calculated by optimizing the distance between fracture endpoints, and fracture breakpoints are then connected. During rendering, this optimal path is rendered on an independent connection mask using preset RGB color channels, thus ensuring the visual representation of the connection features. This connection structure is integrated into the original image through pixel-level overlay, preserving the topological continuity of the fracture relationships while achieving morphological optimization of the fracture network. Figure 3 , Figure 4 and Figure 5 As shown, this is a comparison of the two-dimensional and three-dimensional effects before and after the connection.

[0030] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments for application in other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A core fracture connection method based on multi-constraint dynamic fusion, characterized in that, Includes the following steps: Step A: Crack Feature Enhancement and Extraction; HSV color space conversion combined with morphological closing operation optimization strategy is used to achieve fine segmentation of crack regions in CT images, so as to obtain the precise shape and location of cracks; Step B: Analysis of fracture geometric features; Quantitative morphological analysis of the extracted fractures is performed to obtain the geometric parameters of the fractures, including fracture length, fracture direction, and fracture slope; Step C: Multimodal constraint evaluation; evaluate each breakpoint in the fracture network, and select the locations where fractures need to be connected by combining fracture spatial proximity, directional approximation and spatial collinearity indices. Step D: Connecting fracture breakpoints; Based on the judgment of fracture spatial proximity, directional similarity and spatial collinearity, by optimizing the distance between fracture endpoints, a connection path with the minimum error in geometric space is calculated, and fracture breakpoints are connected.

2. The core fracture connection method based on multi-constraint dynamic fusion according to claim 1, characterized in that: Step A is specifically implemented in the following manner: (1) Convert the input image from RGB color space to HSV color space; (2) In the HSV color space, set the color range for threshold segmentation, extract the crack area and generate a binary mask; (3) Extract the crack outline from the binary mask, filter out short cracks, and obtain the precise shape and position of the crack.

3. The core fracture connection method based on multi-constraint dynamic fusion according to claim 1, characterized in that: Step B, in obtaining the geometric parameters of the crack, specifically includes the following steps: (1) Calculate the crack length; combine the extracted cracks to obtain its contour point set, and accumulate the Euclidean distance between adjacent contour points through geometric topology analysis; (2) Determine the principal direction vector of the crack; Combine the contour point set and calculate the principal direction vector and centroid of the crack by PCA. The principal direction vector represents the main extension direction of the crack. Select the largest eigenvector as the principal direction vector of the crack. (3) Determine the slope of the fracture; define the angle θ between the line containing the fracture and the x-axis, and let the principal direction vectors of the two fractures be V1=[d x1 d y1 ] and V2=[d x2 d y2 This allows us to obtain the angle difference between the two cracks; The angle difference is corrected to limit the angular difference between the two principal directions of the fracture. The corrected angle difference is expressed as: in, This indicates the angle difference between the two cracks. This represents the corrected angle difference; if the corrected angle difference is less than the angle threshold, the two cracks are considered to be similar in direction.

4. The core fracture connection method based on multi-constraint dynamic fusion according to claim 1, characterized in that: In step B, when determining the direction of the crack: The formula for calculating the centroid is as follows: Each contour point is represented as a deviation vector relative to the centroid: By constructing the covariance matrix Vectorization reflects the distribution of data in various directions: Where Var(x) represents the variance of the x-coordinate, Var(y) represents the variance of the y-coordinate, and Cov(x, y) represents the covariance of x and y. For covariance matrix Perform eigenvalue decomposition to obtain a set of eigenvalues. And the corresponding eigenvector V, where the eigenvalue represents the variance along the direction of the corresponding eigenvector, the principal direction vector of the fracture is the eigenvector with the largest eigenvalue, defined as: The feature vector v main This is the main direction of crack extension.

5. The core fracture connection method based on multi-constraint dynamic fusion according to claim 1, characterized in that: Step C specifically includes the following steps: (1) Spatial proximity screening; First, determine the endpoints of the cracks. For any two cracks, calculate the Euclidean distance between all the endpoints and find the minimum value among the calculated distances. If this value is less than a set threshold, then the two cracks are considered to satisfy the spatial proximity. (2) Determination of spatial collinearity; For any two cracks, treat one of them as an infinitely extended straight line defined by the principal direction vector and the centroid, and calculate the perpendicular distance from the nearest endpoint of the other crack to this straight line. Where (x0, y0) are the coordinates of the origin of the centroid, and (v x v y Let be the unit direction vector, and define the endpoint P(x) p y p ), where d is the vertical distance, representing the degree of vertical deviation of the endpoint from the line. The smaller the vertical distance value, the better the collinearity.

Images