Method for identifying kerogen microcomponents

Abstract

The application discloses a method for identifying kerogen microcomponents, and belongs to the technical field of oil and gas exploration. The method comprises the following steps: extracting a plurality of particle graphs from a to-be-detected kerogen image, wherein each particle graph comprises an image of a single particle; performing feature extraction on each particle graph to determine a plurality of characteristic values corresponding to each particle graph; determining the microcomponent corresponding to each particle graph according to the plurality of characteristic values corresponding to each particle graph and a classification model; determining the number of types of microcomponents contained in the to-be-detected kerogen image according to the microcomponent corresponding to each particle graph; and determining the area proportion of each microcomponent in the to-be-detected kerogen image according to the number of types of microcomponents contained in the to-be-detected kerogen image and the microcomponent corresponding to each particle graph. The method is based on a computer to identify the microcomponent of each particle, and then calculates the specific value of the area proportion of each microcomponent, and the calculation accuracy is high.

Description

Technical Field

[0001] This application relates to the field of oil and gas exploration technology, and in particular to a method for identifying microscopic components of kerogen. Background Technology

[0002] Kerogen refers to a general term for various organic compounds and is an important primary material for the formation of petroleum and natural gas. Kerogen separated from rocks is generally a very fine powder, its shape, structure, and composition invisible to the naked eye. However, under a microscope, it appears as fragments with specific morphological and structural characteristics, revealing its organic components. Microscopic components refer to these identifiable organic components. Because different kerogen microscopic components exhibit significantly different hydrocarbon-generating capabilities, accurately identifying these components is crucial for accurately predicting the reserves of oil and gas reservoirs.

[0003] In related technologies, the identification of microscopic components in kerogen typically involves operators observing the kerogen sample field by field under a microscope and estimating the area percentage of each component. However, this method relies on the operator's subjective judgment, and the resulting component proportions are estimates with low accuracy. Summary of the Invention

[0004] In view of this, this application provides a method for identifying the microscopic components of kerogen, which does not require manual judgment and has high accuracy in calculating the proportion of each component of kerogen.

[0005] Specifically, the following technical solutions are included:

[0006] This application provides a method for identifying microscopic components of kerogen, the method comprising:

[0007] Multiple particle images were extracted from the kerogen image to be tested, and each particle image included an image of a single particle.

[0008] Feature extraction is performed on each particle map to determine multiple feature values ​​corresponding to each particle map;

[0009] Based on the multiple feature values ​​and classification model corresponding to each particle image, the microscopic components corresponding to each particle image are determined;

[0010] Based on the microscopic components corresponding to each particle image and the area of ​​each particle image, the area ratio of each microscopic component in the kerogen image to be tested is determined.

[0011] In some embodiments, extracting multiple particle images from the kerogen image to be tested includes:

[0012] Edge extraction is performed on each particle contained in the kerogen image to be tested to determine the contour of each particle, wherein the contour includes edge contour and cavity contour.

[0013] The particles are screened, and the outline of each screened particle is cut out to obtain the multiple particle images.

[0014] In some embodiments, the process of screening the particles and performing image cutout processing on the outline of each screened particle to obtain the plurality of particle images includes:

[0015] The area of ​​each particle is determined based on its outline;

[0016] Based on the area of ​​each particle, determine the number of particles to be processed;

[0017] A binary mask is applied to the multiple particles to be processed to obtain the multiple particle images.

[0018] In some embodiments, determining the plurality of particles to be processed based on the area of ​​each particle includes:

[0019] In response to a particle area being greater than or equal to a threshold, the plurality of particles to be processed are obtained.

[0020] In some embodiments, determining the plurality of particles to be processed based on the area of ​​each particle includes:

[0021] Arrange the particles in descending order of area to obtain an area sequence;

[0022] Select the particles corresponding to the first N areas in the area sequence to obtain the multiple particles to be processed.

[0023] In some embodiments, the step of edge extraction of each particle contained in the kerogen image to be tested, and determining the contour of each particle, includes:

[0024] Gradient-based edge detection is performed on the kerogen image to be tested to obtain a transition image;

[0025] Boundary tracking is performed on the transition image to determine the contour of each particle.

