Clastic rock sedimentary bedding type rapid identification method and system
By combining high-resolution scanning and image processing technologies with mathematical algorithms, the sedimentary bedding types of clastic rocks can be quickly identified, solving the problems of low identification efficiency and insufficient accuracy in existing technologies, and achieving efficient and accurate automated identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-19
- Publication Date
- 2026-04-07
AI Technical Summary
In existing technologies, the identification of sedimentary bedding in clastic rocks mainly relies on manual observation, which leads to low identification efficiency, difficulty in ensuring accuracy, and easy omissions, increasing the burden on staff.
High-resolution scanning equipment was used to acquire core column images. Image preprocessing and meshing were performed, followed by edge detection and skeletonization. Key morphological parameters were calculated, multi-dimensional feature vectors were constructed, and similarity calculation and ranking algorithms were used to identify sedimentary bedding types.
It improves the efficiency and accuracy of identifying sedimentary bedding in clastic rocks, reduces manual workload, enhances the efficiency of geological surveys and core descriptions, and achieves automated and highly accurate identification.
Smart Images
Figure CN121811073A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of petroleum geological exploration technology, specifically to a method and system for rapid identification of sedimentary bedding types in clastic rocks. Background Technology
[0002] Sedimentary bedding, as an important indicator of changes in the sedimentary environment, is a typical feature of clastic rocks. It reflects the conditions of the sedimentary environment, the source of materials, and energy changes during the sedimentary process. Identifying the sedimentary bedding of clastic rocks helps us interpret geological history and analyze the sedimentary environment, especially in analyzing the depth and energy conditions of sedimentary water bodies and determining the transport methods and sedimentary processes of sediments. It plays an irreplaceable role and is one of the important contents of sedimentological research. At the same time, the morphology and structure of sedimentary bedding have a significant impact on the porosity, permeability, and other physical properties of clastic reservoirs. Homogeneous massive bedding sandstones have significantly better physical properties than heterogeneous parallel bedding, cross-bedding, and ripple bedding sandstones. Therefore, identifying sedimentary bedding types is of great value for evaluating the three-dimensional spatial connectivity of reservoirs.
[0003] Currently, most existing methods for identifying clastic sedimentary bedding types rely on manual observation, resulting in low efficiency, difficulty in ensuring accuracy, significant increase in workload for staff, and the potential for omissions in the identification results. Therefore, it is necessary to design a rapid identification method and system for clastic sedimentary bedding types. Summary of the Invention
[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to better and more effectively solve the problem that the identification of clastic rock sedimentary bedding is mostly based on manual observation, resulting in low identification efficiency, difficulty in ensuring accuracy, and a significant increase in the workload of staff. Furthermore, omissions are also prone to occur in the identification results. This invention provides a rapid identification method and system for clastic rock sedimentary bedding types. It achieves the function of extracting sample sedimentary bedding morphological parameters from high-resolution core scan images using image processing technology and quantifying the similarity between sample sedimentary bedding and standard sedimentary bedding using mathematical algorithms for accurate identification of sedimentary bedding types. This not only improves the identification efficiency of clastic rock sedimentary bedding but also ensures the identification accuracy, thereby improving the efficiency of geological workers' field investigations and indoor core description work.
[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for rapid identification of sedimentary bedding types in clastic rocks includes the following steps: Step A: Use a scanning device to perform high-resolution scanning of the core column image and obtain the acquired image. Then, perform image preprocessing and image meshing on the acquired image to obtain the sample image. Step B: Perform edge detection on the sample image and obtain sample texture lines, then perform skeletonization on the sample texture lines and obtain a set of sample texture curves. Step C: Quantitatively calculate the key morphological parameters of each sample curve in the sample striation curve set and obtain the calculation results of the morphological parameters of the sample striation curve respectively. Step D: Construct multidimensional feature vectors for all samples based on the calculation results of the morphological parameters of the laminar curves. Step E: Calculate the similarity between the multidimensional feature vector of each sample and the standard multidimensional feature vector, and obtain the similarity calculation results respectively; Step F: Sort the similarity calculation results of the laminar curves of each sample according to the sedimentary bedding type and obtain an initial similarity sort list. Then, merge and sort each initial similarity sort list to obtain a final similarity sort list. Step G: Obtain the sedimentary bedding type identification results of each sample's stratigraphic curve based on the final similarity ranking list.
[0006] The aforementioned method for rapid identification of clastic rock sedimentary bedding types includes step A, which involves using a scanning device to perform high-resolution scanning of the core column image to obtain the acquired image, followed by image preprocessing and image meshing to obtain sample images. The specific steps are as follows. Step A1: Use a scanning device to perform high-resolution scanning on the core column image and obtain the acquired image. The scanning device includes a core scanner and a high-resolution digital camera. The core column image is a two-dimensional unfolded image or a 360° panoramic image of the core column under uniform illumination conditions. The high-resolution scanning is such that the image resolution can clearly display millimeter-level textures. Step A2 involves performing image preprocessing and image meshing on the acquired images to obtain sample images. The specific steps are as follows: Step A21 involves preprocessing the acquired image to obtain a preprocessed image. This preprocessing includes color correction and homogenization, image enhancement, and geometric correction. Color correction and homogenization eliminate shadows and color differences caused by uneven lighting and ensure consistent brightness across the entire core surface. Image enhancement uses contrast stretching and Laplacian sharpening algorithms to highlight the contours of the bedding planes. Geometric correction corrects image distortion caused by the shooting angle and ensures accurate horizontal and vertical proportions. The specific steps of image enhancement are as follows. Step A211: The acquired image is contrast-stretched using a contrast stretching algorithm, as shown in formula (1). (1) in, Image stretched for contrast in coordinates The pixel grayscale value at that location, Image before contrast stretching in coordinates The pixel grayscale value at that location, and These are the row index and column index of the pixel, respectively. and These are the maximum and minimum gray values of the image before contrast stretching, respectively. and These are the maximum and minimum gray values within the target gray range, respectively. Step A212: The acquired image is sharpened using the Laplacian sharpening algorithm, as shown in formula (2). (2) in, The image after sharpening. The image before sharpening. This is the sharpening intensity factor. For the Laplace operator; Step A22: Perform image meshing on the preprocessed image to obtain a meshed image. Specifically, image meshing involves establishing a Cartesian coordinate system based on the preprocessed image, defining the origin as the first pixel at the bottom left of the image, and defining the horizontal direction... The axis increases from left to right, while the vertical direction... The axes increase from bottom to top.
