Lithology identification method, system and device based on scanning electron microscope

Abstract

The present application relates to the technical field of image recognition, and particularly relates to a lithology identification method, system and device based on a scanning electron microscope, comprising: dividing a backscattered electron image into a plurality of superpixel blocks; determining a composition distribution fineness factor of the backscattered electron image according to the area and gray scale of the superpixel blocks with similar gray scale, and the position distribution; determining a local composition complexity expression factor of the backscattered electron image according to the area size and gray scale difference of adjacent superpixel blocks; determining a block adjustment coefficient of the backscattered electron image to obtain an optimal block window size; performing an enhancement operation on the backscattered electron image to obtain an enhanced backscattered electron image; and using the enhanced backscattered electron image for lithology identification. The present application performs efficient enhancement on the backscattered electron image through an adaptive block window size, thereby ensuring the accuracy of lithology identification.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and more specifically to a method, system, and device for lithology identification based on scanning electron microscopy. Background Technology

[0002] Lithological identification is fundamental to geological research, oil and gas exploration, mineral resource evaluation, engineering geology, and sedimentary diagenetic analysis. Its core task is to determine rock type, mineral composition, structure, porosity, and diagenetic evolution history. Scanning electron microscopy (SEM) is a key supporting technology for modern lithological identification. It has advantages such as high-resolution imaging, micro-area compositional analysis (EDS), and morphology-composition linkage characterization. Furthermore, it can observe rock surface morphology, mineral crystal morphology, grain contact relationships, and pore and fracture structures at the nanometer to micrometer scale, and, combined with energy dispersive spectroscopy (EDS), achieve in-situ mineral composition identification.

[0003] Existing problems: In scanning electron microscopy (SEM) images used for lithological identification, backscattered electron imaging (BSE) primarily relies on atomic number contrast imaging, which generally suffers from weak signal intensity and low signal-to-noise ratio. When the atomic number differences between different minerals are small, this further leads to low image contrast and blurred mineral boundaries. Furthermore, rock samples themselves are characterized by complex mineral compositions and significant variations in mineral content distribution. BSE images are prone to exhibiting similar mineral grayscale levels and unclear microscopic details, directly causing difficulties in mineral identification, inaccurate boundary delineation, and large errors in micro-area composition interpretation, severely impacting the accuracy and reliability of lithological identification. Summary of the Invention

[0004] This invention provides a method, system, and device for lithological identification based on scanning electron microscopy to solve existing problems.

[0005] The lithology identification method, system, and equipment based on scanning electron microscopy of the present invention adopt the following technical solution: One embodiment of the present invention provides a lithology identification method based on scanning electron microscopy, the method comprising the following steps: Acquire backscattered electron images for lithological identification; The backscattered electron image is divided into several superpixel blocks; the component distribution fragmentation factor of the backscattered electron image is determined based on the area and gray level of the superpixel blocks with similar gray levels, as well as their positional distribution. Based on the area size and grayscale difference of adjacent superpixel blocks, the local component complexity of each superpixel block is determined; based on the magnitude of the local component complexity, several component-complex superpixel blocks are selected; based on the number and positional distribution of component-complex superpixel blocks, combined with the magnitude of the local component complexity, the local component complexity representation factor of the backscattered electron image is determined. Based on the magnitudes of the local component complexity factor and the component fragmentation factor of the backscattered electron image, the block adjustment coefficient of the backscattered electron image is determined; based on the magnitude of the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithological identification.

[0006] Furthermore, the specific steps for determining the component distribution fragmentation factor of the backscattered electron image are as follows: The average gray value of each superpixel block is the mean of the gray values ​​of all pixels within each superpixel block. Using the absolute value of the difference between the average gray values ​​of any two superpixel blocks as the clustering distance, clustering operations are performed on all superpixel blocks to obtain several clusters; The average area of ​​all superpixel blocks is obtained and denoted as the relative area threshold. For all superpixel blocks in each cluster, the ratio of the number of superpixel blocks with an area smaller than the relative area threshold to the total number of superpixel blocks is denoted as the proportion of tiny particles. The mean of the areas of all superpixel blocks is denoted as the area expressivity. The mean of the shortest distances between all pairs of superpixel blocks is denoted as the particle distribution dispersion. The mean of the complement normalized value of the area expressivity and the normalized value of the particle distribution dispersion is denoted as the quantity adjustment coefficient. The product of the quantity adjustment coefficient and the proportion of tiny particles is denoted as the fragmented distribution expressivity of each cluster. The component distribution fragmentation factor of the backscattered electron image is determined based on the average gray value of the superpixel block in each cluster and the fragmentation distribution performance value.

