Slip zone soil physical property identification system and method based on SEM image

Abstract

The invention provides a sliding zone soil physical property recognition system and method based on an SEM image, and relates to the technical field of sliding zone soil physical property recognition, and the method comprises the steps: obtaining an initial SEM image corresponding to to-be-recognized sliding zone soil; carrying out binarization processing on the initial SEM image to obtain a binarization image corresponding to the to-be-identified slippery soil; obtaining each sub-region in the binarized image to obtain a sub-region list A; obtaining the number of pixels corresponding to each sub-region in the A to obtain a pixel number list B corresponding to the A; clustering the number of pixels in the A by using a preset clustering algorithm to obtain a cluster list C; according to the C, determining the actual soil particle size corresponding to each cluster; by means of the method, the actual soil particle size can be accurately determined; the method does not need to be carried out in indoor laboratory equipment, the determination period is short, and the operation is simple.

Description

Technical Field

[0001] The invention relates to the technical field of physical property identification of sliding zone soil, and in particular to a system and method for identifying physical property of sliding zone soil based on SEM images. Background Art

[0002] my country has complex geological conditions, diverse topography and landforms, and strong geological transformation. As a key geological element of landslides, the structure and strength characteristics of sliding zones play a decisive role in the stability of landslides. The stability of landslides often depends on the stress state and strength changes of sliding zones. The changes in sliding zone strength depend on the type of soil, microstructural characteristics, material composition, and occurrence environment. The properties of sliding zone soil that are different from the surrounding rock and soil are mainly manifested in different stress and deformation histories, different material compositions, and different structural characteristics. Particle grading and particle mineral composition are very important indicators for analyzing the shear strength of sliding zone soil.

[0003] At present, scanning electron microscope (SEM) is commonly used to analyze the microstructure of sliding belt soil, including micropores, mineral orientation, scratches, dendritic structure, network structure and microorganisms; X-ray diffractometer (XDR) combined with X-ray fluorescence spectroscopy (XRF) is used to analyze the types and contents of minerals in sliding belt soil; SEM images are used to obtain the microscopic pore morphology ratio, fractal dimension of pores and particles, particle roundness, and particle discrete index; nuclear magnetic resonance (NMR) technology is used to test and analyze the internal pore structure of the soil. At present, the work of obtaining the particle grading and particle mineral composition of sliding belt soil is mostly carried out in indoor laboratory equipment, with a long measurement cycle and complex computer operation. Summary of the invention

[0004] In view of the above technical problems, the technical solution adopted by the present invention is: According to a first aspect of the present application, a method for identifying physical properties of sliding zone soil based on SEM images is provided, the method comprising the following steps: S100, obtaining an initial SEM image corresponding to the sliding zone soil to be identified; S200, binarizing the initial SEM image to obtain a binarized image corresponding to the sliding zone soil to be identified; wherein the binarized image includes a plurality of sub-regions, each sub-region corresponding to a soil particle; S300, obtaining each sub-region in the binary image to obtain a sub-region list A=(A1, A2, ..., A i , …, A n ), i=1, 2,...,n; among them, A i is the i-th sub-region in the obtained binary image, and n is the number of sub-regions in the binary image; S400, obtaining the number of pixels corresponding to each sub-region in A, so as to obtain a list of the number of pixels corresponding to A, B=(B1, B2, ..., B i , …, B n ); where B i A i The corresponding number of pixels; S500, clustering the number of pixels in A using a preset clustering algorithm to obtain a cluster list C = (C1, C2, ..., C j , …, C m ), j = 1, 2, …, m; where C j is the jth cluster obtained by clustering the number of pixels in A, and m is the number of clusters obtained by clustering the number of pixels in A; S600: According to C, the actual soil particle size corresponding to each cluster is determined.