[0007] The aforementioned method for rapid identification of clastic rock sedimentary bedding types includes step B, which involves edge detection of the sample image to obtain sample bedding lines, followed by skeletonization of the sample bedding lines to obtain a set of sample bedding curves. The specific steps are as follows. Step B1 involves performing edge detection on the sample image and obtaining sample texture lines. The edge detection includes Gaussian filtering for noise reduction, finding image gradients, non-maximum suppression, and dual-threshold boundary tracking. The specific steps are as follows. Step B11: Perform Gaussian filtering on the sample image to denoise and obtain the denoised sample image. The Gaussian filtering denoising involves assigning weights to points at different positions in the convolution kernel according to a Gaussian distribution, as shown in formula (3). (3) in, For coordinates Gaussian filter kernel weights at the specified location. The standard deviation of the Gaussian distribution; Step B12 involves calculating the image gradient of the denoised sample image and obtaining the image gradient calculation result. Specifically, this is done using... sobel The operator calculates the gradient magnitude and gradient direction of the denoised sample image, as shown in formula (4). ; ; ; (4) in, for x Orientation gradient, for y Orientation gradient, For gradient magnitude, For the gradient direction, The denoised sample image to be processed; Step B13: Based on the image gradient calculation results, perform non-extreme suppression on the edges of the denoised sample image and suppress the sample image. Specifically, quantize the gradient direction to the 0° direction and compare the current pixel with two adjacent pixels in the gradient direction. If the gradient magnitude of the current pixel is not the maximum value, then suppress it; if the gradient magnitude of the current pixel is the maximum value, then do not suppress it. Step B14 involves performing dual-threshold boundary tracking on the suppressed sample image to obtain sample texture lines. The specific steps are as follows: Step B141: Set the upper threshold and the lower threshold. If the number of pixels in the suppressed sample image is greater than the upper threshold, it is called a strong boundary. If the number of pixels in the suppressed sample image is less than the lower threshold, it is not a boundary. Step B142: Take the pixel between the upper threshold and the lower threshold as the false boundary, and then determine whether there are any points in the eight neighborhoods of the pixels in the false boundary that are greater than the upper threshold. If they exist, they are the boundary; if they do not exist, they are discarded. Step B2 involves skeletalizing the sample texture lines and obtaining a set of sample texture curves. Specifically, the sample texture lines are processed into continuous curves with a width of one pixel, and then the continuous curves are converted into vector data to obtain sample texture curves, thereby obtaining a set of sample texture curves.
[0008] The aforementioned method for rapid identification of clastic sedimentary bedding types, in step C, involves quantitatively calculating the key morphological parameters of each sample curve in the sample bedding curve set and obtaining the calculation results of the sample bedding curve morphological parameters. These key morphological parameters include bedding dip angle, bedding curvature, bedding interface morphology, and texture features. The specific steps are as follows. Step C1: Calculate the lamellar dip angle of the sample lamellar curve, as shown in formula (5). (5) in, The dip angle of the laminae. The height of a pixel. This represents the horizontal distance of a pixel in the coordinate system. Step C2: Calculate the laminar curvature of the sample laminar curve, as shown in formula (6). (6) in, For the laminar curvature, This is a laminar curve function; Step C3: Calculate the layer interface morphology of the sample striation curve. Specifically, the layer interface morphology is determined by mathematically fitting the vector curves of the top and bottom interfaces of the layer system and judging whether the morphology is flat, grooved, or wavy, as shown in formula (7). (7) in, This is a curve function representing the morphology of the layered interface. , , , , and All are fitting coefficients; Step C4: Calculate the texture features of the sample lamina curves, wherein the texture features include mean features, variance features, contrast features, homogeneity features, and entropy features. The specific steps are as follows. Step C41: Calculate the mean characteristic of the sample striation curve, as shown in formula (8). (8) in, The mean value feature of the gray-level co-occurrence matrix of the sample lamina curve. and These represent the number of rows and columns of the gray-level co-occurrence matrix, respectively. In direction distance Position and pixel position The grayscale value of the other pixel that appears in a pair. For distance, It is the direction angle; Step C42: Calculate the variance characteristic of the sample striation curve, as shown in formula (9). (9) in, Let V be the variance characteristic of the gray-level co-occurrence matrix of the sample striation curve. The average position of the sample striation curve; Step C43: Calculate the contrast feature of the sample lamellar curve, as shown in formula (10). (10) in, The gray-level co-occurrence matrix contrast feature of the sample lamina curve; Step C44: Calculate the homogeneity characteristic of the sample lamellar curve, as shown in formula (11). (11) in, The gray-level co-occurrence matrix homogeneity feature of the sample lamina curve; Step C45: Calculate the entropy feature of the sample lamina curve, as shown in formula (12). (12) in, The entropy feature of the gray-level co-occurrence matrix of the sample lamina curve. The sample ridge curve at the pixel location The grayscale value at that location.
[0009] In the aforementioned method for rapid identification of clastic sedimentary bedding types, step D involves constructing multidimensional feature vectors for all samples based on the calculated morphological parameters of the bedding curves, as shown in formula (13). (13) in, This is the multidimensional feature vector of the sample.