[0007] Furthermore, the specific steps for determining the component distribution fragmentation factor of the backscattered electron image based on the average gray value of the superpixel blocks in each cluster and the fragmentation distribution performance value are as follows: The mean of the average gray values ​​of all superpixel blocks in each cluster is obtained and recorded as the average gray value of each cluster. The component distribution fragmentation factor of the backscattered electron image is determined by weighting the weighted average of the fragmented distribution values ​​of all clusters using the complement normalized value of the average gray value of each cluster.

[0008] Furthermore, the specific steps for determining the local component complexity of each superpixel block are as follows: Let any superpixel block be denoted as the target superpixel block; All superpixel blocks adjacent to the target superpixel block are designated as reference superpixel blocks; The mean of the absolute values ​​of the differences between the average gray value of the target superpixel block and the average gray value of all reference superpixel blocks is denoted as the drastic change of local components. The target superpixel block and all reference superpixel blocks are collectively referred to as the neighborhood superpixel block; The normalized value of the complement of the sum of the areas of all neighboring superpixel blocks is denoted as the local component density. The normalized value of the number of cluster types to which all neighboring superpixel blocks belong is denoted as the local component type quantization value; The mean of the local component density and the local component type quantization value is denoted as the local component feature adjustment value. The product of the local component feature adjustment value and the local component drastic change is denoted as the local component complexity of the target superpixel block.

[0009] Furthermore, the specific steps for selecting several superpixel blocks with complex composition are as follows: Superpixel blocks whose normalized value of local component complexity is greater than a preset component complexity threshold are denoted as component complex superpixel blocks.

[0010] Furthermore, the specific steps for determining the local component complexity representation factor of the backscattered electron image are as follows: Obtain the ratio of the number of all complex superpixel blocks to the total number of superpixel blocks, and denote it as the proportion of complex micro-regions. The normalized value of the mean of the shortest distances between all pairs of superpixel blocks with complex composition is denoted as the discreteness of the distribution of small regions with complex composition. The average of the proportion of complex micro-regions and the dispersion of their distribution is denoted as the component complexity correction coefficient. The product of the average of the normalized values ​​of the local component complexity of all superpixel blocks and the component complexity correction coefficient is denoted as the local component complexity representation factor of the backscattered electron image.

[0011] Furthermore, the specific steps for determining the block adjustment coefficients of the backscattered electron image are as follows: The product of the factor representing the complexity of local components and the factor representing the fragmentation of component distribution in a backscattered electron image is denoted as the block adjustment coefficient of the backscattered electron image.

[0012] Furthermore, the specific steps for determining the optimal block window size are as follows: If the normalized value of the block adjustment coefficient of the backscattered electron image is less than or equal to the preset first threshold, then the optimal block window size is set to the preset third window. If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than the preset first threshold and less than the preset second threshold, then the optimal block window size is set to the preset second window. If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than or equal to the preset second threshold, then the optimal block window size is set to the preset first window size.

[0013] The lithological identification system based on scanning electron microscopy, employing any one of the lithological identification methods based on scanning electron microscopy, includes the following modules: A scanning electron microscope (SEM) image acquisition module for rock samples is used to acquire backscattered electron images for lithology identification. The rock composition distribution fragmentation analysis module is used to divide the backscattered electron image into several superpixel blocks; based on the area and gray level of the superpixel blocks with similar gray levels, as well as their positional distribution, the composition distribution fragmentation factor of the backscattered electron image is determined. The rock composition local complexity analysis module is used to determine the local composition complexity of each superpixel block based on the area size and grayscale difference of adjacent superpixel blocks; to select several compositionally complex superpixel blocks based on the magnitude of the local composition complexity; and to determine the local composition complexity representation factor of the backscattered electron image based on the number and location distribution of compositionally complex superpixel blocks and the magnitude of the local composition complexity. The lithology identification module is used to determine the block adjustment coefficient of the backscattered electron image based on the magnitude of the local component complexity representation factor and the component distribution fragmentation factor; based on the magnitude of the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithology identification.

[0014] The present invention also proposes a lithology identification device based on scanning electron microscopy, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned lithology identification method based on scanning electron microscopy.