[0005] Further, step S600 includes the following steps: S610, get C j The minimum circle of the sub-area corresponding to each number of pixels in C j The corresponding minimum circle list D j =(D j，1 , D j，2 , …, D j，p , …, D j，f（j） ), p = 1, 2, …, f(j); where D j，p C j The smallest circle of the sub-region corresponding to the p-th number of pixels in C, f(j) is j The number of pixels in the S620, obtain D j The diameter of each smallest circle in the equation is j Corresponding diameter list E j =(E j，1 , E j，2 ,…,E j，p ,…,E j，f（j） ); where E j，p D j，p The diameter of S630, Get E j Fluctuation rate of median diameter η = (1 / f(j)) × ∑ f（j） p=1 (E j，p -((1 / f(j))×∑ f（j） p= 1E j，p )) 2 ; S640, if η≤η', determine C jThe corresponding actual soil particle size W j =((1 / f(j))×∑ f（j） p=1 E j，p ×α; where η' is the preset diameter fluctuation rate threshold; α is the conversion coefficient between the preset pixel diameter and the actual soil particle size.

[0006] Furthermore, after step S640, the method further includes the following steps: S650, if η>η', obtain E j Each diameter and E j The diameter difference corresponding to the average diameter is obtained to obtain E j Corresponding diameter difference table ΔE j =(ΔE j，1 , ΔE j，2 ,…,ΔE j，p ,…,ΔE j，f（j） ); where ΔE j，p For E j，p With E j The diameter difference corresponding to the mean diameter; S651, traversal ΔE j , if ΔE j，p <ΔE', then E j，p Determine as the target diameter to obtain the target diameter list F = (F1, F2, ..., F q , …, F u ), q=1, 2, …, u; where F q is the qth target diameter determined, u is the number of target diameters determined; ΔE' is the preset diameter difference threshold; S652, according to F, determine C j The corresponding actual soil particle size W j =((1 / u)×∑ u q=1 F q ×α.

[0007] Furthermore, after step S500 and before step S600, the method further includes the following steps: S510, setting the same color for each sub-region corresponding to each cluster; wherein different colors are set for any two clusters.

[0008] Furthermore, after step S600, the method further includes the following steps: S700, obtaining the number of pixels corresponding to each cluster; S710, determining the percentage of the number of pixels corresponding to each cluster to the total number of pixels according to the number of pixels corresponding to each cluster and the total number of pixels corresponding to the binary image, so as to determine the particle size distribution range of the sliding zone soil to be identified.

[0009] Furthermore, A i Obtained through the preset feature extraction model.

[0010] Furthermore, the preset clustering algorithm includes a DBSCANA clustering algorithm.

[0011] According to another aspect of the present application, a system for identifying the physical properties of sliding zone soil based on SEM images is also provided, the system comprising: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the method for identifying the physical properties of sliding zone soil based on SEM images as described in the first aspect.

[0012] The present invention has at least the following beneficial effects: The method for identifying the physical properties of sliding zone soil based on SEM images of the present invention comprises the following steps: obtaining an initial SEM image corresponding to the sliding zone soil to be identified; binarizing the initial SEM image to obtain a binarized image corresponding to the sliding zone soil to be identified; obtaining each sub-region in the binarized image to obtain a sub-region list A; obtaining the number of pixels corresponding to each sub-region in A to obtain a pixel number list B corresponding to A; clustering the number of pixels in A using a preset clustering algorithm to obtain a cluster list C; determining the actual soil particle size corresponding to each cluster based on C; through the method of the present invention, the actual soil particle size can be determined more accurately; and the method of the present invention does not need to be performed in indoor laboratory equipment, has a short measurement cycle, and is simple to operate. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 A flow chart of a method for identifying physical properties of sliding zone soil based on SEM images provided in an embodiment of the present invention; Figure 2 A schematic diagram of an initial SEM image provided by an embodiment of the present invention; Figure 3 A schematic diagram of a binarized image provided by an embodiment of the present invention; Figure 4A schematic diagram of setting different colors for soil particles provided by an embodiment of the present invention; Figure 5 A flow chart of the intelligent identification technology of the physical properties of sliding zone soil provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] It should be noted that, based on the present disclosure, those skilled in the art should understand that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement a device and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0017] The following will refer to Figure 1 The flowchart of the method for identifying the physical properties of sliding zone soil based on SEM images is shown, which introduces a method for identifying the physical properties of sliding zone soil based on SEM images.

[0018] The method for identifying the physical properties of sliding zone soil based on SEM images may include the following steps: S100, obtaining an initial SEM image corresponding to the sliding zone soil to be identified.

