A method for identifying laminar structure of marine-continental transitional shale based on rock thin section images

Abstract

The present invention discloses a method for identifying the laminar structure of marine-continental transitional shale based on rock thin-section images, and the main steps are: first, grayscale transformation, Fourier transformation and spectrum feature extraction are performed on the rock thin-section image; then, principal component analysis is performed on the point set in the frequency domain image, the first principal component corresponds to the long axis in the point set distribution, and the second principal component corresponds to the short axis in the point set distribution; with the first and second principal components as coordinate axes, the data points on the frequency domain image are projected onto the first and second principal component axes respectively to obtain the histogram of the frequency domain image on the two coordinate axes; the ratio of the widths of the two histograms is used to quantitatively characterize the morphological characteristics of the distribution of data points on the frequency domain image; the smaller the ratio, the more obvious the laminar structure in the corresponding original image. The method of the present invention directly obtains the structural characteristics of the core through the frequency domain feature analysis of the digital image, without the need for image segmentation, and is suitable for cores with strong heterogeneity and difficult laminar image segmentation, such as marine-continental transitional shale.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas field exploitation, and in particular to a method for identifying marine-continental transitional shale laminar structures based on rock thin section images. Background Art

[0002] The marine-continental transitional shale resources are large and have great potential for exploration and development, but their sedimentary environment is changeable and the rock structure is complex. If there is a laminar structure in the shale, its rock physics, mechanics and other properties are mostly anisotropic, which is significantly different from shales with other structures. Therefore, before rock physics modeling and reservoir evaluation, it is of great significance to identify shales with laminar structures based on rock thin section data. There are two commonly used methods. One is manual identification. The main disadvantages of this method are that it is time-consuming, dependent on the geological level of the interpreter, and difficult to quantify the degree of laminar development. In actual engineering projects, the number of samples is often large. Manual division is not only time-consuming and labor-intensive, but also prone to misselection and wrong selection, and cannot provide quantitative characterization of the laminar structure. The other is to identify laminae through image segmentation based on digital image processing technology.

[0003] Extraction of rock thin sections based on image segmentation technology. The specific method is to convert the rock thin section image into a grayscale image, binarize the rock thin section through color segmentation, count the number of white pixels (i.e., pixels with an amplitude of 1) by row or column, draw a curve graph, and analyze the lamination development through the fluctuation of the curve. If the peak of the curve surges locally, it is considered that there is a lamination structure. The recognition effect of this method is poor and its scope of application is limited. When the lamination is nearly parallel to the horizontal axis, this method has a good recognition effect, but when there is an angle between the lamination structure and the horizontal axis, the recognition effect is poor. When light-colored non-laminar structure minerals are present, misidentification is prone to occur. For example Figure 1 As shown, although Figure 1 The left image shows that the [600,1000] line has an inclined laminar structure, but the curve peak does not increase significantly in the right image. If the entire image is an inclined laminar structure, the recognition effect will be very poor. As shown in Figure 2, when threshold segmentation is performed, the boundary value is difficult to determine. When there are light-colored gravels in the rock slice, white pixels will also be generated. Figure 2 In the right figure, the peak value of the curve at lines 400 and 800 reaches about 550, which is consistent with Figure 1 The peaks of the inclined lamination structure in the right image are comparable, but the rock does not have a lamination structure. Figure 1 and Figure 2 In the middle right figure, the ordinate represents the row number of the image, and the abscissa represents the number of white pixels in each row of the binary image. Therefore, for shale thin section images, which are mostly highly heterogeneous, this method is difficult to segment the image. Summary of the invention

[0004] The purpose of the present invention is to provide a method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images in order to address the above-mentioned shortcomings of the existing methods for identifying laminar structures based on rock thin section data, which makes it difficult to accurately identify the laminar structure of marine-continental transitional shale.