[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, the backscattered electron image is divided into several superpixel blocks. Based on the area, grayscale value, and positional distribution of superpixel blocks with similar grayscale, a component distribution fragmentation factor is determined for the backscattered electron image. This component distribution fragmentation factor allows for adaptive determination of the block window size, achieving image enhancement that better matches the characteristics of rock composition. Based on the area and grayscale differences of adjacent superpixel blocks, a local component complexity factor is determined for the backscattered electron image, thereby determining the block adjustment coefficient. The component distribution fragmentation factor and the local component complexity are then fused to obtain the final block adjustment coefficient. Based on this coefficient, the block window size is adaptively determined, ensuring that the block window adapts to both the overall fragmented distribution characteristics of the rock composition and the drastic changes in composition in local areas, achieving more accurate image enhancement that better meets the needs of lithological identification. Based on the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithological identification. This invention enhances backscattered electron images by adaptively adjusting the block window size, significantly improving image contrast and clarity. It effectively addresses issues such as similar mineral grayscale, blurred boundaries, and loss of detail caused by small differences in atomic number, weak signals, and low signal-to-noise ratio. Furthermore, by strengthening the contrast differences between different mineral phases, it can clearly distinguish mineral types, accurately delineate mineral boundaries, and highlight microstructural features, reducing errors in mineral identification and compositional interpretation. This improves the accuracy, reliability, and efficiency of lithological identification, providing a high-quality image foundation for subsequent micro-area compositional analysis, quantitative mineral statistics, and diagenetic studies. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the steps of the lithology identification method based on scanning electron microscopy of the present invention. Figure 2 This is a structural block diagram of the lithology identification system based on scanning electron microscopy of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the lithology identification method, system, and device based on scanning electron microscopy proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following description, in conjunction with the accompanying drawings, details the specific solutions for the lithology identification method, system, and equipment based on scanning electron microscopy provided by this invention.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a lithology identification method based on scanning electron microscopy according to an embodiment of the present invention. The method includes the following steps: Step S001: Acquire backscattered electron images for lithological identification.

[0022] Obtain scanning electron microscope (SEM) images for lithological identification, specifically: (1) Place the rock sample after polishing, drying and conductive treatment on the scanning electron microscope sample stage and evacuate the sample chamber to the working vacuum level.

[0023] (2) Set the working parameters of the scanning electron microscope, including the accelerating voltage, working distance, beam current and magnification, so that the electron beam is stably focused on the sample surface.

[0024] (3) Use an electron beam to scan the surface of the rock sample point by point to generate secondary electron signals and backscattered electron signals.

[0025] (4) Secondary electron signals and backscattered electron signals are collected by detectors respectively, and after signal amplification and analog-to-digital conversion, corresponding secondary electron images and backscattered electron images are formed.

[0026] Secondary electron images are used to reflect the microscopic morphology, grain shape, structure, and porosity of the rock surface; backscattered electron images, based on atomic number contrast, are used to distinguish different mineral phases and compositional distributions. Within the selected observation area, image acquisition, storage, and output are completed at a preset magnification.

[0027] The image acquisition process described above is a standard and well-known procedure for scanning electron microscopy analysis, and it is widely used in rock microstructure observation and lithology identification.

[0028] It should be noted that secondary electron images (BSEs) primarily reflect the microscopic morphology of the sample surface. They exhibit high signal intensity, good signal-to-noise ratio, rich image detail, and naturally high contrast, and are typically used directly for morphological observation. Therefore, in this embodiment, only the backscattered electron image (BSE) is enhanced during SEM lithology identification. Furthermore, bilateral filtering is first used to denoise the BSE, suppressing scanning noise while preserving image edge information. Then, the filtered and denoised BSE is enhanced to improve image contrast and detail resolution. The BSE is a single-channel grayscale image with a grayscale range of 0 to 255. The grayscale value of each pixel directly corresponds to the local average atomic number of the sample. Bilateral filtering is a well-known technique, and its specific method will not be described here.

[0029] Step S002: Divide the backscattered electron image into several superpixel blocks; determine the component distribution fragmentation factor of the backscattered electron image based on the area and grayscale of the superpixel blocks with similar grayscale, as well as their positional distribution.

[0030] It should be noted that in this embodiment, adaptive histogram equalization (CLAHE) is used to enhance the backscattered electron image, significantly improving its contrast and detail resolution. In adaptive histogram equalization, the size of the block window (also known as the local region size, block size, or neighborhood size) can be adaptively adjusted to suit different scales of minerals, pores, and structural features in the rock sample, thus improving the image enhancement effect. Specifically, when the block window is small, the local detail enhancement effect is significant, highlighting minute mineral boundaries, micropores, and fine structures, resulting in a noticeable improvement in local contrast. However, this may produce block effects, artifacts, or over-enhancement, leading to image distortion. When the block window is large, the image enhancement is smoother, with fewer block effects and artifacts, resulting in a more natural overall visual effect. However, the ability to enhance local microstructures is weakened, detail resolution decreases, and the improvement in local contrast is limited. Adaptive histogram equalization is a well-known technique, and its specific method will not be described here.