[0019] In this embodiment, the sliding zone soil to be identified can be obtained, and then the sliding zone soil to be identified is photographed by SEM. In order to increase the image area, the image taken by SEM can be stitched using a stitching technology to obtain an electron microscope image of a larger area.

[0020] S200, binarizing the initial SEM image to obtain a binarized image corresponding to the sliding zone soil to be identified; wherein the binarized image includes a plurality of sub-regions, each sub-region corresponding to a soil particle.

[0021] In this embodiment, a preset digital image analysis system is used to analyze and process the SEM image to obtain the microstructure parameters of the sample. The image processing mainly includes removing redundant information and binarizing the image. The initial SEM image is as follows: Figure 2 As shown in the figure, after being processed by the digital image analysis system, a binary image with clear boundaries is obtained, such as Figure 3As shown; the sand particles are white and the pores between the particles are black.

[0022] After identifying the binary image, the geometric parameters of the particles and pores in the image can also be counted, including microstructural parameters such as the number of soil particles, average particle area, average particle circumference, equivalent particle diameter, particle shape coefficient, particle shape ratio and fractal dimension.

[0023] S300, obtaining each sub-region in the binary image to obtain a sub-region list A=(A1, A2, ..., A i , …, A n ), i=1, 2,...,n; among them, A i is the i-th sub-region in the obtained binary image, and n is the number of sub-regions in the binary image.

[0024] In this embodiment, the image corresponding to each soil particle in the binary image is a sub-region, and each sub-region can be acquired by image recognition to obtain A.

[0025] In this embodiment, a digital image analysis system is preset, which is designed based on an improved algorithm model of a convolutional neural network.

[0026] 1. Use the global-local Transformer module. The specific steps are as follows: (1) Global feature extraction. The Transformer self-attention mechanism is used to capture the global dependencies in the image. The self-attention calculation is shown as follows: ; Where Q, K, V are the query, key, and value matrices respectively, and dk is the dimension of the key vector. This step allows the model to capture features across the entire image.

[0027] (2) Local feature extraction. Traditional CNN is used to capture local features, such as edges, corners, etc. The CNN layer usually includes multiple convolution and activation functions, which can be expressed as follows: ; Among them, W is the weight of the convolution kernel, b is the bias term, * represents the convolution operation, and Activation is the activation function, such as the linear rectification function.

[0028] (3) Feature fusion. This module combines global features and local features to form a comprehensive feature representation through feature fusion technology.

[0029] (4) Parameter setting. The key parameters of the model include the number of heads in the Transformer layer, the dimension of each head, and the convolution kernel size, number of layers, and number of filters in the CNN layer.

[0030] 2. Convolutional layer encoder. The convolutional layer encoder extracts features from the image by applying a series of convolution operations.

[0031] 3. Feature refinement module. The core function of the feature refinement module is to improve the robustness of feature representation. This module mainly achieves feature refinement by enhancing important features and suppressing unimportant features. In this module, the convolution operation is used to first extract deeper features from the input feature map, and an attention map is generated through the Sigmoid activation function. This map corresponds to the spatial importance of each feature. By applying this attention map to the input feature map, the model can focus more on important features, thereby improving the overall recognition performance. It is particularly effective when processing remote sensing images with complex backgrounds and details, and can help the model identify the most critical parts for the final task from many features.

[0032] 4. Model architecture. The model architecture design includes multiple layers such as input layer, multiple convolutional layers, pooling layer, fully connected layer, output layer, etc. Each layer is responsible for different types of feature extraction and processing. In the model architecture, each convolutional layer consists of several convolution operations and activation functions. This architecture allows the model to extract complex spatial and spectral features from remote sensing images and enhance the model's discrimination ability through layer-by-layer abstraction.

[0033] 5. Loss function. The loss function design is crucial for model training. The loss function defines the difference between the model output and the true value, and the model training goal is to minimize the difference.

[0034] 6. Recognition output. Use a pre-divided test set to evaluate the trained model. The test set contains remote sensing images that have never appeared in the training process, covering various types of land objects and different acquisition conditions to ensure the fairness and comprehensiveness of the evaluation results. During the test, each image is input into the model, and the recognition of the model output response is recorded. Secondly, after the model outputs the result, the result is compared with the actual fifth category to calculate key performance indicators such as accuracy, recall, and F1 score, so as to fully reflect the model's ability in remote sensing image recognition. Among them, the accuracy rate measures the correct proportion of the model's predictions, the recall rate indicates the correct proportion of the model's successful identification, and the F1 score is the harmonic average of the accuracy and recall rates, which is used to balance the relationship between the two. Finally, compared with the traditional machine algorithm, the technical advantages of this model are verified.