[0026] In some embodiments, prior to extracting multiple particle images from the kerogen image to be tested, the method further includes:

[0027] The kerogen image to be tested is acquired, and the image is then subjected to image enhancement, noise reduction, image size adjustment, brightness adjustment, contrast enhancement, smoothing and noise reduction, and grayscale processing in sequence.

[0028] In some embodiments, the plurality of feature values ​​include maximum gray value, minimum gray value, mean gray value, standard deviation of gray value, mean energy, standard deviation of energy, mean entropy, standard deviation of entropy, and information entropy.

[0029] In some embodiments, determining the microscopic components corresponding to each particle image based on multiple feature values ​​and a classification model corresponding to each particle image includes:

[0030] A classification model is obtained, which is trained based on multiple particle maps of a known microscopic component, wherein multiple feature values ​​corresponding to each particle map of the known microscopic component and the microscopic component correspond to the input and output of the classification model.

[0031] By inputting multiple feature values ​​corresponding to each particle image into the classification model, the microscopic components corresponding to each particle image are obtained.

[0032] In some embodiments, the area percentage of each microscopic component in the kerogen image to be tested is obtained according to the following formula:

[0033]

[0034] In the formula: c k The microscopic components identified after classification for each particle k, s is the target microscopic component, and r s Let a be the area percentage of the target micro-component s in the image of the kerogen to be tested. k Let A be the particle area of ​​particle k, and A be the area of ​​the kerogen image to be tested.

[0035] The method for identifying microscopic components of kerogen provided in this application involves image processing of the kerogen image to obtain multiple particle images, calculating multiple feature values ​​corresponding to each particle image, and inputting these feature values ​​into a classification model to obtain the microscopic components corresponding to each particle image, thereby determining the area proportion in the kerogen image. Since this method is based on extracting feature values ​​from each particle and identifying each microscopic component through a classification model, its implementation is entirely computer-driven. Compared to traditional manual calculations, this method allows for computer-based identification of microscopic components in each particle, thereby calculating the specific area proportion of each microscopic component without requiring manual judgment, resulting in higher calculation accuracy. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart illustrating a method for identifying microscopic components of kerogen, provided as an exemplary embodiment of this application;

[0038] Figure 2 A flowchart illustrating another method for identifying microscopic components of kerogen provided as an exemplary embodiment of this application;

[0039] Figure 3 A flowchart illustrating the method for processing the image of the kerogen to be tested in another method for identifying microscopic components of kerogen provided in an exemplary embodiment of this application;

[0040] Figure 4 A flowchart illustrating a method for extracting multiple particle maps from an image of kerogen in an exemplary embodiment of this application for identifying microscopic components of kerogen.

[0041] Figure 5 A flowchart illustrating a method for determining the micro-components corresponding to each particle in another method for identifying kerogen micro-components provided as an exemplary embodiment of this application. Detailed Implementation

[0042] To make the technical solutions and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0043] Kerogen refers to the insoluble organic matter in source rocks and is an important primary material for oil and gas formation. Kerogen is a general term for various organic substances, and current industry standards classify it into twelve microscopic components, belonging to four microgroups. The twelve microscopic components are: phytoplankton, saprophytic amorphous bodies, resinous bodies, corky bodies, cutinous bodies, pollen bodies, fungal spore bodies, humic amorphous bodies, basidiospore amorphous bodies, hydrogen-rich vitrinite, normal vitrinite, and filamentous bodies. The four microgroups are: saprophytic group, chitinous group, vitrinite group, and inertinite group.

[0044] Currently, the methods for classifying kerogen microstructures are mainly based on the methods recommended in standard SY / T 5125-2014.

[0045] The basic steps of this method are:

[0046] ① Prepare microfilms of kerogen samples using either the polyvinyl alcohol method or the glycerol method.

[0047] ② Divide the thin slice into several fields of view and observe each field of view under transmitted light and reflected fluorescence. The observer determines the type of organic component appearing in the field of view and estimates the area ratio of the component in the entire field of view.

[0048] ③ Based on the analysis results of all fields of view, count the kerogen components that appear in the sample and estimate the percentage of each component in the entire slice area.