[0031] It should be further explained that backscattered electron imaging (BSE) focuses on composition and phase distribution. Its core function is atomic number-based contrast imaging, where image brightness directly reflects the local average atomic number of the sample: the higher the atomic number, the brighter the image; the lower the atomic number, the darker the image. For example, pyrite with a high atomic number appears bright white in a BSE image, while quartz with a low atomic number appears dark gray, which can visually demonstrate the distribution characteristics, symbiotic relationships, and alteration zone range of minerals.

[0032] First, a superpixel segmentation algorithm is used to divide the backscattered electron image into several superpixel blocks.

[0033] Among them, the superpixel segmentation algorithm is a well-known technology, and the specific method will not be introduced here.

[0034] Within each superpixel block, the grayscale values ​​of the pixels are similar, meaning that each superpixel block corresponds to a microstructural unit or grain region for each type of rock composition. Therefore, superpixel blocks with similar grayscale values ​​correspond to a type of rock composition.

[0035] Therefore, the component distribution fragmentation factor of the backscattered electron image can be determined based on the area and gray level of superpixel blocks with similar gray levels, as well as their positional distribution.

[0036] Therefore, the size of the block window can be adaptively determined based on the fragmentation factor of the component distribution to achieve image enhancement that better matches the characteristics of the rock composition. In backscattered electron images, the more fragmented the rock composition distribution and the darker the grayscale, the smaller the block window needs to be to enhance local details.

[0037] Step S003: Determine the local component complexity of each superpixel block based on the area size and grayscale difference of adjacent superpixel blocks; select several component-complex superpixel blocks based on the magnitude of the local component complexity; determine the local component complexity representation factor of the backscattered electron image based on the number and positional distribution of component-complex superpixel blocks and the magnitude of the local component complexity.

[0038] It should be noted that the above describes the overall representation of rock components in backscattered electron images. Furthermore, to more accurately adapt to the compositional variations in local areas, it is necessary to analyze the compositional differences between adjacent superpixel blocks and determine the complexity of local compositions.

[0039] Therefore, the local component complexity of each superpixel block is first determined based on the area size and grayscale difference of adjacent superpixel blocks.

[0040] It should be noted that the higher the complexity of the local composition, the smaller the block window needs to be for adaptive histogram equalization, in order to preserve clear phase boundaries in complex composition regions and improve the enhancement effect of fine structures.

[0041] Then, based on the degree of local component complexity of each superpixel block, several component-complex superpixel blocks are selected from all superpixel blocks.

[0042] Finally, based on the number and location distribution of the complex superpixel blocks, and combined with the magnitude of the local component complexity of each superpixel block, the local component complexity representation factor of the backscattered electron image is determined.

[0043] Step S004: Determine the block adjustment coefficient of the backscattered electron image based on the magnitude of the local component complexity factor and the component distribution fragmentation factor; determine the optimal block window size based on the magnitude of the block adjustment coefficient, perform enhancement operation on the backscattered electron image, and obtain the enhanced backscattered electron image for lithological identification.

[0044] It should be noted that in this embodiment, the component distribution fragmentation factor and the local component complexity are fused to obtain the final block adjustment coefficient, and the block window size is adaptively determined based on the coefficient, so that the block window can adapt to the overall fragmentation distribution characteristics of rock components and match the drastic component changes in local areas, thereby achieving more accurate image enhancement that better meets the needs of lithology identification.

[0045] Therefore, firstly, the block adjustment coefficient of the backscattered electron image is determined based on the magnitude of the local component complexity factor and the component distribution fragmentation factor of the backscattered electron image.

[0046] Then, the optimal block window size is determined based on the block adjustment coefficient of the backscattered electron image.

[0047] Based on the optimal block window size, adaptive histogram equalization is performed on the backscattered electron image to obtain the enhanced backscattered electron image.

[0048] Lithology identification is performed based on secondary electron images and enhanced backscattered electron images.

[0049] It should be noted that lithological identification is a fundamental technical step in geological exploration, oil and gas resource evaluation, and ore analysis, and lithological identification based on scanning electron microscopy is a well-known and routine technique in this field. Specifically: (1) The microscopic morphological features of the rock surface are obtained by acquiring secondary electron images, and mineral phases with different atomic numbers are distinguished by enhanced backscattered electron images; (2) Combine energy dispersive spectroscopy to perform elemental qualitative and semi-quantitative analysis on micro-areas of rocks to determine the main mineral composition and elemental distribution; (3) Based on the rock’s microstructure, mineral morphology, elemental composition and occurrence state, comprehensively determine the rock type, mineral composition, structure and diagenetic characteristics.