[0010] In the aforementioned method for rapid identification of clastic sedimentary bedding types, step E involves calculating the similarity between the multidimensional feature vector of each sample and the standard multidimensional feature vector, and obtaining the similarity calculation results. The similarity calculation includes Euclidean distance similarity, cosine similarity, and... Tanimoto The specific steps for calculating coefficient similarity and dynamic time warping similarity are as follows: Step E1 involves calculating the Euclidean distance similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (14). (14) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Euclidean distance similarity between them For the multidimensional feature vector of the sample, For sample multidimensional feature vectors The i3D eigenvalues Standard multidimensional feature vector The j 3D eigenvalues; Step E2 involves calculating the cosine similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. The cosine similarity is used to compare the directional differences between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (15). (15) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Cosine similarity between them; Step E3, compare the sample multidimensional feature vector with the standard multidimensional feature vector. Tanimoto The coefficient similarity is calculated, wherein... Tanimoto The coefficient similarity is used to measure the degree of similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (16). (16) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Between Tanimoto Coefficient similarity; Step E4: Calculate the dynamic time warping similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. This dynamic time warping similarity is used to compare laminar thickness sequences or curves that are similar in shape but differ in length due to different deposition rates, as shown in formula (17). ; ; (17) in, It is a distance matrix. This is the cumulative distance matrix. For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Dynamic time-warped similarity between them For the alignment path of the first One alignment point pair, The total number of alignment point pairs, The index of the alignment point pair.
[0011] The aforementioned method for rapid identification of clastic rock sedimentary bedding types includes step F, which involves sorting the similarity calculation results of the lamellar curves of each sample according to the sedimentary bedding type to obtain an initial similarity ranking list. Then, each initial similarity ranking list is merged and sorted to obtain a final similarity ranking list. The specific steps are as follows. Step F1: Sort the similarity calculation results of the laminar curves of each sample according to the sedimentary bedding type and obtain an initial similarity ranking list, wherein the initial similarity ranking list includes a sublist of Euclidean distance similarity ranking, a sublist of cosine similarity ranking, and so on. Tanimoto The coefficient similarity sorting sublist and the dynamic time warping similarity sorting sublist, wherein the sedimentary bedding types include horizontal bedding, parallel bedding, oblique bedding, trough cross-bedding, platy cross-bedding, wavy bedding and massive bedding; Step F2 involves merging and sorting each initial similarity ranking list to obtain a final similarity ranking list. Specifically, the merging and sorting process employs... Borda The specific steps for merging and sorting using the counting method are as follows. Step F21: For each sedimentary bedding type, sort the sublists by Euclidean distance similarity and by cosine similarity. Tanimoto The positions in the coefficient similarity ranking sublist and the dynamic time warping similarity ranking sublist are assigned scores respectively. If the sedimentary bedding type ranks first in the sublist, it receives N points; if it ranks second, it receives N-1 points, and so on. If the sedimentary bedding type ranks N in the sublist, it receives 1 point, as shown in formula (18). ; ; ; (18) in, The score for sedimentary bedding type in the Euclidean distance similarity sublist. The number of sedimentary bedding types, This represents the ranking of sedimentary bedding types within the Euclidean distance similarity sublist. The score for sedimentary bedding type in the cosine similarity sorted sublist. This represents the ranking of sedimentary bedding types within the cosine similarity sorting sublist. For sedimentary bedding types in Tanimoto Scores in the coefficient similarity sorted sublist For sedimentary bedding types in Tanimoto Ranked in the coefficient similarity sublist The score for sedimentary bedding type in the dynamic time-regularity similarity sublist. The ranking of sedimentary bedding types in the dynamic time-regularity similarity sublist; Step F22: Sum the scores of each sedimentary bedding type in each sublist and obtain the fusion ranking score. Then, sort all the fusion ranking scores in descending order to obtain the final similarity ranking list. The process of obtaining the fusion ranking score is shown in formula (19). (19) in, The fusion and sorting of sedimentary bedding types, It is a sedimentary bedding type.
[0012] In the aforementioned method for rapid identification of sedimentary bedding types in clastic rocks, step G involves obtaining the sedimentary bedding type identification results of each sample's lamellar curve based on the final similarity ranking list. Specifically, the sedimentary bedding type ranked first in the final similarity ranking list is taken as the sedimentary bedding type identification result of the sample's lamellar curve.
[0013] A rapid identification system for clastic rock sedimentary bedding types includes an image acquisition module, an image processing module, a parameter calculation module, a feature vector construction module, a similarity calculation module, a similarity ranking module, and a bedding type output module. The image acquisition module uses a scanning device to perform high-resolution scanning of core columns to obtain acquired images, then performs image preprocessing and image meshing on the acquired images to obtain sample images. The image processing module performs edge detection on the sample images to obtain sample bedding lines, then performs skeletonization processing on the sample bedding lines to obtain a set of sample bedding curves. The parameter calculation module quantitatively calculates and analyzes the key morphological parameters of each sample curve in the set of sample bedding curves. The module obtains the morphological parameters of the sample laminar curves; the feature vector construction module constructs multidimensional feature vectors for each sample based on the morphological parameters of all sample laminar curves; the similarity calculation module calculates the similarity between each sample multidimensional feature vector and the standard multidimensional feature vector and obtains the similarity calculation results; the similarity sorting module sorts the similarity calculation results of each sample laminar curve according to the sedimentary bedding type and obtains an initial similarity sorting list, then merges and sorts each initial similarity sorting list to obtain a final similarity sorting list; the bedding type output module obtains the sedimentary bedding type identification results of each sample laminar curve based on the final similarity sorting list.