[0049] ④ Apply different weighting coefficients to different components according to the formula given in the standard to calculate the kerogen type index (TI value). The TI value can be further used for source rock evaluation and oil and gas reservoir reserve assessment.

[0050] This method has at least the following problems:

[0051] ① The determination of the composition of the target particles depends on the operator's subjective understanding, which leads to a lack of consistency in the judgment results of different operators.

[0052] ② To improve the accuracy of particle area estimation, a denser grid needs to be used to divide the field of view, which will increase the number of fields of view and increase the computational load.

[0053] ③ Because each field of view is analyzed separately, when the target particle covers multiple fields of view, it may be identified as different components in different fields of view. In this case, secondary component identification and correction of the results are required.

[0054] To address the problems existing in related technologies, this application provides a method for identifying microscopic components of kerogen, the flowchart of which is shown below. Figure 1 As shown. The method includes:

[0055] Step 101: Extract multiple particle images from the kerogen image to be tested. Each particle image includes an image of a single particle.

[0056] Step 102: Extract features from each particle image and determine multiple feature values ​​corresponding to each particle image.

[0057] Step 103: Based on the multiple feature values ​​and classification model corresponding to each particle image, determine the microscopic components corresponding to each particle image.

[0058] Step 104: Based on the microscopic components corresponding to each particle image and the area of ​​each particle image, determine the area ratio of each microscopic component in the kerogen image to be tested.

[0059] In some embodiments, extracting multiple particle images from the kerogen image to be tested includes:

[0060] Edge extraction is performed on each particle contained in the kerogen image to be tested to determine the contour of each particle, where the contour includes edge contour and cavity contour.

[0061] The particles are screened, and the outline of each particle is cut out to obtain multiple particle images.

[0062] In some embodiments, the particles are screened, and the outline of each screened particle is cut out to obtain multiple particle images, including:

[0063] The area of ​​each particle is determined based on its outline.

[0064] Based on the area of ​​each particle, determine the number of particles to be processed.

[0065] Binary masking is applied to the multiple particles to be processed, resulting in multiple particle images.

[0066] In some embodiments, determining the plurality of particles to be processed based on the area of ​​each particle includes:

[0067] In response to a particle area being greater than or equal to a threshold, multiple particles are obtained to be processed.

[0068] In some embodiments, determining the plurality of particles to be processed based on the area of ​​each particle includes:

[0069] Arrange the particles in descending order of area to obtain an area sequence.

[0070] Select the particles corresponding to the first N areas in the area sequence to obtain multiple particles to be processed.

[0071] In some embodiments, edge extraction is performed on each particle contained in the kerogen image to be tested, and the contour of each particle is determined by:

[0072] Gradient-based edge detection is performed on the kerogen image to be tested to obtain a transition image.

[0073] Boundary tracking is performed on the transition image to determine the contour of each particle.

[0074] In some embodiments, before extracting multiple particle maps from the kerogen image to be tested, the method further includes:

[0075] The kerogen image to be tested is acquired, and then image enhancement, noise reduction, image size adjustment, brightness adjustment, contrast enhancement, smoothing and noise reduction, and grayscale processing are performed on the kerogen image in sequence.

[0076] In some embodiments, the multiple feature values ​​include maximum gray value, minimum gray value, mean gray value, standard deviation of gray value, mean energy, standard deviation of energy, mean entropy, standard deviation of entropy, and information entropy.

[0077] In some embodiments, the microscopic components corresponding to each particle image are determined based on multiple feature values ​​and a classification model corresponding to each particle image, including:

[0078] A classification model is obtained by training multiple particle maps of a known microscopic component. The multiple feature values ​​corresponding to each particle map of the known microscopic component and the corresponding microscopic component are used as the input and output of the classification model.

[0079] By inputting multiple feature values ​​corresponding to each particle image into the classification model, the microscopic components corresponding to each particle image are obtained.