[0005] The present invention provides a method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images, and the steps are as follows:

[0006] S1. Grayscale conversion of rock slice images; after grayscale conversion, the image data format needs to be converted to double precision, and the grayscale range of the image changes from the original [0,255] to [0,1]. The purpose is to prevent data overflow or insufficient precision in the subsequent Fourier transform process to obtain the spectrum.

[0007] S2, two-dimensional Fourier transform and spectrum feature extraction, as follows:

[0008] S21. Perform a two-dimensional Fourier transform on the rock slice image after grayscale transformation to obtain a frequency spectrum. The two-dimensional Fourier transform formula is as follows:

[0009]

[0010] Where f(x, y) is a digital image of size M×N, F(u, v) is the frequency spectrum corresponding to f(x, y); x and y represent the image space domain coordinate variables, u and v are frequency domain discrete variables, u=0,1,2,…,M-1,υ=0,1,2,…,N-1, j 2 =-1.

[0011] S22, phase transform the frequency spectrum to move the transformation origin to the center of the frequency spectrum, then take the logarithm of the amplitude of the image, and finally binarize the image by setting a threshold to make the spectrum characteristics more obvious. When the image is binarized, each rock slice image is processed using the same threshold setting standard, and preferably, the threshold is 0.5 times the maximum amplitude of the frequency spectrum pixel point.

[0012] S3. Analysis of distribution characteristics of frequency domain images: The major axis and minor axis of the frequency domain image distribution range are used to quantitatively characterize the shape of the frequency domain image. The specific method is to perform principal component analysis on the point set in the frequency domain image, extract the first and second principal components, the first principal component corresponds to the major axis in the point set distribution, and the second principal component corresponds to the minor axis in the point set distribution. The distribution shape of the point set is analyzed by comparing the distribution characteristics of the point set on the major axis and the minor axis.

[0013] S4. Quantitative characterization of frequency domain image distribution features, as follows:

[0014] S41. With the extracted first and second principal components as horizontal coordinates and the frequency as vertical coordinates, the data points on the frequency domain graph are projected onto the first and second principal component axes respectively to obtain the histogram distribution of the frequency domain graph on the two coordinate axes.

[0015] S42, eliminating data points near the boundary of the histogram distribution, the total amount of eliminated data points accounting for 10% of the total amount of data points, the purpose of which is to reduce the influence of singular points on the frequency spectrum characteristics.

[0016] S43, using the ratio of the width of the data point distribution histogram on the first principal component axis to the width of the data point distribution histogram on the second principal component axis to quantitatively characterize the morphological characteristics of the data point distribution on the frequency domain image; the smaller the ratio of the widths, the more the distribution morphology of the points in the frequency domain image tends to be elliptical, and the more obvious the laminar structure in the corresponding original image. Preferably, when the ratio of the widths of the two histograms is less than 0.8, it is determined that there is a laminar structure on the rock thin section image.

[0017] Compared with the prior art, the present invention is beneficial in that:

[0018] (1) Compared with manual recognition methods and digital image analysis methods based on image segmentation technology, this method directly obtains the structural characteristics of the core through frequency domain feature analysis of digital images without image segmentation. It is particularly suitable for cores with strong heterogeneity and difficult laminar image segmentation, such as marine-continental transitional shale, and is also applicable to rocks with incompletely parallel stratification.

[0019] (2) Low engineering investment. This method can effectively reduce the manpower and time costs of rock thin section laminar structure sorting, and meets the economic index requirements of actual engineering.

[0020] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 , illustration of existing conventional recognition methods.

[0022] Figure 2 , illustration of existing conventional recognition methods.