[0050] The aforementioned lithological identification process based on scanning electron microscopy is mainly used to achieve observation of rock micromorphology, mineral identification, composition analysis and microstructure characterization, and is a conventional and well-known analytical method in this field.

[0051] Preferably, in some possible implementations of the embodiments of the present invention, the average gray value of all pixels in each superpixel block is recorded as the average gray value of each superpixel block.

[0052] The average gray value of each superpixel block can represent the average atomic number (i.e., core component feature) of that region.

[0053] Using the absolute value of the difference between the average gray values ​​of any two superpixel blocks as the clustering distance, a hierarchical clustering algorithm is used to perform clustering operations on all superpixel blocks in the backscattered electron image, resulting in several clusters.

[0054] Each cluster represents a type of rock composition, and each superpixel within a cluster corresponds to a microstructural unit or particle region of that rock composition. Hierarchical clustering algorithms are well-known techniques, and their specific methods will not be described here.

[0055] The average gray value of each cluster corresponds to the average atomic number of the rock component of that type, and can be used to characterize the phase properties of that rock component.

[0056] The mean area of ​​all superpixel blocks in the backscattered electron image is obtained and denoted as the relative area threshold.

[0057] For all superpixel blocks in each cluster, the ratio of the number of superpixel blocks with an area smaller than the relative area threshold to the total number of superpixel blocks is denoted as the proportion of tiny particles, and the average area of ​​all superpixel blocks is denoted as the area performance. The mean of the shortest distances between all pairs of superpixel blocks is denoted as the particle distribution discreteness. Normalize the area representation of the complement value. Normalized value of particle distribution discreteness The mean value is denoted as the quantity adjustment coefficient. The product of the quantity adjustment coefficient and the proportion of small particles is denoted as the fragmented distribution performance value of each cluster.

[0058] If each cluster contains only one superpixel block, or if each cluster contains no superpixel blocks with an area smaller than the relative area threshold, then the fragmented distribution representation value of each cluster is directly set to 0. In this embodiment, the shortest distance between two superpixel blocks is the Euclidean distance between the geometric centroids of the two superpixel blocks.

[0059] It should be noted that in this embodiment, the maximum value of the shortest distances between all pairs of superpixel blocks in the backscattered electron image is obtained and denoted as the maximum discrete distance. The maximum area among all superpixel blocks in the backscattered electron image is obtained and denoted as the maximum superpixel block area. Therefore, with As the complement normalized value representing the area, As a normalized value for the dispersion of particle distribution, the quantity adjustment factor is a dimensionless data value between 0 and 1.

[0060] It should be further noted that the smaller the area of ​​the superpixel blocks in a cluster and the greater the distance between the superpixel blocks (i.e., the smaller the area representation and the greater the dispersion of particle distribution), the smaller the microstructural units or particle regions of this type of rock component, and the more dispersed their distribution. At the same time, the greater the proportion of tiny particles, the more pronounced the fine fragmentation of this type of rock component.

[0061] Preferably, in some possible implementations of the embodiments of the present invention, the mean of the average grayscale values ​​of all superpixel blocks in each cluster is obtained and recorded as the average grayscale value of each cluster.

[0062] The average gray value of each cluster complement normalized value As a weight, the fractional distribution performance values ​​of all clusters are weighted and averaged to determine the fractional distribution factor of the backscattered electron image.

[0063] It should be noted that, in this embodiment, the grayscale range of the backscattered electron image is 0 to 255, therefore... As The complement normalized value is used. When the fine distribution of each rock component is more obvious, a smaller segment window should be selected to precisely enhance the small particles. However, different components have different "fine distribution thresholds". For example, even if pyrite particles are small, their atomic number contrast is high, so the enhancement intensity can be slightly lower; quartz particles are fine and have low contrast, so a slightly higher enhancement intensity is required. Therefore, in this embodiment, the complement normalized value of the average gray value of each cluster is used as the weight. The fine distribution performance values ​​of all clusters are weighted, summed, and then divided by the total number of clusters. That is, the fine distribution performance values ​​of all clusters are weighted and averaged to determine the component distribution fine distribution factor of the backscattered electron image. In the backscattered electron image, when the rock component distribution is fine and the gray level is dark, a smaller segment window is needed to enhance local details.

[0064] Preferably, in some possible implementations of the embodiments of the present invention, any superpixel block in the backscattered electron image is denoted as the target superpixel block.

[0065] All superpixel blocks adjacent to the target superpixel block are designated as reference superpixel blocks.