[0080] In some embodiments, the area percentage of each micro-component in the kerogen image to be tested is obtained according to the following formula:

[0081]

[0082] In the formula: c k The microscopic components identified after classification for each particle k, s is the target microscopic component, and r s Let a be the area percentage of the target micro-component s in the image of the kerogen to be tested. k Let A be the particle area of ​​particle k, and A be the area of ​​the kerogen image to be tested.

[0083] Therefore, the method for identifying microscopic components of kerogen provided in this application involves image processing of the kerogen image to obtain multiple particle images, calculating multiple feature values ​​corresponding to each particle image, and inputting these feature values ​​into a classification model to obtain the microscopic components corresponding to each particle image, thereby determining their area proportion in the kerogen image. Since this method is based on extracting feature values ​​from each particle and identifying each microscopic component through a classification model, its implementation is entirely computer-based. Compared to traditional manual calculations, this method allows for computer-based identification of microscopic components in each particle, thereby calculating the specific area proportion of each microscopic component without requiring manual judgment, resulting in higher calculation accuracy.

[0084] This application also provides a method for identifying microscopic components of kerogen, the flowchart of which is shown below. Figure 2 As shown, the method includes:

[0085] Step 201: Obtain the kerogen image to be tested, and perform image enhancement, noise reduction, image size adjustment, brightness adjustment, contrast enhancement, smoothing and noise reduction, and grayscale processing on the kerogen image to be tested in sequence.

[0086] By sequentially performing image enhancement, noise reduction, image size adjustment, brightness adjustment, contrast enhancement, smoothing and noise reduction, and grayscale processing on the kerogen image to be tested, the image preprocessing of the kerogen image to be tested is achieved, and the particle outline of the kerogen image after preprocessing is clearer.

[0087] Specifically, the flowchart of the steps is as follows: Figure 3 As shown, this step may further include:

[0088] Step 2011: Obtain the image of the kerogen to be tested.

[0089] In some embodiments, the image of the kerogen to be tested can be obtained by preparing a suitable microfilm from the kerogen sample using the polyvinyl alcohol method or the glycerol method, and then photographing the microfilm under transmitted light through a microscope.

[0090] In some embodiments, the kerogen image to be tested is a color image having three color channels.

[0091] In some embodiments, by dividing a microscopic slide into multiple fields of view, and taking a picture of each field of view under transmitted light from a microscope, multiple images of the kerogen to be tested can be obtained. By processing each image of the kerogen to be tested, the number of particles in each field of view is reduced, making the imaging of each particle in each field of view clearer, which facilitates the subsequent extraction of particle contours in the image of the kerogen to be tested.

[0092] Understandably, stitching together multiple images of the kerogen to be tested can yield a complete image of the kerogen to be tested.

[0093] Step 2012: Image enhancement is performed on the kerogen image to be tested.

[0094] When a microscope photographs a thin section of microstructure, the edges of the kerogen image may appear blurry due to insufficient illumination from the transmitted light. Therefore, image enhancement processing can resolve this issue, making the edges of the kerogen image clearer.

[0095] Optionally, for each chee root image to be tested, it is first split into an illumination component I and a reflection component R, wherein the illumination component I is mainly low-frequency and the reflection component R is mainly high-frequency; then, the illumination component I and the reflection component R are subjected to Fourier transform and high-pass filtering respectively to obtain the processed illumination component and reflection component; based on the processed illumination component and reflection component, the processed chee root image to be tested is obtained, which is the chee root image to be tested after image enhancement adjustment.

[0096] Optionally, the image enhancement processing method also includes adaptive Gamma grayscale correction, which involves performing Gamma correction on the kerogen image to be tested to obtain an enhanced kerogen image.

[0097] Step 2013: Eliminate salt-and-pepper noise in the kerogen image to be tested.

[0098] By performing salt-and-pepper noise reduction on the enhanced kerogen image, randomly appearing black and white pixels can be removed, improving the clarity of the kerogen image without damaging the contour information.

[0099] Optionally, each test cheese root image can be filtered separately to obtain multiple filtered test cheese root images. These filtered test cheese root images are the test cheese root images after salt-and-pepper noise removal.

[0100] Step 2014: Adjust the size of the kerogen image to be tested.

