A rock thin section analysis method for quantitative identification of shale sedimentary structures

Abstract

The present invention discloses a rock thin section analysis method for quantitatively distinguishing mud shale sedimentary structures, which generates a complete scanned and stitched image by scanning and splicing multiple rock thin section photos, obtains the corresponding RGB image, performs sharpening processing, then converts it into a grayscale image, performs Gaussian blur processing, and then performs binarization processing to obtain a binarized image, performs morphological reconstruction operation on the binarized image to obtain a reconstructed image, identifies the closed boundaries or contours of rock particles in the reconstructed image, constructs and analyzes the total number of rock mineral particles, the total area of ​​rock mineral particles, and the average area of ​​rock mineral particles. The sedimentary structure type is divided according to the three trend graphs. This method effectively avoids the inherent uncertainty of manual statistics and significantly improves the accurate discrimination of mud shale felsic sedimentary structure types.

Description

Technical Field

[0001] The invention relates to the field of analysis and identification of shale felsic genetic types, and in particular to a rock thin section analysis method for quantitatively identifying shale sedimentary structures. Background Art

[0002] The felsic in shale is usually input from external sources, and its depositional causes are complex and diverse, including gravity flow deposition, wind-driven lake deposition, and static water deposition, which makes the classification and identification of depositional types extremely challenging. Accurately distinguishing the depositional types of felsic helps to identify the depositional stage, which is of great significance for the reconstruction of the paleo-sedimentary environment of shale and reservoir evaluation.

[0003] Complex sedimentation manifests itself as a variety of sedimentary structures in rock thin sections. For example, different types of gravity flow sedimentation can form features such as normal grain order bedding, parallel bedding, reverse grain order bedding and massive bedding. However, the traditional identification of sedimentary structure types mainly relies on manual identification, which is based on large-scale observations of cores and the classification of sedimentary structure types through subjective analysis. This method is easily affected by human factors and has large errors and uncertainties. In addition, manual identification fails to make full use of the vertical development characteristics of mineral grains in rock thin sections to more intuitively and accurately reveal the type of sedimentation at a smaller scale.

[0004] Existing methods are mainly based on qualitative analysis, which is difficult to meet the needs of quantitative analysis in practical applications. Therefore, in view of the intuitive performance of the vertical development characteristics of mineral particles on the thin section scale and the urgent need for quantitative classification of sedimentary structural types in engineering practice, it is urgent to develop an automated analysis method based on the thin section scale, using rock thin section images to quantitatively characterize the vertical development characteristics of mineral particles and realize the accurate identification of the sedimentary structural type of shale felsic. Summary of the invention

[0005] In view of the above technical problems, the present invention aims to construct an automated quantitative identification method for shale felsic sedimentary structure types by using on-site rock thin section scanning image data and quantitatively characterizing the vertical development characteristics of mineral particles.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows: A rock thin section analysis method for quantitative identification of shale sedimentary structure comprises the following steps:

[0007] S01: Generate a complete scanned stitched image by scanning and stitching multiple rock thin-section photos. Since a single rock thin-section photo is limited by the limited field of view of a microscope, in order to obtain a wider range of thin-section scale information, a complete scanned stitched image can be generated by scanning and stitching multiple rock thin-section photos. This stitched image can not only provide a wider field of view, but also more comprehensively display the development characteristics of mineral particles on the thin-section scale. Since rock thin-section scanning and stitching technology is already a mature image processing method, it will not be described in detail.

[0008] S02: Place the scanned stitching image obtained in S01 in the longitudinal depth direction. The size of the scanned stitching image is H×W pixels, where H is the longitudinal dimension of the scanned stitching image, that is, from top to bottom, the depth of the rock formation is from shallow to deep, and obtain the RGB image of the scanned stitching image and the corresponding RGB color value matrix I RGB ,Right now:

[0009] I RGB = {I RGB (i, j)}, I RGB (i,j)=(R(i,j),G(i,j),B(i,j))

[0010] Among them, R, G, and B correspond to the red, green, and blue channel pixel values ​​respectively, i=1, 2,…, H; j=1, 2,…, W.