[0014] The beneficial effects of this invention are as follows: This invention provides a rapid identification method and system for sedimentary bedding types in clastic rocks. First, a scanning device is used to perform high-resolution scanning of a core column image to obtain acquired images. Then, the acquired images undergo image preprocessing and image meshing to obtain sample images. Next, edge detection is performed on the sample images to obtain sample bedding lines. Then, the sample bedding lines are processed into a skeleton to obtain a set of sample bedding curves. Subsequently, the key morphological parameters of each sample curve in the sample bedding curve set are quantitatively calculated, and the morphological parameter calculation results are obtained. Then, the morphological parameter calculation results of all sample bedding curves are used to construct sample multidimensional feature vectors. Then, the similarity between each sample multidimensional feature vector and the standard multidimensional feature vector is calculated, and the similarity calculation results are obtained. Finally, the similarity calculation results of each sample bedding curve are categorized according to sedimentary bedding type. The method and system for rapid identification of sedimentary bedding types in clastic rocks are developed. The initial similarity ranking lists are then sorted and merged to obtain final similarity ranking lists. Based on these final similarity ranking lists, the sedimentary bedding type identification results for each sample's lamellar curve are obtained. This effectively realizes the function of extracting morphological parameters of sedimentary bedding from high-resolution core scan images using image processing technology and quantifying the similarity between sample sedimentary bedding and standard sedimentary bedding using mathematical algorithms for accurate identification of sedimentary bedding types. This not only improves the identification efficiency of clastic rock sedimentary bedding but also ensures the identification accuracy, improving the efficiency of field investigations and indoor core description work for geological workers. It also facilitates the digital logging of geological outcrops and cores, exhibiting significant advantages such as high automation, high identification accuracy, and independence from expert experience. Attached Figure Description
[0015] Picture 1 This is an overall flowchart of a method for rapid identification of clastic rock sedimentary bedding types according to the present invention; Picture 2 This is a schematic diagram of image meshing according to the present invention; Picture 3 This is a schematic diagram of the standard sedimentary bedding type of the present invention; Picture 4 This is a schematic diagram of the sample deposition layering of the present invention, wherein... Picture 4 a represents massive bedding. Picture 4 b represents massive bedding. Picture 4 c represents parallel bedding; Picture 5 This is a schematic diagram of the sample deposition layering of the present invention, wherein... Picture 5 d represents cross-bedding. Picture 5 e represents cross-bedding. Picture 5 f represents cross-bedding. Detailed Implementation
[0016] The present invention will now be further described with reference to the accompanying drawings.
[0017] like Picture 1 As shown, the present invention provides a method for rapid identification of clastic rock sedimentary bedding types, comprising the following steps: Step A involves using a scanning device to perform high-resolution scanning of the core column image and acquiring the acquired image. The acquired image is then preprocessed and meshed to obtain the sample image. The specific steps are as follows: Step A1: Use a scanning device to perform high-resolution scanning on the core column image and obtain the acquired image. The scanning device includes a core scanner and a high-resolution digital camera. The core column image is a two-dimensional unfolded image or a 360° panoramic image of the core column under uniform illumination conditions. The high-resolution scanning is such that the image resolution can clearly display millimeter-level textures. Step A2 involves performing image preprocessing and image meshing on the acquired images to obtain sample images. The specific steps are as follows: Step A21 involves preprocessing the acquired image to obtain a preprocessed image. This preprocessing includes color correction and homogenization, image enhancement, and geometric correction. Color correction and homogenization eliminate shadows and color differences caused by uneven lighting and ensure consistent brightness across the entire core surface. Image enhancement uses contrast stretching and Laplacian sharpening algorithms to highlight the contours of the bedding planes. Geometric correction corrects image distortion caused by the shooting angle and ensures accurate horizontal and vertical proportions. The specific steps of image enhancement are as follows. Step A211: The acquired image is contrast-stretched using a contrast stretching algorithm, as shown in formula (1). (1) in, Image stretched for contrast in coordinates The pixel grayscale value at that location, Image before contrast stretching in coordinates The pixel grayscale value at that location, and These are the row index and column index of the pixel, respectively. and These are the maximum and minimum gray values of the image before contrast stretching, respectively. and These are the maximum and minimum gray values within the target gray range, respectively. Step A212: The acquired image is sharpened using the Laplacian sharpening algorithm, as shown in formula (2). (2) in, The image after sharpening. The image before sharpening. This is the sharpening intensity factor. For the Laplace operator; like Picture 2 As shown, step A22 involves meshing the preprocessed image to obtain a meshed image. Specifically, this meshing involves establishing a Cartesian coordinate system based on the preprocessed image, defining the origin as the first pixel at the bottom left of the image, and defining the horizontal direction... The axis increases from left to right, while the vertical direction... The axes increase from bottom to top.
[0018] Step B involves performing edge detection on the sample image to obtain sample texture lines, then performing skeletonization on the sample texture lines to obtain a set of sample texture curves. The specific steps are as follows. Step B1 involves performing edge detection on the sample image and obtaining sample texture lines. The edge detection includes Gaussian filtering for noise reduction, finding image gradients, non-maximum suppression, and dual-threshold boundary tracking. The specific steps are as follows. Step B11: Perform Gaussian filtering on the sample image to denoise and obtain the denoised sample image. The Gaussian filtering denoising involves assigning weights to points at different positions in the convolution kernel according to a Gaussian distribution, as shown in formula (3). (3) in, For coordinates Gaussian filter kernel weights at the specified location. The standard deviation of the Gaussian distribution; Step B12 involves calculating the image gradient of the denoised sample image and obtaining the image gradient calculation result. Specifically, this is done using... sobel The operator calculates the gradient magnitude and gradient direction of the denoised sample image, as shown in formula (4). ; ; ; (4) in, for x Orientation gradient, for y Orientation gradient, For gradient magnitude, For the gradient direction, The denoised sample image to be processed; Step B13: Based on the image gradient calculation results, perform non-extreme suppression on the edges of the denoised sample image and suppress the sample image. Specifically, quantize the gradient direction to the 0° direction and compare the current pixel with two adjacent pixels in the gradient direction. If the gradient magnitude of the current pixel is not the maximum value, then suppress it; if the gradient magnitude of the current pixel is the maximum value, then do not suppress it. Step B14 involves performing dual-threshold boundary tracking on the suppressed sample image to obtain sample texture lines. The specific steps are as follows: Step B141: Set the upper threshold and the lower threshold. If the number of pixels in the suppressed sample image is greater than the upper threshold, it is called a strong boundary. If the number of pixels in the suppressed sample image is less than the lower threshold, it is not a boundary. Step B142: Take the pixel between the upper threshold and the lower threshold as the false boundary, and then determine whether there are any points in the eight neighborhoods of the pixels in the false boundary that are greater than the upper threshold. If they exist, they are the boundary; if they do not exist, they are discarded. Step B2 involves skeletalizing the sample texture lines and obtaining a set of sample texture curves. Specifically, the sample texture lines are processed into continuous curves with a width of one pixel, and then the continuous curves are converted into vector data to obtain sample texture curves, thereby obtaining a set of sample texture curves.