[0101] The size of the salt-and-pepper noise-reduced kerogen image to be tested is adjusted to facilitate subsequent image processing. This resizing allows the salt-and-pepper noise-reduced kerogen sub-image to be adjusted to a preset size.

[0102] Optionally, the kerogen image to be tested is a rectangular image, with dimensions including the length and width of the rectangle. For example, the preset dimensions could be a length of 3264 pixels and a width of 2448 pixels.

[0103] Step 2015: Adjust the brightness of the kerogen image to be tested.

[0104] The brightness of the kerogen image to be tested, which has already been resized, is adjusted to ensure that the grayscale value of the background in the kerogen image is lower than the grayscale value of the particles. The background refers to the portion of the kerogen image other than the particles.

[0105] Optionally, K-means clustering is performed on each kerogen image to be tested to obtain the gray values ​​of the background and particles in each color channel of the kerogen image to be tested. Then, based on the gray values ​​of the background and particles in each color channel, the average gray value of the background and the average gray value of the particles in the three color channels are calculated. In response to the fact that the average gray value of the background is greater than the average gray value of the particles, the gray values ​​of the three color channels corresponding to each pixel are converted into the difference between 256 and the gray value of the current color channel.

[0106] Step 2016: Enhance the contrast of the kerogen image to be tested.

[0107] By enhancing the contrast of the kerogen image under test, which has already undergone brightness adjustment, the difference in gray levels between the background and the grain parts in the kerogen image can be made greater, which facilitates the subsequent extraction of grain contours.

[0108] Specifically, contrast enhancement is performed on the three color channels of each kerogen image to be tested. Contrast enhancement methods can include Gamma transformation, histogram equalization, and nonlinear brightness stretching, among others.

[0109] Specifically, the standard deviation of the grayscale values ​​in the image of the kerogen sample to be tested is calculated using the following formula:

[0110]

[0111] In the formula, i and j are the row and column of each pixel in the image of the cheesin to be tested, respectively, and f ij Let denot be a pixel in the cheese root image to be tested, M be the total number of pixels in the cheese root image to be tested, μ be the mean gray value of the cheese root image to be tested, and σ be the standard deviation of the gray value of the cheese root image to be tested.

[0112] By enhancing the contrast of each test chellogen image, the grayscale standard deviation σ of each test chellogen image is made not less than 110, thereby making the difference between the grayscale values ​​of the particle part and the grayscale values ​​of the background part greater, making the image clearer.

[0113] Step 2017: Smooth and denoise the kerogen image to be tested.

[0114] Since enhancing contrast can degrade pixel quality, smoothing and denoising the contrast-enhanced kerogen image can eliminate image blurring caused by noise and improve the image clarity.

[0115] Optionally, two-dimensional Gaussian filtering, rectangular window filtering, bilateral filtering, median filtering, and other methods can be used to smooth and reduce noise in the kerogen image under test.

[0116] Optionally, smoothing and denoising can be performed on the three color channels of each test kerogen image to obtain the images corresponding to the three channels respectively. Then, the images corresponding to the three channels can be superimposed to obtain the denoised test kerogen image.

[0117] Step 2018: Perform grayscale processing on the kerogen image to be tested.

[0118] By performing grayscale processing on each smoothed and denoised kerogen image, that is, converting the three color channels of the image into one color channel, it is easier to perform edge detection on the particles in the subsequent process.

[0119] Optionally, the grayscale value of the pixel can be set to the average grayscale value of the three color channels included in the pixel to achieve grayscale processing of the kerogen image to be tested.

[0120] Step 202: Extract multiple particle images from the kerogen image to be tested.

[0121] Each particle image includes an image of a single particle.

[0122] For this step, see Figure 4 Specifically, it can include:

[0123] Step 2021: Edge extraction is performed on each particle contained in the kerogen image to be tested to determine the contour of each particle.

[0124] The contours include edge contours and void contours. Edge contours are the contours formed by the outer edges of particles; void contours are the contours formed by the outer contours of the hollow parts in particles with hollow interiors.

[0125] Edge extraction for each particle in the kerogen image under test can be obtained through a transition image. Specifically, gradient-based edge detection is performed on the kerogen image under test to obtain a transition image; and boundary tracking is performed on the transition image to determine the contour of each particle.