[0035] 7. Model verification. The application process of the rapid detection technology of particle gradation of sliding zone soil using image recognition is as follows: 1) Perform a large amount of image recognition and screening work, matching one by one; 2) Organize training samples. The statistical variables obtained from the image recognition results are the input data. The statistical variables calculated from the grading curve obtained by particle grading screening are the output data. The number of samples should be no less than 100 groups to ensure the learning efficiency and accurate implicit relationship of the artificial neural network. After the functional relationship between the two model parameters is obtained using the neural network algorithm, the grading test of the sliding zone soil particles can be carried out.

[0036] S400, obtaining the number of pixels corresponding to each sub-region in A, so as to obtain a list of the number of pixels corresponding to A, B=(B1, B2, ..., B i , …, B n ); where B i A i The corresponding number of pixels.

[0037] In this embodiment, each sub-region corresponds to a sub-region image, and the sub-region image includes a number of pixels. Each pixel in each sub-region can be acquired, thereby obtaining B.

[0038] S500, clustering the number of pixels in A using a preset clustering algorithm to obtain a cluster list C = (C1, C2, ..., C j , …, C m ), j = 1, 2, …, m; where C j is the jth cluster obtained by clustering the number of pixels in A, and m is the number of clusters obtained by clustering the number of pixels in A.

[0039] In this embodiment, the preset clustering algorithm may be a DBSCAN clustering algorithm, which clusters the number of pixels in A and can cluster sub-regions with similar numbers into clusters. It can be understood that when the number of pixels is similar, the size of the sub-regions is generally also relatively similar.

[0040] Furthermore, after step S500 and before step S600, the method further includes the following steps: S510, setting the same color for each sub-region corresponding to each cluster; wherein different colors are set for any two clusters.

[0041] In this embodiment, a digital image analysis system is used to perform microscopic structural analysis on the photographs to obtain Figure 4The sample parameter calculation diagram is shown. Each soil particle is represented by a different color, and the microstructure parameters of each soil particle are also obtained. These include the area of ​​each soil particle and the equivalent diameter of the soil particle. The area of ​​the soil particle is the cross-sectional area of ​​each particle in the figure, expressed in pixel size; the equivalent diameter of the soil particle is the diameter of a circle with the same area as the soil particle, which is used to reflect the size of the soil particle.

[0042] S600: According to C, the actual soil particle size corresponding to each cluster is determined.

[0043] Further, step S600 includes the following steps: S610, get C j The minimum circle of the sub-area corresponding to each number of pixels in C j The corresponding minimum circle list D j =(D j，1 , D j，2 , …, D j，p , …, D j，f（j） ), p = 1, 2, …, f(j); where D j，p C j The smallest circle of the sub-region corresponding to the p-th number of pixels in C, f(j) is j The number of pixels in .

[0044] S620, obtain D j The diameter of each smallest circle in the equation is j Corresponding diameter list E j =(E j，1 , E j，2 , …, E j，p ,…,E j，f（j） ); where E j，p D j，p diameter.

[0045] S630, Get E j Fluctuation rate of median diameter η = (1 / f(j)) × ∑ f（j） p=1 (E j，p -((1 / f(j))×∑ f（j） p= 1E j，p )) 2 .

[0046] S640, if η≤η', determine C j The corresponding actual soil particle size W j =((1 / f(j))×∑ f（j） p=1 E j，p×α; where η' is the preset diameter fluctuation rate threshold; α is the conversion coefficient between the preset pixel diameter and the actual soil particle size.

[0047] In this embodiment, if η≤η', it means D j If the diameter of each smallest circle in is relatively balanced, then C can be directly determined j The corresponding actual soil particle size W j ; In determining C j The corresponding actual soil particle size is based on a large number of minimum circle diameters, which can avoid the situation where the individual minimum circle diameter changes suddenly and causes erroneous results.