[0019] Step C involves quantitatively calculating the key morphological parameters of each sample curve in the sample striation curve set and obtaining the calculation results for the morphological parameters of the sample striation curve. These key morphological parameters include striation dip angle, striation curvature, layer interface morphology, and texture features. The specific steps are as follows. Step C1: Calculate the lamellar dip angle of the sample lamellar curve, as shown in formula (5). (5) in, The dip angle of the laminae. The height of a pixel. This represents the horizontal distance of a pixel in the coordinate system. Step C2: Calculate the laminar curvature of the sample laminar curve, as shown in formula (6). (6) in, For the laminar curvature, This is a laminar curve function; Step C3: Calculate the layer interface morphology of the sample striation curve. Specifically, the layer interface morphology is determined by mathematically fitting the vector curves of the top and bottom interfaces of the layer system and judging whether the morphology is flat, grooved, or wavy, as shown in formula (7). (7) in, This is a curve function representing the morphology of the layered interface. , , , , and All are fitting coefficients; Step C4: Calculate the texture features of the sample lamina curves, wherein the texture features include mean features, variance features, contrast features, homogeneity features, and entropy features. The specific steps are as follows. Step C41: Calculate the mean characteristic of the sample striation curve, as shown in formula (8). (8) in, The mean value feature of the gray-level co-occurrence matrix of the sample lamina curve. and These represent the number of rows and columns of the gray-level co-occurrence matrix, respectively. In direction distance Position and pixel position The grayscale value of the other pixel that appears in a pair. For distance, It is the direction angle; Step C42: Calculate the variance characteristic of the sample striation curve, as shown in formula (9). (9) in, Let V be the variance characteristic of the gray-level co-occurrence matrix of the sample striation curve. The average position of the sample striation curve; Step C43: Calculate the contrast feature of the sample lamellar curve, as shown in formula (10). (10) in, The gray-level co-occurrence matrix contrast feature of the sample lamina curve; Step C44: Calculate the homogeneity characteristic of the sample lamellar curve, as shown in formula (11). (11) in, The gray-level co-occurrence matrix homogeneity feature of the sample lamina curve; Step C45: Calculate the entropy feature of the sample lamina curve, as shown in formula (12). (12) in, The entropy feature of the gray-level co-occurrence matrix of the sample lamina curve. The sample ridge curve at the pixel location The grayscale value at that location.
[0020] Step D: Construct multidimensional feature vectors for all samples by calculating the morphological parameters of the laminar curves, as shown in formula (13). (13) in, This is the multidimensional feature vector of the sample.
[0021] Step E involves calculating the similarity between each sample's multidimensional feature vector and the standard multidimensional feature vector, and obtaining the similarity calculation results. The similarity calculation includes Euclidean distance similarity, cosine similarity, and... Tanimoto The specific steps for calculating coefficient similarity and dynamic time warping similarity are as follows: Step E1 involves calculating the Euclidean distance similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (14). (14) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Euclidean distance similarity between them For the multidimensional feature vector of the sample, For sample multidimensional feature vectors The i 3D eigenvalues Standard multidimensional feature vector The j 3D eigenvalues; Step E2 involves calculating the cosine similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. The cosine similarity is used to compare the directional differences between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (15). (15) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Cosine similarity between them; Step E3, compare the sample multidimensional feature vector with the standard multidimensional feature vector. Tanimoto The coefficient similarity is calculated, wherein... Tanimoto The coefficient similarity is used to measure the degree of similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (16). (16) in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Between Tanimoto Coefficient similarity; Step E4: Calculate the dynamic time warping similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. This dynamic time warping similarity is used to compare laminar thickness sequences or curves that are similar in shape but differ in length due to different deposition rates, as shown in formula (17). ; ; (17) in, It is a distance matrix. This is the cumulative distance matrix. For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Dynamic time-warped similarity between them For the alignment path of the first One alignment point pair, The total number of alignment point pairs, The index of the alignment point pair.
[0022] Step F involves sorting the similarity calculation results of the stratigraphic curves for each sample according to the sedimentary bedding type and obtaining an initial similarity ranking list. Then, each initial similarity ranking list is merged and sorted to obtain a final similarity ranking list. The specific steps are as follows. like Picture 3 As shown, in step F1, the similarity calculation results of the laminar curves of each sample are sorted according to the sedimentary bedding type, and an initial similarity sorting list is obtained. This initial similarity sorting list includes a Euclidean distance similarity sorting sublist, a cosine similarity sorting sublist, and... Tanimoto The coefficient similarity sorting sublist and the dynamic time warping similarity sorting sublist, wherein the sedimentary bedding types include horizontal bedding, parallel bedding, oblique bedding, trough cross-bedding, platy cross-bedding, wavy bedding and massive bedding; Step F2 involves merging and sorting each initial similarity ranking list to obtain a final similarity ranking list. Specifically, the merging and sorting process employs... Borda The specific steps for merging and sorting using the counting method are as follows. Step F21: For each sedimentary bedding type, sort the sublists by Euclidean distance similarity and by cosine similarity. TanimotoThe positions in the coefficient similarity ranking sublist and the dynamic time warping similarity ranking sublist are assigned scores respectively. If the sedimentary bedding type ranks first in the sublist, it receives N points; if it ranks second, it receives N-1 points, and so on. If the sedimentary bedding type ranks N in the sublist, it receives 1 point, as shown in formula (18). ; ; ; (18) in, The score for sedimentary bedding type in the Euclidean distance similarity sublist. The number of sedimentary bedding types, This represents the ranking of sedimentary bedding types within the Euclidean distance similarity sublist. The score for sedimentary bedding type in the cosine similarity sorted sublist. This represents the ranking of sedimentary bedding types within the cosine similarity sorting sublist. For sedimentary bedding types in Tanimoto Scores in the coefficient similarity sorted sublist For sedimentary bedding types in Tanimoto Ranked in the coefficient similarity sublist The score for sedimentary bedding type in the dynamic time-regularity similarity sublist. The ranking of sedimentary bedding types in the dynamic time-regularity similarity sublist; Step F22: Sum the scores of each sedimentary bedding type in each sublist and obtain the fusion ranking score. Then, sort all the fusion ranking scores in descending order to obtain the final similarity ranking list. The process of obtaining the fusion ranking score is shown in formula (19). (19) in, The fusion and sorting of sedimentary bedding types, It is a sedimentary bedding type.