[0126] Alternatively, edge detection methods for the kerogen image to be tested can include Sobel kernel, Roberts kernel, Preweitt kernel, Laplace kernel, Canny method, morphological methods, pre-trained neural networks, etc.

[0127] Step 2022: Determine the area of ​​each particle based on its outline.

[0128] By analyzing the outline of each particle, the total number of pixels within that outline can be calculated, thereby determining the area of ​​each particle, which facilitates subsequent particle cutout processing.

[0129] Specifically, in response to a particle having a hollow profile, the area of ​​the particle is the area enclosed by the outer profile minus the area enclosed by the hollow profile; in response to a particle not having a hollow profile, the area of ​​the particle is the area enclosed by the outer profile.

[0130] Step 2023: Determine the multiple particles to be processed based on the area of ​​each particle.

[0131] The purpose of this step is to remove multiple particles with excessively small areas from the kerogen image to be tested, thereby reducing the amount of computation.

[0132] This step can include the following two possible implementation methods:

[0133] In one possible implementation, multiple particles are obtained in response to the particle area being greater than or equal to a threshold.

[0134] By comparing the area of ​​each particle with a threshold, particles whose area is greater than or equal to the threshold are identified as particles to be processed.

[0135] Optionally, the threshold can be 15% of the area of ​​the kerogen image to be tested.

[0136] In another possible implementation, the areas of the particles are arranged in descending order to obtain an area sequence; the particles corresponding to the first N areas in the area sequence are selected to obtain multiple particles to be processed.

[0137] By comparing the area sequence of each particle with N, particles whose area sequence is less than or equal to N are identified as particles to be processed.

[0138] Optionally, N can be 5.

[0139] Step 2024: Perform binary masking on the multiple particles to be processed to obtain the mask value of each pixel in the kerogen image to be tested.

[0140] The purpose of this step is to obtain multiple particles with mask values ​​based on the outline of each particle to be processed, which facilitates subsequent image matting processing.

[0141] Specifically, the mask values ​​of the pixels containing the background and the hollow parts of the particles to be processed in the image of the kerogen to be tested are set to 0, and the mask value of the pixels containing the particles to be processed is set to 1.

[0142] Step 2025: Extract multiple particle images based on the mask value of each pixel in the image of the cheesin to be tested.

[0143] Based on the binary mask of each particle to be processed, multiple particle images can be obtained by performing image matting on multiple particles to be processed, which can be used for subsequent feature extraction.

[0144] Specifically, by performing image matting operations on the input image of the cheesin to be tested, and a completely black image of the same size as the image of the cheesin to be tested, multiple particle images can be obtained.

[0145] The formula for image matting can be:

[0146] T k =α k F+(1-α k B

[0147] In the formula, T k For a particle map containing only particle k, α kF is a binary mask for particle k, F is the image of the cheese root to be tested, and B is a completely black image of the same size as the image of the cheese root to be tested.

[0148] Step 203: Extract features from each particle image and determine multiple feature values ​​corresponding to each particle image.

[0149] By extracting features from each particle image, we can obtain the feature value of that particle. The feature value can characterize the particle's properties and prepare data for subsequent identification of particle components.

[0150] In some embodiments, there are multiple ways to extract features from each particle image. The following are some examples of these methods:

[0151] In one possible implementation, each particle image has three color channels. Feature extraction is performed on each color channel of each particle image to determine multiple feature values ​​corresponding to each particle image. These multiple feature values ​​include the maximum gray value, minimum gray value, mean gray value, standard deviation of gray value, mean energy, standard deviation of energy, mean entropy, standard deviation of entropy, and information entropy.

[0152] The formula for feature extraction can be:

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161]

[0162] In the formula, i and j are the row and column of each pixel in the image, respectively, and f ij Let be a specific pixel in the image, M be the total number of pixels in the image, h(t) be the total number of pixels with gray level t, p(t) be the frequency of gray level t, and f be the frequency of the pixel with gray level t. max f is the maximum grayscale value of the image. min Let μ be the minimum gray level of the image, μ be the mean gray level of the image, and σ be the standard deviation of the gray level of the image. ELet σ be the mean energy of the image. E Let μ be the standard deviation of the image energy. s Let σ be the mean entropy of the image. s Let be the standard deviation of the image entropy, and H be the information entropy of the image.