[0048] Furthermore, after step S640, the method further includes the following steps: S650, if η>η', obtain E j Each diameter and E j The diameter difference corresponding to the average diameter is obtained to obtain E j Corresponding diameter difference table ΔE j =(ΔE j，1 , ΔE j，2 ,…,ΔE j，p ,…,ΔE j，f（j） ); where ΔE j，p For E j，p With E j The diameter difference corresponds to the mean diameter.

[0049] S651, traversal ΔE j , if ΔE j，p <ΔE', then E j，p Determine as the target diameter to obtain the target diameter list F = (F1, F2, ..., F q , …, F u ), q=1, 2, …, u; where F q is the qth target diameter determined, u is the number of the determined target diameters; ΔE' is the preset diameter difference threshold.

[0050] S652, according to F, determine C j The corresponding actual soil particle size W j =((1 / u)×∑ u q=1 F q ×α.

[0051] In this embodiment, there may be situations where soil particles are not round, for example, individual soil particles are long strips. In this case, the diameter of the smallest circle of the long strips of soil particles will be larger, which will have a greater impact on the measurement results. The above steps can avoid the occurrence of the above situation, making the determined soil particle size more consistent with the actual situation.

[0052] Furthermore, after step S600, the method further includes the following steps: S700: Obtain the number of pixels corresponding to each cluster.

[0053] S710, determining the percentage of the number of pixels corresponding to each cluster to the total number of pixels according to the number of pixels corresponding to each cluster and the total number of pixels corresponding to the binary image, so as to determine the particle size distribution range of the sliding zone soil to be identified.

[0054] In this embodiment, the area of ​​each soil particle and the equivalent diameter of the soil particle can be obtained. The area of ​​the soil particle is the cross-sectional area of ​​each particle in the figure, expressed in pixel size; the equivalent diameter of the soil particle is the diameter of a circle of the same area as the soil particle, which is used to reflect the size of the soil particles.

[0055] The "equivalent diameter of soil particles" and "area of ​​soil particles" obtained based on digital image processing technology are respectively equivalent to the actual soil particle size and soil particle mass. At the same time, the pixels are converted into the actual size of soil particles according to the physical resolution of the SEM image, thereby obtaining the conversion relationship between the "equivalent diameter of soil particles" and the actual diameter of soil particles. Then, the percentage of the "area of ​​soil particles" of each particle group to the sum of all the "areas of soil particles" in the image is calculated by statistical methods, and finally the particle size distribution range of the sample is determined. The accuracy of this method is related to the recognition accuracy of the digital image analysis system. Therefore, controlling the minimum particle size that can be recognized by the digital image analysis system is a key link; in this embodiment, the technical process for intelligent recognition of the physical properties of sliding belt soil can be referred to. Figure 5 shown.

[0056] The SEM image-based method for identifying the physical properties of sliding zone soil in this embodiment comprises the following steps: obtaining an initial SEM image corresponding to the sliding zone soil to be identified; binarizing the initial SEM image to obtain a binarized image corresponding to the sliding zone soil to be identified; obtaining each sub-region in the binarized image to obtain a sub-region list A; obtaining the number of pixels corresponding to each sub-region in A to obtain a pixel number list B corresponding to A; clustering the number of pixels in A using a preset clustering algorithm to obtain a cluster list C; determining the actual soil particle size corresponding to each cluster based on C; through the method of the present invention, the actual soil particle size can be determined more accurately; and the method of the present invention does not need to be performed in indoor laboratory equipment, has a short measurement cycle, and is simple to operate.

[0057] By combining SEM and image recognition technology, the particle gradation and particle mineral composition of the sliding zone soil are intelligently identified and analyzed. The device is designed to be handheld and portable, which is convenient for field operations. The built-in chip system can obtain the particle gradation and mineral composition of the sliding zone soil in the project area through data set training and image intelligent recognition, thereby improving the efficiency and intelligence of landslide stability analysis in landslide control projects.

[0058] In an exemplary embodiment, a system for identifying the physical properties of sliding zone soil based on SEM images is also provided, characterized in that the system includes: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the method for identifying the physical properties of sliding zone soil based on SEM images as in the above-mentioned embodiment.