[0011] S03: In order to enhance the edges and details of the RGB color image of the rock thin section scanned and stitched in S02, the three channels of R, G, and B of the RGB image obtained in S02 are sharpened respectively;

[0012] S04: In order to make the subsequent analysis of rock mineral grain development characteristics, especially the detection of mineral grain edges and the construction of contours more efficient, the RGB image after sharpening in S03 is converted into a grayscale image, and the grayscale image brightness value matrix I is extracted. grey Grayscale provides more efficient and robust input for subsequent mineral particle edge detection, contour extraction and other processing by reducing the amount of data, extracting brightness features and enhancing contrast; the weighted average method is often used in grayscale processing.

[0013] S05: In order to achieve image smoothing and noise reduction and reduce the interference of high-frequency artifacts on subsequent morphological image operations, the grayscale image obtained by S04 is Gaussian blurred and the image matrix I is obtained. gussian ; Gaussian blur makes full use of the grayscale image brightness value matrix I grey The local information of the core scan is used to smooth and reduce the noise of the grayscale image by weighted averaging the pixel area through the Gaussian kernel G(u, v) weight.

[0014] S06: In order to highlight the main edge contour features of rock particles, the Gaussian blurred image after the Gaussian blur processing in S05 is binarized to obtain a binary image, and the corresponding binary image matrix I is obtained binary .

[0015] S07: Considering that rock particles usually have irregular shapes and complex boundaries, the binary image in S06 may contain noise (such as small particles) and incomplete areas (such as gaps between particles). In order to accurately extract the rock particle contours in the subsequent steps, the binary image obtained in S06 is subjected to morphological reconstruction operations to obtain a reconstructed image, including the following steps:

[0016] S07-1: In order to maintain the overall shape of the rock particles, not change their main contours, and remove small particle noise points, the binary image in S06 is opened and the image matrix I after the opening operation is obtained. open ;

[0017] S07-2: In order to further fill the incomplete areas in the rock particles and improve the connectivity and integrity of the rock mineral particles, the image obtained after the opening operation in S07-1 is closed again, and the image matrix I after the closed operation is obtained close ,

[0018] S08: Using the Canny edge detection algorithm to identify the closed boundaries or contours of rock particles in the reconstructed image, as well as the shapes of the closed boundaries or contours;

[0019] After completing the S07 morphological reconstruction operation, in order to clearly depict and extract the outline of rock mineral particles, the Canny edge detection algorithm is used to identify the outline and shape of rock particles and convert complex images into clear edge information. Based on the binary edge information, the boundary of the connected area, that is, the outline of the rock particles, is extracted. Given that the Canny algorithm is an existing mature technology in the field of edge detection. The Canny algorithm needs to provide a pixel threshold for strong edges and a pixel threshold for weak edges to depict edges. The above two thresholds are obtained by experiments.

[0020] S09: A rectangular study area is delineated on the scanned stitched image, and the center of the rectangular study area overlaps with the center of the scanned stitched image. After completing S08 rock grain edge detection and contour extraction, a rectangular study area is delineated near the center of the original image in order to focus the analysis. Considering that the original-size scanned stitched image contains too many rock mineral particles, the subsequent vertical development characterization of rock mineral particles only revolves around the delineated rectangular study area. The specific rock mineral particle characterization includes the following steps:

[0021] The rectangular research area is divided into a series of sub-areas vertically, that is, H pixels are divided into multiple sub-areas. Assume that each sub-area contains n ′ pixels, then the vertical subdivision area. Ceil is the ceiling function.

[0022] For each sub-region, the total number of closed boundaries or contours obtained in S08 in the sub-region is counted, that is, the number of rock mineral particles in the sub-region is counted. A trend graph of the number of corresponding rock mineral particles and the total number of rock mineral particles corresponding to each sub-region is drawn, in which the ordinate is the longitudinal position of the sub-region and the abscissa is the total number of closed boundaries or contours in the sub-region.

[0023] For each sub-region, the total area corresponding to the closed boundary obtained in S08 in the sub-region is counted, that is, the total area of ​​the rock mineral particles in the sub-region is counted. A trend chart of the corresponding total area of ​​rock mineral particles and the total area of ​​rock mineral particles corresponding to each sub-region is drawn, in which the ordinate is the longitudinal position of the sub-region, and the abscissa is the total area composed of all closed boundaries or contours in each sub-region.

[0024] For each sub-region, the average area of ​​the rock mineral particles in the sub-region is calculated, and the corresponding average area of ​​the rock mineral particles and the corresponding average area of ​​the rock mineral particles in each sub-region are plotted. The ordinate of the rock mineral particle average area change trend graph is the longitudinal position of the sub-region, and the abscissa is the average area of ​​all closed boundaries or contours in each sub-region.