[0023] like Picture 4 and Picture 5 As shown, in step G, the sedimentary bedding type identification result of each sample's stratigraphic curve is obtained according to the final similarity ranking list. Specifically, the sedimentary bedding type ranked first in the final similarity ranking list is taken as the sedimentary bedding type identification result of the sample's stratigraphic curve.
[0024] A rapid identification system for clastic rock sedimentary bedding types includes an image acquisition module, an image processing module, a parameter calculation module, a feature vector construction module, a similarity calculation module, a similarity ranking module, and a bedding type output module. The image acquisition module uses a scanning device to perform high-resolution scanning of core columns to obtain acquired images, then performs image preprocessing and image meshing on the acquired images to obtain sample images. The image processing module performs edge detection on the sample images to obtain sample bedding lines, then performs skeletonization processing on the sample bedding lines to obtain a set of sample bedding curves. The parameter calculation module quantitatively calculates and analyzes the key morphological parameters of each sample curve in the set of sample bedding curves. The module obtains the morphological parameters of the sample laminar curves; the feature vector construction module constructs multidimensional feature vectors for each sample based on the morphological parameters of all sample laminar curves; the similarity calculation module calculates the similarity between each sample multidimensional feature vector and the standard multidimensional feature vector and obtains the similarity calculation results; the similarity sorting module sorts the similarity calculation results of each sample laminar curve according to the sedimentary bedding type and obtains an initial similarity sorting list, then merges and sorts each initial similarity sorting list to obtain a final similarity sorting list; the bedding type output module obtains the sedimentary bedding type identification results of each sample laminar curve based on the final similarity sorting list.
[0025] In summary, the present invention provides a method and system for rapid identification of sedimentary bedding types in clastic rocks. First, a scanning device is used to perform high-resolution scanning of a core column image to obtain acquired images. Then, the acquired images undergo image preprocessing and image meshing to obtain sample images. Next, edge detection is performed on the sample images to obtain sample laminar lines. Then, the sample laminar lines are skeletonized to obtain a set of sample laminar curves. Subsequently, the key morphological parameters of each sample curve in the set of sample laminar curves are quantitatively calculated, and the morphological parameter calculation results are obtained. Then, multi-dimensional feature vectors are constructed from the morphological parameter calculation results of all sample laminar curves. Then, similarity calculations are performed between each sample multi-dimensional feature vector and the standard multi-dimensional feature vector, and the similarity calculation results are obtained. Finally, the similarity calculation results of each sample laminar curve are further processed according to the sedimentary bedding type. The method and system for rapid identification of sedimentary bedding types in clastic rocks effectively utilize image processing technology to extract morphological parameters of sedimentary bedding from high-resolution core scan images and employ mathematical algorithms to quantify the similarity between sample sedimentary bedding and standard sedimentary bedding for accurate identification of sedimentary bedding types. This not only improves the identification efficiency of sedimentary bedding in clastic rocks but also ensures the identification accuracy, thereby enhancing the efficiency of field investigations and indoor core description work for geological workers. It also facilitates the digital logging of geological outcrops and cores, exhibiting significant advantages such as high automation, high identification accuracy, and independence from expert experience.
[0026] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for rapid identification of sedimentary bedding types in clastic rocks, characterized in that: Includes the following steps, Step A: Use a scanning device to perform high-resolution scanning of the core column image and obtain the acquired image. Then, perform image preprocessing and image meshing on the acquired image to obtain the sample image. Step B: Perform edge detection on the sample image and obtain sample texture lines, then perform skeletonization on the sample texture lines and obtain a set of sample texture curves. Step C: Quantitatively calculate the key morphological parameters of each sample curve in the sample striation curve set and obtain the calculation results of the morphological parameters of the sample striation curve respectively. Step D: Construct multidimensional feature vectors for all samples based on the calculation results of the morphological parameters of the laminar curves. Step E: Calculate the similarity between the multidimensional feature vector of each sample and the standard multidimensional feature vector, and obtain the similarity calculation results respectively; Step F: Sort the similarity calculation results of the laminar curves of each sample according to the sedimentary bedding type and obtain an initial similarity sort list. Then, merge and sort each initial similarity sort list to obtain a final similarity sort list. Step G: Obtain the sedimentary bedding type identification results of each sample's stratigraphic curve based on the final similarity ranking list.
2. The method for rapid identification of clastic rock sedimentary bedding types according to claim 1, characterized in that: Step A involves using a scanning device to perform high-resolution scanning of the core column image and acquiring the acquired image. The acquired image is then preprocessed and meshed to obtain the sample image. The specific steps are as follows: Step A1: Use a scanning device to perform high-resolution scanning on the core column image and obtain the acquired image. The scanning device includes a core scanner and a high-resolution digital camera. The core column image is a two-dimensional unfolded image or a 360° panoramic image of the core column under uniform illumination conditions. The high-resolution scanning is such that the image resolution can clearly display millimeter-level textures. Step A2 involves performing image preprocessing and image meshing on the acquired images to obtain sample images. The specific steps are as follows: Step A21 involves preprocessing the acquired image to obtain a preprocessed image. This preprocessing includes color correction and homogenization, image enhancement, and geometric correction. Color correction and homogenization eliminate shadows and color differences caused by uneven lighting and ensure consistent brightness across the entire core surface. Image enhancement uses contrast stretching and Laplacian sharpening algorithms to highlight the contours of the bedding planes. Geometric correction corrects image distortion caused by the shooting angle and ensures accurate horizontal and vertical proportions. The specific steps of image enhancement are as follows. Step A211: The acquired image is contrast-stretched using a contrast stretching algorithm, as shown in formula (1). （1） in, Image stretched for contrast in coordinates The pixel grayscale value at that location, Image before contrast stretching in coordinates The pixel grayscale value at that location, and These are the row index and column index of the pixel, respectively. and These are the maximum and minimum gray values of the image before contrast stretching, respectively. and These are the maximum and minimum gray values within the target gray range, respectively. Step A212: The acquired image is sharpened using the Laplacian sharpening algorithm, as shown in formula (2). （2） in, The image after sharpening. The image before sharpening. This is the sharpening intensity factor. For the Laplace operator; Step A22: Perform image meshing on the preprocessed image to obtain a meshed image. Specifically, image meshing involves establishing a Cartesian coordinate system based on the preprocessed image, defining the origin as the first pixel at the bottom left of the image, and defining the horizontal direction... The axis increases from left to right, while the vertical direction... The axes increase from bottom to top.