[0066] Calculate the absolute value of the difference between the average gray value of the target superpixel block and the average gray value of each reference superpixel block. The mean of the absolute values ​​of the differences between the average gray value of the target superpixel block and the average gray values ​​of all reference superpixel blocks is denoted as the drastic change of local components.

[0067] The target superpixel block and all reference superpixel blocks are collectively referred to as the neighborhood superpixel block.

[0068] Sum of the areas of all neighboring superpixels The complement normalized value is denoted as the local component density.

[0069] In this embodiment, As The complement normalized value, Let be the local area of ​​the target superpixel block. It represents the maximum local area of ​​all superpixel blocks in the backscattered electron image.

[0070] The number of clusters to which all neighboring superpixel blocks belong. The normalized value is denoted as the local component type quantization value.

[0071] In this embodiment, The ratio of the total number of clusters to the total number of clusters, as The normalized value.

[0072] The mean of the local component density and the local component type quantization value is denoted as the local component feature adjustment value. The product of the local component feature adjustment value and the local component drastic change is denoted as the local component complexity of the target superpixel block.

[0073] It should be noted that: the smaller the local region occupied by the target superpixel block and its neighboring superpixel blocks, and the more component types they correspond to (i.e., the greater the local component density and the higher the quantization value of the local component type), the denser the boundaries of a large number of different rock components within the smaller local region. Simultaneously, the greater the grayscale difference between the target superpixel block and its neighboring superpixel blocks (i.e., the more drastic the local component changes), the more drastic the component changes, and the higher the complexity of the local components corresponding to the target superpixel block.

[0074] Preferably, in some possible implementations of the embodiments of the present invention, in a backscattered electron image, a superpixel block whose normalized value of local component complexity is greater than a preset component complexity threshold is denoted as a component complex superpixel block.

[0075] In this embodiment, the min-max normalization method is used to normalize the local component complexity of all superpixel blocks in the backscattered electron image to a range of 0 to 1. The min-max normalization method is a well-known technique, and its specific method will not be described here. A preset component complexity threshold of 0.7 is used as an example for explanation.

[0076] It should be noted that, because the min-max normalization method is used, there will definitely be at least one value greater than 0.7 among the normalized values ​​of the local composition complexity of all superpixel blocks in the backscattered electron image. If the local composition complexity of all superpixel blocks in the backscattered electron image is consistent, it means that the composition of the entire rock sample is absolutely homogeneous, which is almost impossible. Therefore, an alarm should be triggered to indicate that an anomaly has been detected, and the operation should be stopped and the process guided for troubleshooting.

[0077] Preferably, in some possible implementations of the embodiments of the present invention, the ratio of the number of all component-complex superpixel blocks to the total number of all superpixel blocks is obtained and denoted as the proportion of component-complex micro-regions.

[0078] Obtain the mean of the shortest distances between all pairs of complex superpixel blocks. normalized value , which is denoted as the discreteness of the distribution of complex, small regions.

[0079] in, The maximum value among the shortest distances between all pairs of superpixel blocks in the backscattered electron image. If there is only one superpixel block with complex composition, then the dispersion of the distribution of the small complex region is directly set to 0.

[0080] The average of the proportion of complex micro-regions and the dispersion of their distribution is denoted as the component complexity correction coefficient. The product of the average of the normalized values ​​of the local component complexity of all superpixel blocks and the component complexity correction coefficient is denoted as the local component complexity representation factor of the backscattered electron image.

[0081] It should be noted that in backscattered electron images, when numerous small local regions with high compositional complexity are distributed throughout the image, adaptive histogram equalization uses smaller block windows to enhance the local contrast of these small regions and highlight the details of local mineral composition. The local compositional complexity representation factor in backscattered electron images is a dimensionless data value between 0 and 1.

[0082] Preferably, in some possible implementations of the embodiments of the present invention, the product of the local component complexity representation factor of the backscattered electron image and the component distribution fragmentation factor of the backscattered electron image is denoted as the block adjustment coefficient of the backscattered electron image.

[0083] Preferably, in some possible implementations of the embodiments of the present invention, if the normalized value of the block adjustment coefficient of the backscattered electron image is less than or equal to a preset first threshold, then the optimal block window size is set to a preset third window.

[0084] If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than the preset first threshold and less than the preset second threshold, then the optimal block window size is set to the preset second window.

[0085] If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than or equal to the preset second threshold, then the optimal block window size is set to the preset first window size.

[0086] The preset first threshold is 0.3, the preset second threshold is 0.7, and the preset first window is [missing information]. The default second window is The default third window is This will be described using an example. In this embodiment, the block adjustment coefficients of multiple historical backscattered electron images are obtained in the manner described above. The min-max normalization method is used to normalize the block adjustment coefficients of all backscattered electron images to between 0 and 1. The multiple historical backscattered electron images correspond to similar rock samples.