[0025] S10: Analyze and observe the trend chart of the total number of rock mineral particles, the total area of ​​rock mineral particles and the average area of ​​rock mineral particles. When the trends of the three curves are consistent, the trend of the curve of the total number of rock mineral particles is used as the distinguishing mark; when the trends of the three curves are inconsistent, if the total area of ​​particles and the average area of ​​particles show opposite trends, the trend of the curve of the average area of ​​rock mineral particles is used as the distinguishing mark; if the three curves are inconsistent, and the total area of ​​particles and the average area of ​​particles show the same trend, the total number of rock mineral particles is still used as the distinguishing mark.

[0026] S101: If the identification mark shows a downward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by the gravity flow deposition is normal grain sequence bedding;

[0027] S102: According to the vertical depth from deep to shallow, if the identification mark increases first and then decreases, showing a repeated change trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is parallel bedding;

[0028] S103: If the identification mark shows an upward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by the gravity flow deposition is inverse grain sequence bedding;

[0029] S104: According to the vertical depth from deep to shallow, if the change of the identification mark is not obvious, and there is no obvious rising, falling and repeated trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is massive bedding.

[0030] Through S01-S10, the vertical development characteristics of rock mineral particles are quantitatively characterized at the rock thin section scale, and the automatic division of different sedimentary structures of shale felsic is realized.

[0031] Furthermore, in S03, the sharpening process is performed on the R channel of the RGB image according to the following formula:

[0032]

[0033] Among them, L(m, n) is the sharpening convolution kernel which can be selected according to the experiment, m and n respectively represent the indexes corresponding to the sharpening convolution kernel matrix L, L(0, 0) represents the center of the sharpening convolution kernel matrix L, and the G channel and the B channel are sharpened in the same way as the R channel.

[0034] Taking the sharpening convolution kernel matrix L of size 3×3 as an example, the values ​​of m and n can be m={-1,0,1} and n={-1,0,1} respectively. The sharpening convolution kernel matrix can be expressed in index form as follows:

[0035]

[0036] Further, in said S04, grey Use the following formula to obtain:

[0037] I grey (i, j) = 0.299R sharp (i,j)+0.587G sharp (i,j)+0.114B sharp (i, j)

[0038] Among them, I grey = {I grey (i, j)}, i=1,2,…,H; j=1,2,…,W.

[0039] Furthermore, the image matrix I is obtained in S05 gussian The process is:

[0040] The formula for Gaussian blur is:

[0041]

[0042] Among them, the Gaussian convolution kernel k controls the size of the convolution kernel, σ controls the intensity of the blur; I gussian = {I guassian (i, j)}, i=1,2,…,H; j=1,2,…,W.

[0043] Further, the S06 obtains I binary The process is:

[0044] The binarization process converts the continuous pixel values ​​of the Gaussian blurred image between 0 and 255 into discrete values ​​containing only 0 and 255, thereby obtaining the corresponding binary image matrix I binary .

[0045]

[0046] Among them, P * is the pixel value threshold, which is usually automatically selected according to Otsu's threshold method. binary = {I binary (i, j)}, i=1,2,…,H; j=1,2,…,W.

[0047] Furthermore, in S07-1, open The calculation process is as follows:

[0048] The opening operation formula is:

[0049]

[0050] Among them, B1 is the kernel matrix, x, y represent the pixel coordinates of the binary image, that is, I binary The index of each element in the kernel matrix B1, x = 1, 2, ... H, y = 1, 2, ... W, a', b' represent the index variables of each element in the kernel matrix B1, the purpose is to traverse each index value; a, b represent the index of the element in the kernel matrix B1, indicating that the current use of an element in the kernel matrix B1 and the element I in the binary image matrix binary Index to perform the opening operation on.

[0051] Taking the kernel matrix B1 of size 3×3 as an example, the values ​​of u and v can be u={-1,0,1} and v={-1,0,1} respectively. The kernel matrix B1 can be expressed in index form as follows:

[0052]

[0053] Furthermore, in S07-2, close The calculation process is as follows:

[0054]

[0055] Among them, B2 is the kernel matrix, which is usually selected based on experiments.