3. The method for rapid identification of clastic rock sedimentary bedding types according to claim 2, characterized in that: Step B involves performing edge detection on the sample image to obtain sample texture lines, then performing skeletonization on the sample texture lines to obtain a set of sample texture curves. The specific steps are as follows. Step B1 involves performing edge detection on the sample image and obtaining sample texture lines. The edge detection includes Gaussian filtering for noise reduction, finding image gradients, non-maximum suppression, and dual-threshold boundary tracking. The specific steps are as follows. Step B11: Perform Gaussian filtering on the sample image to denoise and obtain the denoised sample image. The Gaussian filtering denoising involves assigning weights to points at different positions in the convolution kernel according to a Gaussian distribution, as shown in formula (3). （3） in, For coordinates Gaussian filter kernel weights at the specified location. The standard deviation of the Gaussian distribution; Step B12 involves calculating the image gradient of the denoised sample image and obtaining the image gradient calculation result. Specifically, this is done using... sobel The operator calculates the gradient magnitude and gradient direction of the denoised sample image, as shown in formula (4). ； ； ； （4） in, for x Orientation gradient, for y Orientation gradient, For gradient magnitude, For the gradient direction, The denoised sample image to be processed; Step B13: Based on the image gradient calculation results, perform non-extreme suppression on the edges of the denoised sample image and suppress the sample image. Specifically, quantize the gradient direction to the 0° direction and compare the current pixel with two adjacent pixels in the gradient direction. If the gradient magnitude of the current pixel is not the maximum value, then suppress it; if the gradient magnitude of the current pixel is the maximum value, then do not suppress it. Step B14 involves performing dual-threshold boundary tracking on the suppressed sample image to obtain sample texture lines. The specific steps are as follows: Step B141: Set the upper threshold and the lower threshold. If the number of pixels in the suppressed sample image is greater than the upper threshold, it is called a strong boundary. If the number of pixels in the suppressed sample image is less than the lower threshold, it is not a boundary. Step B142: Take the pixel between the upper threshold and the lower threshold as the false boundary, and then determine whether there are any points in the eight neighborhoods of the pixels in the false boundary that are greater than the upper threshold. If they exist, they are the boundary; if they do not exist, they are discarded. Step B2 involves skeletalizing the sample texture lines and obtaining a set of sample texture curves. Specifically, the sample texture lines are processed into continuous curves with a width of one pixel, and then the continuous curves are converted into vector data to obtain sample texture curves, thereby obtaining a set of sample texture curves.
4. The method for rapid identification of clastic rock sedimentary bedding types according to claim 3, characterized in that: Step C involves quantitatively calculating the key morphological parameters of each sample curve in the sample striation curve set and obtaining the calculation results for the morphological parameters of the sample striation curve. These key morphological parameters include striation dip angle, striation curvature, layer interface morphology, and texture features. The specific steps are as follows. Step C1: Calculate the lamellar dip angle of the sample lamellar curve, as shown in formula (5). （5） in, The dip angle of the laminae. The height of a pixel. This represents the horizontal distance of a pixel in the coordinate system. Step C2: Calculate the laminar curvature of the sample laminar curve, as shown in formula (6). （6） in, For the laminar curvature, This is a laminar curve function; Step C3: Calculate the layer interface morphology of the sample striation curve. Specifically, the layer interface morphology is determined by mathematically fitting the vector curves of the top and bottom interfaces of the layer system and judging whether the morphology is flat, grooved, or wavy, as shown in formula (7). （7） in, This is a curve function representing the morphology of the layered interface. , , , , and All are fitting coefficients; Step C4: Calculate the texture features of the sample lamina curves, wherein the texture features include mean features, variance features, contrast features, homogeneity features, and entropy features. The specific steps are as follows. Step C41: Calculate the mean characteristic of the sample striation curve, as shown in formula (8). （8） in, The mean value feature of the gray-level co-occurrence matrix of the sample lamina curve. and These represent the number of rows and columns of the gray-level co-occurrence matrix, respectively. In direction distance Position and pixel position The grayscale value of the other pixel that appears in a pair. For distance, It is the direction angle; Step C42: Calculate the variance characteristic of the sample striation curve, as shown in formula (9). （9） in, Let V be the variance characteristic of the gray-level co-occurrence matrix of the sample striation curve. The average position of the sample striation curve; Step C43: Calculate the contrast feature of the sample lamellar curve, as shown in formula (10). （10） in, The gray-level co-occurrence matrix contrast feature of the sample lamina curve; Step C44: Calculate the homogeneity characteristic of the sample striation curve, as shown in formula (11). （11） in, The gray-level co-occurrence matrix homogeneity feature of the sample lamina curve; Step C45: Calculate the entropy feature of the sample lamina curve, as shown in formula (12). （12） in, The gray-level co-occurrence matrix entropy feature of the sample lamina curve. The sample ridge curve at the pixel location The grayscale value at that location.
5. The method for rapid identification of clastic rock sedimentary bedding types according to claim 4, characterized in that: Step D: Construct multidimensional feature vectors for all samples by calculating the morphological parameters of the laminar curves, as shown in formula (13). （13） in, This is the multidimensional feature vector of the sample.