[0163] In one possible implementation, each particle image has three color channels, and the normalization matrix for each color channel is obtained. The formula for calculating the normalization matrix can be:

[0164]

[0165] In the formula, i and j are the gray levels of two pixels, respectively, and p ij P represents the number of arrangements of all i,j pixels in the particle image that satisfy (δ,θ), where δ is the distance between two pixels, θ is the direction between two pixels, and P is the distance between p pixels. ij The matrix formed Let be the normalized matrix of matrix P.

[0166] Based on the normalization matrix of each color channel, feature extraction is performed on each particle image to determine multiple feature values ​​corresponding to each particle image. These feature values ​​include energy, contrast, dissimilarity, inverse moment, correlation coefficient, and entropy.

[0167] The formula for feature extraction can be:

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174] In the formula, For matrix A certain value in, μ X and μ Y Each is a matrix The mean of rows and columns in the array, σ X and σ Y For matrix The standard deviations of the rows and columns in the image are: W is the energy of the image, η is the contrast of the image, λ is the dissimilarity of the image, ε is the inverse moment of the image, γ is the correlation coefficient of the image, and h is the entropy of the image.

[0175] In one possible implementation, the granular image is divided into several units, and these units are grouped into several unit blocks. By calculating the gradient vector of each pixel in the granular image, a gradient histogram is generated for all pixels within each unit according to the gradient vector direction. Based on the gradient histograms generated for all pixels within each unit according to the gradient vector direction, all gradient histograms within each unit block are obtained. Based on all gradient histograms within each unit block, multiple unit block vectors X are generated. Each unit block vector X is normalized to obtain feature vectors for multiple unit blocks. Based on the feature vectors of multiple unit blocks, multiple feature values ​​corresponding to the granular image are obtained.

[0176] Those skilled in the art will understand that when using only a single method to calculate feature values ​​for component identification, the accuracy of the identification result is low; when using two or more methods to calculate feature values ​​for component identification, the accuracy of the identification result is high. In other words, the more feature value calculation methods used, the higher the accuracy of the identification result. In application, selective settings can be made according to the actual situation, which will not be elaborated here.

[0177] Step 204: Based on the multiple feature values ​​and classification model corresponding to each particle image, determine the microscopic components corresponding to each particle image.

[0178] The purpose of this step is to identify the components of the particles in each particle image to obtain the corresponding microscopic components for each particle image, which are then used to calculate the area ratio of different microscopic components.

[0179] For this step, see Figure 5 Specifically, it can include:

[0180] Step 2041: Obtain the classification model.

[0181] By acquiring a classification model, multiple particle map microstructures can be identified based on that model.

[0182] In some embodiments, the classification model is trained based on multiple particle maps of a known microscopic component, wherein multiple feature values ​​corresponding to each particle map of the known microscopic component and the microscopic component correspondence serve as the input and output of the classification model.

[0183] Optionally, the classification model can be a multi-class support vector machine or a neural network model.

[0184] Step 2042: Determine the microscopic components corresponding to each particle map based on the classification model.

[0185] By identifying the microscopic components corresponding to each particle map, the area percentage of different microscopic components can be calculated.

[0186] In some embodiments, multiple feature values ​​corresponding to each particle map are input into the classification model to obtain the microscopic components corresponding to each particle map.

[0187] Step 205: Based on the microscopic components corresponding to each particle image and the area of ​​each particle image, determine the area ratio of each microscopic component in the kerogen image to be tested.

[0188] This step allows us to obtain the area percentage of each microscopic component.

[0189] In some embodiments, the area percentage of each micro-component in the kerogen image to be tested can be obtained by the following formula:

[0190]

[0191] In the formula: c k The microscopic components identified after classification for each particle k, s is the target microscopic component, and r s Let a be the area percentage of the target micro-component s in the image of the kerogen to be tested. k Let A be the particle area of ​​particle k, and A be the area of ​​the kerogen image to be tested.