[0056] Compared with the prior art, the present invention has at least the following beneficial effects:

[0057] 1. The present invention uses image processing methods to scan and splice rock slice images, starting from a smaller rock slice scale, fully utilizing the vertical development characteristics of rock mineral particles, realizing the automated characterization of the vertical development of rock mineral particles, and establishing a discrimination standard for different sedimentary structural types. This method effectively avoids the inherent uncertainty of manual statistics and significantly improves the accurate discrimination of shale felsic sedimentary structural types.

[0058] 2. The rock mineral grain development characterization curve proposed in the present invention can realize the quantitative characterization of the vertical development characteristics of mineral grains, thereby realizing the quantitative division of sedimentary structure types, and enhancing the accuracy and reliability of traditional subjective division. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a technical flow chart of the present invention.

[0060] Figure 2 This is the original image of a rock slice scanned and stitched in Example 1.

[0061] Figure 3 This is a sharpened image of a rock slice scanned and stitched image in Example 1.

[0062] Figure 4 This is a Gaussian blurred image of a rock slice scanned and stitched image in Example 1.

[0063] Figure 5 This is a morphological reconstruction image of a rock slice scanned and stitched image in Example 1.

[0064] Figure 6 This is a depiction of the edge contours of rock mineral particles in a rock slice scanned and stitched image in Example 1.

[0065] Figure 7 This is a white rectangular study area map delineated by a rock thin section scanned stitching image in Example 1.

[0066] Figure 8 This is a trend chart of the number of rock mineral particles, the total area, and the average area of ​​rock mineral particles in the study area corresponding to the scanned stitching image of a certain rock slice in Example 1.

[0067] Fig. 9 This is the result diagram of quantitative identification of different sedimentary structural types of shale felsic in Example 1. DETAILED DESCRIPTION

[0068] In order to enable those skilled in the art to better understand the present invention, the present invention is further described below.

[0069] In order to further illustrate the effectiveness of the technical method, taking the scanned and stitched images of a certain stratum core as an example, combined with the drawings in the embodiments of the present invention, the rock thin section analysis method for quantitatively distinguishing shale sedimentary structure proposed in the present invention is further explained.

[0070] Example 1, see Figure 1 A rock thin section analysis method for quantitative identification of shale sedimentary structures, the method comprising the following steps:

[0071] S01: Since a single rock thin section photograph is limited by the limited field of view of a microscope, in order to obtain a wider range of thin section scale information, multiple rock thin section photographs can be scanned and stitched together to generate a complete scanned stitched image, as shown in Example 1 of this implementation. Figure 2 As shown in Figure 2, the stitched image can not only provide a wider field of view, but also more comprehensively display the development characteristics of mineral particles on the thin section scale. Since rock thin section scanning and stitching technology is already a mature image processing method, it will not be described in detail.

[0072] S02: Place the rock slice scanned mosaic image obtained in S01 in the vertical depth direction, and the image size is H×W pixels. Among them, H corresponds to the vertical dimension of the image, H=21684, that is, there are 21684 pixels from top to bottom corresponding to the depth of the rock layer from shallow to deep. Obtain the RGB color value matrix I corresponding to the image RGB ,Right now:

[0073] I RGB = {I RGB (i, j)}, I RGB (i,j)=(R(i,j),G(i,j),B(i,j))

[0074] Among them, R, G, and B correspond to the pixel values ​​of the red, green, and blue channels respectively, i = 1, 2, ..., 21684; j = 1, 2, ..., W;

[0075] S03: In order to enhance the edges and details of the rock thin section scanned and stitched RGB color image in S02, each channel of the RGB image in S02 is sharpened separately. Taking the R channel as an example, the sharpening operation can be performed according to the following formula:

[0076]

[0077] Among them, L(m, n) is the sharpening convolution kernel. The sharpening convolution kernel adopted in this implementation case is Similarly, similar sharpening operations are performed on the G channel and the B channel according to the corresponding formula. The resulting sharpened image is as shown in Example 1. Figure 3 shown.

[0078] S04: In order to make the subsequent rock mineral grain development feature analysis, especially the mineral grain edge detection and contour construction more efficient, the sharpened RGB three-channel image in S03 is converted into a grayscale image, and the grayscale image brightness value matrix I is extracted. grey Grayscale processing provides more efficient and robust input for subsequent mineral particle edge detection, contour extraction and other processing by reducing the amount of data, extracting brightness features and enhancing contrast. Grayscale processing often uses the weighted average method:

[0079] I grey (i,j)=0.299R sharp (i,j)+0.587G sharp (i,j)+0.114B sharp (i, j)

[0080] Among them, the corresponding grayscale image matrix, I grey = {I grey (i, j)}, i=1,2,…,H; j=1,2,…,W.