6. The method for rapid identification of clastic rock sedimentary bedding types according to claim 5, characterized in that: Step E involves calculating the similarity between each sample's multidimensional feature vector and the standard multidimensional feature vector, and obtaining the similarity calculation results. The similarity calculation includes Euclidean distance similarity, cosine similarity, and... Tanimoto The specific steps for calculating coefficient similarity and dynamic time warping similarity are as follows: Step E1 involves calculating the Euclidean distance similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (14). （14） in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Euclidean distance similarity between them For the multidimensional feature vector of the sample, For sample multidimensional feature vectors The i 3D eigenvalues Standard multidimensional feature vector The j 3D eigenvalues; Step E2 involves calculating the cosine similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. The cosine similarity is used to compare the directional differences between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (15). （15） in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Cosine similarity between them; Step E3, compare the sample multidimensional feature vector with the standard multidimensional feature vector. Tanimoto The coefficient similarity is calculated, wherein... Tanimoto The coefficient similarity is used to measure the degree of similarity between the sample multidimensional feature vector and the standard multidimensional feature vector, as shown in formula (16). （16） in, For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Between Tanimoto Coefficient similarity; Step E4: Calculate the dynamic time warping similarity between the sample multidimensional feature vector and the standard multidimensional feature vector. This dynamic time warping similarity is used to compare laminar thickness sequences or curves that are similar in shape but differ in length due to different deposition rates, as shown in formula (17). ； ； （17） in, The distance matrix is... This is the cumulative distance matrix. For sample multidimensional feature vectors Compared with standard multidimensional feature vectors Dynamic time-warped similarity between them For the first alignment path One alignment point pair, The total number of alignment point pairs, The index of the alignment point pair.
7. The method for rapid identification of clastic rock sedimentary bedding types according to claim 6, characterized in that: Step F involves sorting the similarity calculation results of the stratigraphic curves for each sample according to the sedimentary bedding type and obtaining an initial similarity ranking list. Then, each initial similarity ranking list is merged and sorted to obtain a final similarity ranking list. The specific steps are as follows. Step F1: Sort the similarity calculation results of the laminar curves of each sample according to the sedimentary bedding type and obtain an initial similarity ranking list, wherein the initial similarity ranking list includes a Euclidean distance similarity ranking sublist, a cosine similarity ranking sublist, and... Tanimoto The coefficient similarity sorting sublist and the dynamic time warping similarity sorting sublist, wherein the sedimentary bedding types include horizontal bedding, parallel bedding, oblique bedding, trough cross-bedding, platy cross-bedding, wavy bedding and massive bedding; Step F2 involves merging and sorting each initial similarity ranking list to obtain a final similarity ranking list. Specifically, the merging and sorting process employs... Borda The specific steps for merging and sorting using the counting method are as follows. Step F21: For each sedimentary bedding type, sort the sublists by Euclidean distance similarity and by cosine similarity. Tanimoto The positions in the coefficient similarity ranking sublist and the dynamic time warping similarity ranking sublist are assigned scores respectively. If the sedimentary bedding type ranks first in the sublist, it receives N points; if it ranks second, it receives N-1 points, and so on. If the sedimentary bedding type ranks N in the sublist, it receives 1 point, as shown in formula (18). ； ； ； （18） in, The score for sedimentary bedding type in the Euclidean distance similarity sublist. The number of sedimentary bedding types, This represents the ranking of sedimentary bedding types within the Euclidean distance similarity sublist. The score for sedimentary bedding type in the cosine similarity sorted sublist. This represents the ranking of sedimentary bedding types within the cosine similarity sorting sublist. For sedimentary bedding types in Tanimoto Scores in the coefficient similarity sorted sublist For sedimentary bedding types in Tanimoto Ranked in the coefficient similarity sublist The score for sedimentary bedding type in the dynamic time-regularity similarity sublist. The ranking of sedimentary bedding types in the dynamic time-regularity similarity sublist; Step F22: Sum the scores of each sedimentary bedding type in each sublist and obtain the fusion ranking score. Then, sort all the fusion ranking scores in descending order to obtain the final similarity ranking list. The process of obtaining the fusion ranking score is shown in formula (19). （19） in, The fusion and sorting of sedimentary bedding types, It is a sedimentary bedding type.
8. The method for rapid identification of clastic rock sedimentary bedding types according to claim 7, characterized in that: Step G: Obtain the sedimentary bedding type identification result of each sample's stratigraphic curve based on the final similarity ranking list. Specifically, the sedimentary bedding type ranked first in the final similarity ranking list is taken as the sedimentary bedding type identification result of the sample's stratigraphic curve.
9. A rapid identification system for clastic sedimentary bedding types, wherein the specific identification process of the rapid identification system for clastic sedimentary bedding types is based on the rapid identification method for clastic sedimentary bedding types according to any one of claims 1-8: comprising an image acquisition module, an image processing module, a parameter calculation module, a feature vector construction module, a similarity calculation module, a similarity ranking module, and a bedding type output module, wherein the image acquisition module is used to perform high-resolution scanning of core column images using a scanning device and obtain acquired images, and then perform image preprocessing and image meshing on the acquired images in sequence to obtain sample images; The image processing module is used to perform edge detection on the sample image and obtain sample texture lines, and then perform skeletonization processing on the sample texture lines to obtain a set of sample texture curves. The parameter calculation module is used to quantitatively calculate the key morphological parameters of each sample curve in the sample striation curve set and obtain the calculation results of the morphological parameters of the sample striation curve respectively. The feature vector construction module constructs multidimensional feature vectors for all samples based on the calculation results of the morphological parameters of the laminar curves of all samples. The similarity calculation module is used to calculate the similarity between the multidimensional feature vector of each sample and the standard multidimensional feature vector and obtain the similarity calculation results respectively. The similarity ranking module is used to sort the similarity calculation results of the laminar curves of each sample according to the sedimentary bedding type and obtain an initial similarity ranking list, and then merge and sort each initial similarity ranking list to obtain a final similarity ranking list. The bedding type output module is used to obtain the deposition bedding type identification results of each sample's bedding curve based on the final similarity sorting list.