[0059] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0060] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are only for illustration, not for limiting the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A method for identifying the physical properties of sliding zone soil based on SEM images, characterized in that: The method comprises the following steps: S100, obtaining an initial SEM image corresponding to the sliding zone soil to be identified; S200, binarizing the initial SEM image to obtain a binarized image corresponding to the sliding zone soil to be identified; wherein the binarized image includes a plurality of sub-regions, each sub-region corresponding to a soil particle; S300, obtaining each sub-region in the binary image to obtain a sub-region list A=(A1, A2, ..., A i , …, A n ), i=1, 2,...,n; among them, A i is the i-th sub-region in the obtained binary image, and n is the number of sub-regions in the binary image; S400, obtaining the number of pixels corresponding to each sub-region in A, so as to obtain a list of the number of pixels corresponding to A, B=(B1, B2, ..., B i , …, B n ); where B i A i The corresponding number of pixels; S500, clustering the number of pixels in A using a preset clustering algorithm to obtain a cluster list C = (C1, C2, ..., C j , …, C m ), j = 1, 2, …, m; where C j is the jth cluster obtained by clustering the number of pixels in A, and m is the number of clusters obtained by clustering the number of pixels in A; S600: According to C, the actual soil particle size corresponding to each cluster is determined.

2. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 1 is characterized in that: Step S600 includes the following steps: S610, get C j The minimum circle of the sub-area corresponding to each number of pixels in C j The corresponding minimum circle list D j =(D j，1 , D j，2 , …, D j，p , …, D j，f（j） ), p = 1, 2, …, f(j); where D j，p C j The smallest circle of the sub-region corresponding to the p-th number of pixels in C, f(j) is j The number of pixels in the S620, obtain D j The diameter of each smallest circle in the equation is j Corresponding diameter list E j =(E j，1 , E j，2 ,…,E j，p ,…,E j，f（j） ); where E j，p D j，p The diameter of S630, Get E j Fluctuation rate of median diameter η = (1 / f(j)) × ∑ f（j） p=1 (E j，p -((1 / f(j))×∑ f（j） p=1 E j，p )) 2 ; S640, if η≤η', determine C j The corresponding actual soil particle size W j =((1 / f(j))×∑ f（j） p=1 E j，p ×α; where η' is the preset diameter fluctuation rate threshold; α is the conversion coefficient between the preset pixel diameter and the actual soil particle size.

3. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 2 is characterized in that: After step S640, the method further includes the following steps: S650, if η>η', obtain E j Each diameter and E j The diameter difference corresponding to the average diameter is obtained to obtain E j Corresponding diameter difference table ΔE j =(ΔE j，1 , ΔE j，2 ,…,ΔE j，p ,…,ΔE j，f（j） ); where ΔE j，p For E j，p With E j The diameter difference corresponding to the mean diameter; S651, traversal ΔE j , if ΔE j，p <ΔE', then E j，p Determine as the target diameter to obtain the target diameter list F = (F1, F2, ..., F q , …, F u ), q=1, 2, …, u; where F q is the qth target diameter determined, u is the number of target diameters determined; ΔE' is the preset diameter difference threshold; S652, according to F, determine C j The corresponding actual soil particle size W j =((1 / u)×∑ u q=1 F q ×α.

4. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 1 is characterized in that: After step S500 and before step S600, the method further includes the following steps: S510, setting the same color for each sub-region corresponding to each cluster; wherein different colors are set for any two clusters.

5. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 1 is characterized in that: After step S600, the method further includes the following steps: S700, obtaining the number of pixels corresponding to each cluster; S710, determining the percentage of the number of pixels corresponding to each cluster to the total number of pixels according to the number of pixels corresponding to each cluster and the total number of pixels corresponding to the binary image, so as to determine the particle size distribution range of the sliding zone soil to be identified.

6. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 1 is characterized in that: A i Obtained through the preset feature extraction model.

7. The method for identifying the physical properties of sliding zone soil based on SEM images according to claim 2 is characterized in that: The preset clustering algorithm includes the DBSCANA clustering algorithm.

8. A system for identifying the physical properties of sliding zone soil based on SEM images, characterized in that: The system includes: a storage medium and a processor; wherein the storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement the method for identifying the physical properties of sliding zone soil based on SEM images as described in any one of claims 1-7.

Images