[0192] Therefore, the method for identifying microscopic components of kerogen provided in this application involves image processing of the kerogen image to be tested to obtain multiple particle images, calculating multiple feature values ​​corresponding to each particle image, and inputting these feature values ​​into a classification model to obtain the microscopic components corresponding to each particle image, thereby determining their area proportion in the kerogen image to be tested. Since this method is based on extracting feature values ​​from each particle and identifying each microscopic component through a classification model, its implementation is entirely computer-based. Compared to traditional manual calculations, this method allows for computer-based identification of microscopic components in each particle, thereby calculating the specific area proportion of each microscopic component without requiring manual judgment, resulting in higher calculation accuracy.

[0193] In this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The term "multiple" refers to two or more unless otherwise expressly defined.

[0194] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only.

[0195] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for identifying microscopic components of kerogen, characterized in that, The method includes: Acquire the kerogen image to be tested, and sequentially perform image enhancement, noise reduction, image size adjustment, brightness adjustment, contrast enhancement, smoothing and noise reduction, and grayscale processing on the kerogen image to be tested. Edge extraction is performed on each particle contained in the kerogen image to be tested to determine the contour of each particle, wherein the contour includes edge contour and cavity contour; the cavity contour is the contour formed by the outer contour of the hollow part in the particle. Based on the outline of each particle, the area of ​​each particle is determined; based on the area of ​​each particle, multiple particles to be processed are determined; binary masking is performed on the multiple particles to be processed to extract the multiple particle images; the kerogen image to be tested includes multiple kerogen sub-images to be tested, the mask value of the background and the hollow part of the particle to be processed in the kerogen sub-image to be tested is 0, and the mask value of the pixel where the particle to be processed is located is 1; each particle image includes an image of a single particle; For each particle image, feature extraction is performed to determine multiple feature values ​​corresponding to each particle image; the multiple feature values ​​include maximum gray value, minimum gray value, mean gray value, standard deviation of gray value, mean energy, standard deviation of energy, mean entropy, standard deviation of entropy, and information entropy; Based on the multiple feature values ​​and classification model corresponding to each particle image, the microscopic components corresponding to each particle image are determined; Based on the microscopic components corresponding to each particle image and the area of ​​each particle image, the area ratio of each microscopic component in the kerogen image to be tested is determined.

2. The method for identifying microscopic components of kerogen according to claim 1, characterized in that, The step of determining the multiple particles to be processed based on the area of ​​each particle includes: In response to a particle area being greater than or equal to a threshold, the plurality of particles to be processed are obtained.

3. The method for identifying microscopic components of kerogen according to claim 2, characterized in that, The step of determining the multiple particles to be processed based on the area of ​​each particle includes: Arrange the particles in descending order of area to obtain an area sequence; Select the particles corresponding to the first N areas in the area sequence to obtain the multiple particles to be processed.

4. The method for identifying microscopic components of kerogen according to claim 1, characterized in that, The step of edge extraction for each particle in the kerogen image to be tested, and determining the contour of each particle, includes: Gradient-based edge detection is performed on the kerogen image to be tested to obtain a transition image; Boundary tracking is performed on the transition image to determine the contour of each particle.

5. The method for identifying microscopic components of kerogen according to claim 1, characterized in that, The determination of the microscopic components corresponding to each particle image based on multiple feature values ​​and classification models includes: A classification model is obtained, which is trained based on multiple particle maps of a known microscopic component, wherein multiple feature values ​​corresponding to each particle map of the known microscopic component and the microscopic component correspond to the input and output of the classification model. By inputting multiple feature values ​​corresponding to each particle image into the classification model, the microscopic components corresponding to each particle image are obtained.

6. The method for identifying microscopic components of kerogen according to claim 1, characterized in that, The area percentage of each micro-component in the tested kerogen image is obtained using the following formula: In the formula: c k is the microcomponent identified after classification for each particle k, s is the target microcomponent, r s is the area proportion of the target microcomponent s in the kerogen image to be measured, a k is the particle area of the particle k, and A is the area of the kerogen image to be measured.

Images