[0081] S05: In order to achieve image smoothing and noise reduction and reduce the interference of high-frequency artifacts on subsequent morphological image operations, the grayscale image matrix in S04 is Gaussian blurred to obtain the corresponding processed image matrix I gussian Gaussian blur makes full use of the local information of the grayscale image matrix and performs weighted averaging on the pixel area through the Gaussian kernel G(u,v) weight to achieve smoothing and noise reduction of the core scanning and splicing grayscale image. The corresponding processing formula of Gaussian blur is:

[0082]

[0083] In this implementation case, k=3 is used, and the Gaussian convolution kernel used is The resulting Gaussian blurred image is shown in Example 1. Figure 4 shown.

[0084] S06: In order to highlight the main edge contour features of rock particles, the image after Gaussian blur in S05 is binarized. Binarization converts the continuous pixel values ​​of the grayscale image between 0 and 255 into discrete values ​​containing only 0 and 255, thereby obtaining the corresponding binary image matrix I binary The conversion formula corresponding to binarization is:

[0085]

[0086] Among them, P * is the pixel value threshold, which is usually automatically selected according to Otsu's threshold method. Since Otsu's threshold method is not the technical content of the present invention, it will not be described in detail.

[0087] S07: Considering that rock particles usually have irregular shapes and complex boundaries, the binary image in S06 may contain noise (such as small particles) and incomplete areas (such as gaps between particles). In order to accurately extract the rock particle contours in the subsequent steps, the binary image in S06 is subjected to morphological reconstruction. The main steps include:

[0088] S07-1: In order to maintain the overall shape of the rock particles, not change their main contours, and remove small particle noise points, the binary image in S06 is opened. The formula for the opening operation is:

[0089]

[0090] Among them, the kernel matrix used

[0091] S07-2: In order to further fill the incomplete areas in the rock particles and improve the connectivity and integrity of the rock mineral particles, the image obtained by the S071 open operation is closed again. The closing operation formula is as follows:

[0092]

[0093] Among them, the kernel matrix used After completing the morphological reconstruction operation, the corresponding image is as shown in Example 1 Figure 5 shown.

[0094] S08: After completing the morphological reconstruction operation of S07, in order to clearly depict and extract the contours of rock mineral particles, the Canny edge detection algorithm is used to identify the contours and shapes of rock particles, and convert complex images into clear edge information. Based on the binary edge information, the boundaries of connected areas, that is, the contours of rock particles, are extracted. In view of the fact that the Canny algorithm is an existing mature technology in the field of edge detection, the present invention focuses on using the Canny algorithm to depict rock particle contour information, so the specific implementation method of the algorithm will not be repeated. It is worth noting that the Canny algorithm needs to provide a pixel threshold for strong edges and a pixel threshold for weak edges to depict edges. The two thresholds used in this implementation case are 50 and 150 respectively. After completing the edge depiction of rock mineral particles, the corresponding image is this implementation case 1. Figure 6 Shown

[0095] S09: After completing S08 rock particle edge detection and contour extraction, in order to focus the analysis, a rectangular research area is delineated near the center of the original image, as shown in Example 1 of this implementation Figure 7 Rectangular selection area. Considering that the original size scanned spliced ​​image contains too many rock mineral particles, the subsequent vertical development characterization of rock mineral particles only revolves around the defined rectangular study area. The specific rock mineral particle characterization includes the following steps:

[0096] S09-1: Divide the defined rectangular research area into a series of sub-areas in the vertical direction, and divide the vertical H = 21738 pixels into multiple sub-areas. Assume that each sub-area contains n ′ = 139 pixels, then the vertical subdivision area. Ceil is the ceiling function.

[0097] S092: Count the total number of closed boundaries or contours obtained in S08 in each sub-region, that is, count the number of rock mineral particles in the sub-region. Draw a corresponding trend chart of the number of rock mineral particles and the corresponding sub-regions, such as the example 1 in this implementation example. Figure 8 shown.

[0098] S093: Count the total area corresponding to the closed boundary obtained in S08 in each sub-region, that is, count the total area of ​​rock mineral particles in the sub-region. Draw a trend chart of the corresponding total area of ​​rock mineral particles and the corresponding change trend of each sub-region, such as the one in this implementation example 1 Figure 8 shown.