[0087] Secondly, please refer to Figure 2 This illustrates a lithology identification system based on scanning electron microscopy provided by an embodiment of the present invention, which includes the following modules: A scanning electron microscope (SEM) image acquisition module for rock samples is used to acquire backscattered electron images for lithology identification. The rock composition distribution fragmentation analysis module is used to divide the backscattered electron image into several superpixel blocks; based on the area and gray level of the superpixel blocks with similar gray levels, as well as their positional distribution, the composition distribution fragmentation factor of the backscattered electron image is determined. The rock composition local complexity analysis module is used to determine the local composition complexity of each superpixel block based on the area size and grayscale difference of adjacent superpixel blocks; to select several compositionally complex superpixel blocks based on the magnitude of the local composition complexity; and to determine the local composition complexity representation factor of the backscattered electron image based on the number and location distribution of compositionally complex superpixel blocks and the magnitude of the local composition complexity. The lithology identification module is used to determine the block adjustment coefficient of the backscattered electron image based on the magnitude of the local component complexity representation factor and the component distribution fragmentation factor; based on the magnitude of the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithology identification.

[0088] Thirdly, the present invention also provides a lithological identification device based on scanning electron microscopy, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned lithological identification method based on scanning electron microscopy.

[0089] This invention is now complete.

[0090] In summary, in this embodiment of the invention, the backscattered electron image is divided into several superpixel blocks. Based on the area, grayscale value, and positional distribution of superpixel blocks with similar grayscale, a component distribution fragmentation factor of the backscattered electron image is determined. Based on the area and grayscale differences of adjacent superpixel blocks, a local component complexity factor of the backscattered electron image is determined. This determines the block adjustment coefficient of the backscattered electron image, used to obtain the optimal block window size. The backscattered electron image is then enhanced to obtain the enhanced backscattered electron image for lithological identification. This invention, through adaptive block window size, efficiently enhances the backscattered electron image, ensuring the accuracy of lithological identification.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A lithological identification method based on scanning electron microscopy, characterized in that, The method includes the following steps: Acquire backscattered electron images for lithological identification; The backscattered electron image is divided into several superpixel blocks; the mean gray value of all pixels in each superpixel block is recorded as the average gray value of each superpixel block; the absolute value of the difference between the average gray values ​​of any two superpixel blocks is used as the clustering distance; clustering is performed on all superpixel blocks to obtain several clusters; the component distribution fragmentation factor of the backscattered electron image is determined based on the area, gray value, and positional distribution of the superpixel blocks in the clusters. Based on the area size and grayscale difference of adjacent superpixel blocks, the local component complexity of each superpixel block is determined; based on the magnitude of the local component complexity, several component-complex superpixel blocks are selected; based on the number and positional distribution of component-complex superpixel blocks, combined with the magnitude of the local component complexity, the local component complexity representation factor of the backscattered electron image is determined. Based on the magnitudes of the local component complexity factor and the component fragmentation factor of the backscattered electron image, the block adjustment coefficient of the backscattered electron image is determined; based on the magnitude of the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithological identification.

2. The lithological identification method based on scanning electron microscopy according to claim 1, characterized in that, The specific steps for determining the component distribution fragmentation factor of the backscattered electron image are as follows: The average area of ​​all superpixel blocks is obtained and denoted as the relative area threshold. For all superpixel blocks in each cluster, the ratio of the number of superpixel blocks with an area smaller than the relative area threshold to the total number of superpixel blocks is denoted as the proportion of tiny particles. The mean of the areas of all superpixel blocks is denoted as the area expressivity. The mean of the shortest distances between all pairs of superpixel blocks is denoted as the particle distribution dispersion. The mean of the complement normalized value of the area expressivity and the normalized value of the particle distribution dispersion is denoted as the quantity adjustment coefficient. The product of the quantity adjustment coefficient and the proportion of tiny particles is denoted as the fragmented distribution expressivity of each cluster. The component distribution fragmentation factor of the backscattered electron image is determined based on the average gray value of the superpixel block in each cluster and the fragmentation distribution performance value.

3. The lithological identification method based on scanning electron microscopy according to claim 2, characterized in that, The specific steps for determining the component distribution fragmentation factor of the backscattered electron image based on the average gray value of the superpixel block in each cluster and the fragmentation distribution performance value are as follows: The mean of the average gray values ​​of all superpixel blocks in each cluster is obtained and recorded as the average gray value of each cluster. The component distribution fragmentation factor of the backscattered electron image is determined by weighting the weighted average of the fragmented distribution values ​​of all clusters using the complement normalized value of the average gray value of each cluster.