[0023] Figure 3 , marine-continental transitional shale lamination identification diagram. DETAILED DESCRIPTION

[0024] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0025] The present invention provides a method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images, comprising the following steps:

[0026] (1) Grayscale transformation of rock thin section images

[0027] Rock slice images are color digital images, which are composed of a finite number of elements. Each element has a corresponding position and value. These elements are called pixels. Each pixel is represented by three components: R, G, and B. The value range of each channel is [0, 255]. In order to facilitate the processing of rock slice images, it is first necessary to perform grayscale transformation to convert the color image into a grayscale image, so that the rock slice image can display more details, thereby improving the contrast of the image (contrast stretching). The different structures of rock slices give each pixel a different grayscale. Grayscale transformation can selectively highlight the target features in the image or suppress unnecessary features in the image, and can make the distribution of pixels more uniform. After the grayscale conversion, the image data format needs to be converted to double precision (double type). At this time, the grayscale range of the image changes from the original [0, 255] to [0, 1]. The purpose is to prevent data overflow or insufficient precision in the subsequent Fourier transform process to obtain the spectrum. After the grayscale transformation of the color rock slice image, the following is obtained Figure 3 Grayscale image shown in a.

[0028] (2) Two-dimensional Fourier transform and spectrum feature extraction. Fourier transform can represent any periodic function as the sum of sine functions or cosine functions of different frequencies (Fourier series). The frequency of the image converted from the spatial domain to the frequency domain is an indicator of the intensity of the grayscale change in the image, and is the gradient of the grayscale in the plane space. The faster the grayscale change, the higher the frequency value. Therefore, if there is a laminar structure in the thin film, the image frequency is low along the laminar structure, and the image frequency is high in the direction perpendicular to the laminar structure. Therefore, if there is laminar structure in the image, its distribution in the frequency domain will show a certain directionality, such as Figure 3 As shown in e.

[0029] First, the grayscale transformed rock thin section image ( Figure 3 a) Perform a two-dimensional Fourier transform to obtain a frequency spectrum ( Figure 3 b), the formula used is as follows:

[0030]

[0031] Where f(x, y) is a digital image of size M×N, F(u, v) is the frequency spectrum corresponding to f(x, y); x and y represent the image space domain coordinate variables, u and v are frequency domain discrete variables, u=0,1,2,…,M-1,υ=0,1,2,…,N-1, j 2 =-1.

[0032] Then the frequency spectrum is phase transformed to move the origin of the transformation to the center of the frequency spectrum, and we get Figure 3 c. The amplitude range after Fourier transform is large, which is not conducive to observing the characteristics of frequency domain images. It is necessary to take the logarithm of the amplitude of the image. Figure 3 d is the image after logarithmic transformation.

[0033] Finally, in order to analyze the distribution shape of data points in the frequency domain image, the image is binarized by setting a threshold to make the spectrum features more obvious, such as Figure 3 In order to provide reliable comparison between different images, each rock thin section image should be processed with the same threshold setting standard, and the threshold is usually 0.5 times the maximum amplitude of the frequency spectrum pixel point.

[0034] (3) Analysis of the distribution characteristics of frequency domain images. The shape of the frequency domain image reflects the structural information of the original image. If the image is circular or approximately circular, it indicates that there is no obvious lamellar structure in the image. If the frequency domain image is elliptical, it indicates that the original image has a lamellar structure. The shape of the frequency domain image is quantitatively characterized by the major and minor axes of the distribution range of the frequency domain image. The specific method is: perform principal component analysis on the point set in the frequency domain image, extract the first and second principal components, where the first principal component corresponds to the major axis of the point set distribution, and the second principal component corresponds to the minor axis of the point set distribution. By comparing the distribution characteristics of the point set on the major axis and the minor axis, the distribution shape of the point set can be effectively analyzed, such as Figure 3 f.