[0099] S094: According to S092 and S093, for each sub-region, the average area of ​​the rock mineral particles in the sub-region is calculated. A corresponding trend chart of the average area of ​​the rock mineral particles and the corresponding sub-region is drawn, as shown in Example 1 of this embodiment. Figure 8 shown.

[0100] S10: Combined with the total number of rock mineral particles, total area and average area of ​​rock mineral particles drawn in S09, further analyze the change trend and classify different sedimentary structure types. It mainly includes the following steps:

[0101] S101: Analyze and observe the trend chart of the total number of rock mineral particles, the total area of ​​rock mineral particles, and the average area of ​​rock mineral particles. When the trends of the three curves are consistent, the trend of the total number of rock mineral particles is used as a distinguishing mark; when the trends of the three curves are inconsistent, if the total area of ​​particles and the average area of ​​particles show opposite trends, the trend of the average area of ​​rock mineral particles is used as a distinguishing mark; when the three curves are inconsistent, and the total area of ​​particles and the average area of ​​particles show the same trend, the total number of rock mineral particles is still used as a distinguishing mark.

[0102] S101: If the identification mark shows a downward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by the gravity flow deposition is normal grain sequence bedding;

[0103] S102: According to the vertical depth from deep to shallow, if the identification mark increases first and then decreases, showing a repeated change trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is parallel bedding;

[0104] S103: If the identification mark shows an upward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by the gravity flow deposition is inverse grain sequence bedding;

[0105] S104: According to the vertical depth from deep to shallow, if the change of the identification mark is not obvious, and there is no obvious rising, falling and repeated trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is massive bedding.

[0106] S11: According to this implementation case 1 Fig. 9 As shown, combined with this implementation case 1 Figure 8 The total number, total area and average area trend diagram of rock mineral particles are divided into inverse grain sequence bedding, normal grain sequence bedding, parallel bedding, massive bedding, parallel bedding and massive bedding according to the depth of the stratum from deep to shallow.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A rock thin section analysis method for quantitative identification of shale sedimentary structure, characterized in that: The steps include: S01: Generate a complete scanned and stitched image by scanning and stitching multiple rock thin section photos; S02: Place the scanned stitched image obtained in S01 in the longitudinal depth direction. The scanned stitched image size is H×W pixels, where H is the longitudinal dimension of the scanned stitched image, i.e., the depth of the rock formation from top to bottom is from shallow to deep, and obtain the RGB image of the scanned stitched image and the corresponding RGB color value matrix I RGB ,Right now: I RGB ={I RGB (i,j)},I RGB (i,j)=(R(i,j),G(i,j),B(i,j)) Among them, R, G, and B correspond to the red, green, and blue channel pixel values ​​respectively, i = 1, 2, ..., H; j = 1, 2, ..., W; S03: sharpening the R, G, and B channels of the RGB image obtained in S02 respectively; S04: Convert the RGB image after sharpening in S03 into a grayscale image, and extract the grayscale image brightness value matrix I grey ; S05: Perform Gaussian blur processing on the grayscale image obtained in S04, and obtain the image matrix I gussian ; S06: Binarize the Gaussian blurred image after the Gaussian blur processing in S05 to obtain a binary image, and obtain the corresponding binary image matrix I binary ; S07: Performing a morphological reconstruction operation on the binary image obtained in S06 to obtain a reconstructed image, including the following steps: S07-1: Perform an opening operation on the binary image in S06, and obtain an image matrix I after the opening operation. open ; S07-2: Perform a closing operation on the image obtained after the opening operation in S07-1, and obtain an image matrix I after the closing operation. close ; S08: Using the Canny edge detection algorithm to identify the closed boundaries or contours of rock particles in the reconstructed image, as well as the shapes of the closed boundaries or contours; S09: A rectangular study area is defined on the scanned stitched image, the center of which overlaps with the center of the scanned stitched image, and the defined rectangular study area is subdivided into a series of sub-areas in the longitudinal direction, that is, H pixels in the longitudinal direction are subdivided into multiple sub-areas. Assuming that each sub-area contains n′ pixels, the longitudinal subdivision is area; Ceil is the upward rounding function; For each sub-region, the total number of closed boundaries or contours obtained in S08 in the sub-region is counted, that is, the number of rock mineral particles in the sub-region is counted; a trend graph of the corresponding number of rock mineral particles and the total number of rock mineral particles corresponding to each sub-region is drawn, wherein the ordinate in the trend graph of the total number of rock mineral particles is the longitudinal position of the sub-region, and the abscissa is the total number of closed boundaries or contours in the sub-region; For each sub-region, the total area corresponding to the closed boundary obtained in S08 in the sub-region is counted, that is, the total area of ​​the rock mineral particles in the sub-region is counted; a trend graph of the corresponding total area of ​​rock mineral particles and the total area of ​​rock mineral particles corresponding to each sub-region is drawn, in which the ordinate is the longitudinal position of the sub-region, and the abscissa is the total area composed of all closed boundaries or contours in each sub-region; For each sub-region, the average area of ​​the rock mineral particles in the sub-region is calculated, and a trend chart of the corresponding average area of ​​the rock mineral particles and the average area of ​​the rock mineral particles corresponding to each sub-region is drawn. The ordinate of the trend chart of the average area of ​​the rock mineral particles is the longitudinal position of the sub-region, and the abscissa is the average area of ​​all closed boundaries or contours in each sub-region; S10: Analyze and observe the trend graph of the total number of rock mineral particles, the total area of ​​rock mineral particles, and the average area of ​​rock mineral particles. When the trends of the three curves are consistent, the trend of the total number of rock mineral particles is used as a distinguishing mark; when the trends of the three curves are inconsistent, if the total area of ​​particles and the average area of ​​particles show opposite trends, the trend of the average area of ​​rock mineral particles is used as a distinguishing mark; when the three curves are inconsistent, and the total area of ​​particles and the average area of ​​particles show the same trend, the total number of rock mineral particles is still used as a distinguishing mark; S101: If the identification mark shows a downward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by gravity flow deposition is normal grain sequence bedding; S102: According to the vertical depth from deep to shallow, if the identification mark increases first and then decreases, showing a repeated change trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is parallel bedding; S103: If the identification mark shows an upward trend from deep to shallow in the vertical direction, it indicates that the sedimentary structure type formed by the gravity flow deposition is inverse grain sequence bedding; S104: According to the vertical depth from deep to shallow, if the change of the identification mark is not obvious, and there is no obvious rising, falling and repeated trend, it indicates that the sedimentary structure type formed by the gravity flow deposition is massive bedding.

2. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 1, characterized in that: In S03, the sharpening process is performed on the R channel of the RGB image according to the following formula: Among them, L(m,n) is the sharpening convolution kernel, m and n represent the indexes corresponding to the sharpening convolution kernel matrix L respectively, L(0,0) represents the center of the sharpening convolution kernel matrix L, and the G channel and B channel are sharpened in the same way as the R channel.

3. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 2, characterized in that: S04 I grey Use the following formula to obtain: I grey (i,j)=0.299R sharp (i,j)+0.587G sharp (i,j)+0.114B sharp (i,j) where I grey ={I grey (i, j)}, i = 1, 2, …, H; j = 1, 2, …, W.

4. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 3, characterized in that: The image matrix I is obtained in S05 gussian The process is: The formula for Gaussian blur is: Among them, the Gaussian convolution kernel k controls the size of the convolution kernel, σ controls the intensity of the blur; I gussian = {I guassian (i, j)}, i=1,2,…,H; j=1,2,…,W.

5. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 4, characterized in that: In S06, I binary The process is: Among them, P * is the pixel value threshold, I binary = {I binary (i, j)}, i=1,2,…,H; j=1,2,…,W.

6. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 5, characterized in that: S07-1 I open The calculation process is as follows: The opening operation formula is: Among them, B1 is the kernel matrix, x, y represent the pixel coordinates of the binary image, that is, I binary The index of each element in the kernel matrix B1, x = 1, 2, ... H, y = 1, 2, ... W, a', b' represent the index variables of each element in the kernel matrix B1, the purpose is to traverse each index value; a, b represent the index of the element in the kernel matrix B1, indicating that the current use of an element in the kernel matrix B1 and the element I in the binary image matrix binary Index to perform the opening operation on.

7. A rock thin section analysis method for quantitative identification of shale sedimentary structure according to claim 6, characterized in that: S07-2 I close The calculation process is as follows: Among them, B2 is the kernel matrix.

Images