4. The lithological identification method based on scanning electron microscopy according to claim 2, characterized in that, The specific steps involved in determining the local component complexity of each superpixel block are as follows: Let any superpixel block be denoted as the target superpixel block; All superpixel blocks adjacent to the target superpixel block are designated as reference superpixel blocks; The mean of the absolute values ​​of the differences between the average gray value of the target superpixel block and the average gray value of all reference superpixel blocks is denoted as the drastic change of local components. The target superpixel block and all reference superpixel blocks are collectively referred to as the neighborhood superpixel block; The normalized value of the complement of the sum of the areas of all neighboring superpixel blocks is denoted as the local component density. The normalized value of the number of cluster types to which all neighboring superpixel blocks belong is denoted as the local component type quantization value; The mean of the local component density and the local component type quantization value is denoted as the local component feature adjustment value. The product of the local component feature adjustment value and the local component drastic change is denoted as the local component complexity of the target superpixel block.

5. The lithological identification method based on scanning electron microscopy according to claim 1, characterized in that, The specific steps involved in selecting several superpixel blocks with complex composition are as follows: Superpixel blocks whose normalized value of local component complexity is greater than a preset component complexity threshold are denoted as component complex superpixel blocks.

6. The lithological identification method based on scanning electron microscopy according to claim 1, characterized in that, The specific steps for determining the local component complexity representation factor of the backscattered electron image are as follows: Obtain the ratio of the number of all complex superpixel blocks to the total number of superpixel blocks, and denote it as the proportion of complex micro-regions. The normalized value of the mean of the shortest distances between all pairs of superpixel blocks with complex composition is denoted as the discreteness of the distribution of small regions with complex composition. The average of the proportion of complex micro-regions and the dispersion of their distribution is denoted as the component complexity correction coefficient. The product of the average of the normalized values ​​of the local component complexity of all superpixel blocks and the component complexity correction coefficient is denoted as the local component complexity representation factor of the backscattered electron image.

7. The lithological identification method based on scanning electron microscopy according to claim 1, characterized in that, The specific steps for determining the block adjustment coefficients of the backscattered electron image are as follows: The product of the factor representing the complexity of local components and the factor representing the fragmentation of component distribution in a backscattered electron image is denoted as the block adjustment coefficient of the backscattered electron image.

8. The lithological identification method based on scanning electron microscopy according to claim 1, characterized in that, The specific steps involved in determining the optimal block window size are as follows: If the normalized value of the block adjustment coefficient of the backscattered electron image is less than or equal to the preset first threshold, then the optimal block window size is set to the preset third window. If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than the preset first threshold and less than the preset second threshold, then the optimal block window size is set to the preset second window. If the normalized value of the block adjustment coefficient of the backscattered electron image is greater than or equal to the preset second threshold, then the optimal block window size is set to the preset first window size.

9. A lithological identification system based on scanning electron microscopy, employing the lithological identification method based on scanning electron microscopy as described in any one of claims 1-8, characterized in that, The system includes the following modules: A scanning electron microscope (SEM) image acquisition module for rock samples is used to acquire backscattered electron images for lithology identification. The rock composition distribution fragmentation analysis module is used to divide the backscattered electron image into several superpixel blocks; the mean gray value of all pixels in each superpixel block is recorded as the average gray value of each superpixel block; the absolute value of the difference between the average gray values ​​of any two superpixel blocks is used as the clustering distance; clustering operation is performed on all superpixel blocks to obtain several clusters; the composition distribution fragmentation factor of the backscattered electron image is determined based on the area, gray value, and positional distribution of the superpixel blocks in the clusters. The rock composition local complexity analysis module is used to determine the local composition complexity of each superpixel block based on the area size and grayscale difference of adjacent superpixel blocks; to select several compositionally complex superpixel blocks based on the magnitude of the local composition complexity; and to determine the local composition complexity representation factor of the backscattered electron image based on the number and location distribution of compositionally complex superpixel blocks and the magnitude of the local composition complexity. The lithology identification module is used to determine the block adjustment coefficient of the backscattered electron image based on the magnitude of the local component complexity representation factor and the component distribution fragmentation factor; based on the magnitude of the block adjustment coefficient, the optimal block window size is determined, and the backscattered electron image is enhanced to obtain the enhanced backscattered electron image for lithology identification.

10. A lithology identification device based on scanning electron microscopy, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the lithology identification method based on scanning electron microscopy as described in any one of claims 1-8.

Images