[0035] (4) Quantitative characterization of the distribution characteristics of the frequency domain image. With the extracted first and second principal components as the horizontal axis, the data points on the frequency domain image are projected onto the first and second principal component axes respectively, and the histogram distribution of the frequency domain image on the two axes can be obtained (Figure 3g), where the F1 histogram is the distribution of the data points on the frequency domain image on the second principal component axis (short axis), and the F2 histogram is the distribution of the data points on the frequency domain image on the first principal component axis (long axis). In order to reduce the impact of singular points on the spectral characteristics, data points near the distribution boundary are removed, and the total amount of removed data points accounts for about 10% of the total number of data points. The new histogram obtained after removing the boundary points is as follows: Figure 3 h. The ratio of the width of the data point distribution histogram on the first principal component axis to the width of the data point distribution histogram on the second principal component axis is used to quantitatively characterize the morphological characteristics of the data point distribution on the frequency domain image. Figure 3 The histogram width ratio in h is 0.6519. The range of the histogram width ratio is between [0,1]. The smaller the ratio, the more the distribution of points in the frequency domain image tends to be elliptical, and the more obvious the laminar structure in the corresponding original image. Usually, when the histogram width ratio is less than 0.8, it is determined that there is a laminar structure on the rock thin section image.

[0036] The above description is only a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of the technical solution of the present invention.

Claims

1. A method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images, characterized in that: Here are the steps: S1, grayscale transformation of rock thin section image; S2, two-dimensional Fourier transform and spectrum feature extraction, as follows: S21, performing a two-dimensional Fourier transform on the rock thin section image after the grayscale transformation to obtain a frequency spectrum; S22, performing phase transformation on the frequency spectrum to move the transformation origin to the center of the frequency spectrum, then taking the logarithm of the amplitude of the image, and finally binarizing the image by setting a threshold to make the spectrum features more obvious; S3. Frequency domain image distribution feature analysis: The long axis and short axis of the frequency domain image distribution range are used to quantitatively characterize the shape of the frequency domain image. The specific method is to perform principal component analysis on the point set in the frequency domain image, extract the first and second principal components, the first principal component corresponds to the long axis in the point set distribution, and the second principal component corresponds to the short axis in the point set distribution; the distribution shape of the point set is analyzed by comparing the distribution features of the point set on the long axis and the short axis. S4. Quantitative characterization of frequency domain image distribution features, as follows: S41, using the extracted first and second principal components as abscissas and the frequency as ordinates, projecting the data points on the frequency domain graph onto the first and second principal component axes respectively, to obtain a histogram distribution of the frequency domain graph on two coordinate axes; S42, removing data points near the boundary of the histogram distribution to reduce the influence of singular points on the spectrum characteristics; S43, using the ratio of the width of the data point distribution histogram on the first principal component axis to the width of the data point distribution histogram on the second principal component axis to quantitatively characterize the morphological characteristics of the data point distribution on the frequency domain image; The smaller the ratio of the widths, the more the distribution of points in the frequency domain image tends to be elliptical, and the more obvious the laminar structure in the corresponding original image.

2. The method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images according to claim 1, characterized in that: In step S1, after the grayscale conversion, the data format of the image needs to be converted into a double-precision type, and the grayscale range of the image changes from the original [0,255] to [0,1].

3. The method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images according to claim 1, characterized in that: In step S21, the two-dimensional Fourier transform formula is as follows: Where f(x, y) is a digital image of size M×N, F(u, v) is the frequency spectrum corresponding to f(x, y); x and y represent the image space domain coordinate variables, u and v are frequency domain discrete variables, u=0,1,2,…,M-1,υ=0,1,2,…,N-1,j 2 =-1.

4. The method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images according to claim 3, characterized in that: In step S22, when the image is binarized by setting a threshold, each rock thin section image is processed using the same threshold setting standard, and the threshold is 0.5 times the maximum amplitude of the frequency spectrum pixel point.

5. The method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images according to claim 1, characterized in that: In step S42, the total amount of eliminated data points accounts for 10% of the total amount of data points.

6. The method for identifying the laminar structure of marine-continental transitional shale based on rock thin section images according to claim 1, characterized in that: In step S43, when the ratio of the widths of the two histograms is less than 0.8, it is determined that a laminar structure exists on the rock slice image.

Images