A shield ring-in geological identification and characterization method based on superpixel segmentation algorithm

Abstract

The application provides a shield ring-in-geology identification and characterization method based on a superpixel segmentation algorithm. A superpixel segmentation algorithm is used to create a layer mask matrix by identifying and positioning each soil layer in the range of the soil layer to be crossed; the soil layer mask matrix is combined according to the arrangement position to obtain a pseudo-color matrix; the corresponding shield tunnel structure ring number is obtained, the ring number is used to segment the inter-ring area of the geological cross-section diagram to form a geological distribution matrix of each ring; the geological distribution matrix of each ring is scanned column by column, and the layer height in each ring is combined to obtain a proportion queue of each ring; according to the shape of the tunnel cross-section, a three-dimensional space matrix is constructed and slice analysis is performed; the distribution area and area proportion of each soil layer from high to low corresponding to the slice are calculated, and the same calculation is performed on the three-dimensional space matrix to complete the distribution characteristic analysis of the soil layer in the shield tunneling ring. Effective support is provided for defining the comprehensive geotechnical physical and mechanical parameters in the tunneling ring, and accurate geological information of the tunneling cross-section is provided for shield construction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of geological identification and characterization in the shield ring, and particularly relates to a geological identification and characterization method in the shield ring based on a superpixel segmentation algorithm. BACKGROUND

[0002] In the process of shield tunnel construction, the operator needs to make dynamic adjustments to the tunneling parameters of the shield machine according to different strata. However, the correlation between the actual geological and rock characteristics in front of the shield and the tunneling parameters of the shield machine is not clear. Construction personnel only identify the relationship between the overall rock and soil physical and mechanical properties in front of the shield and the tunneling parameter setting of the shield machine from a qualitative point of view, and adjust it based on artificial experience. The comprehensive quantitative index of the physical and mechanical properties of the rock and soil of the shield excavation face is missing, making it difficult to use artificial intelligence, machine learning and other methods to deeply explore the deep correlation between geological characteristics and on-board tunneling parameters.

[0003] From the current construction point of view, the main way to obtain the soil layer distribution characteristics in each tunneling ring of the shield is to measure the depth of different soil layers in each ring from the geological exploration report and the CAD longitudinal section construction drawing. However, as the length of the shield tunnel construction increases, the number of tunneling rings corresponding to each tunnel also increases (the number of tunneling rings in some tunnels has exceeded 3000). It takes a lot of manual cost to measure the height of each soil layer in the ring in turn, and the measurement result review cost is also high. Moreover, there are often multiple different soil layers stacked in each ring, and it is urgently needed to quantify the comprehensive rock and soil physical and mechanical properties of each ring. SUMMARY

[0004] To solve the technical problems in the background art, the application provides a geological identification and characterization method in the shield ring based on a superpixel segmentation algorithm. An improved superpixel segmentation algorithm is used to automatically locate the tunnel top and bottom buried depth interface and the distribution position of each soil layer on the geological profile. Based on the interpolation algorithm, the corresponding soil layer thickness of each tunneling ring of the shield is obtained, the in-ring proportion of each soil layer is calculated from the perspective of three-dimensional space, and the accurate geological information required for shield construction is given.

[0005] To achieve the above purpose, the application provides the following technical solution: a geological identification and characterization method in the shield ring based on a superpixel segmentation algorithm, which comprises at least the following steps:

[0006] The geological profile of the shield engineering project is preprocessed to generate a PNG format picture, and the PNG format picture is pixel processed to determine the range of the soil layer penetrated by the shield excavation face.

[0007] The superpixel segmentation algorithm is used to identify and position each soil layer in the soil layer range to be crossed, and a corresponding soil layer mask matrix is created; the soil layer mask matrix is combined according to the arrangement position to obtain a pseudo-color matrix;

[0008] The corresponding number of shield tunnel structure rings is obtained, the number of shield tunnel structure rings is used for inter-ring region segmentation of the geological section graph, and a geological distribution matrix of each ring is formed based on the pseudo-color matrix;

[0009] The geological distribution matrix of each ring is scanned column by column, and the layer height in each ring is combined to obtain a proportion queue of each ring; according to the shape of the tunnel section, a three-dimensional space matrix of the soil layer distribution of each ring is constructed, and the three-dimensional space matrix is analyzed by slicing;

[0010] The distribution area and area proportion of each soil layer from high to low corresponding to the slice are calculated, and the same calculation is performed on the three-dimensional space matrix to complete the distribution characteristic analysis of the soil layer in the shield tunneling ring.

[0011] In further embodiments, the generation process of the PNG format picture is as follows:

[0012] The calibration elements in the geological section graph are deleted, only the layered soil layer information and the design interfaces of the tunnel top boundary and the tunnel bottom boundary are retained to obtain a processed geological section graph;

[0013] The different soil layers of the processed geological section graph are subjected to differential color filling treatment, the tunnel top boundary and the tunnel bottom boundary are subjected to special color filling treatment, and are converted into a PNG format.

[0014] In further embodiments, the explicit way of crossing the soil layer range is as follows:

[0015] A three-dimensional matrix of color images of red, green and blue of the PNG format picture is obtained, the interface positions of the tunnel top boundary and the tunnel bottom boundary are obtained by picking up the fixed RGB (255, 0, 0) value, and the three-dimensional matrix of color images is segmented according to the position of the tunnel top boundary in the three-dimensional matrix. The corresponding values of the three-dimensional matrix of color images outside the tunnel top boundary are filled with 0, and the rows and columns not belonging to the tunnel top boundary, the tunnel bottom boundary and the soil layer in the tunneling section are deleted.

[0016] In further embodiments, the step of identifying and positioning each soil layer is as follows:

[0017] Each pixel between the tunnel top boundary and the tunnel bottom boundary is represented by a five-dimensional feature vector , and K superpixels are preset, the K superpixels correspond to different soil layers;

[0018] The distance between the pixel and each superpixel is calculated ,in, Represents pixels The five-dimensional eigenvectors, It is the first The five-dimensional feature vector of each superpixel ;

[0019] Create a clustering objective function E, and minimize the clustering objective function E to determine the pixels. shortest distance According to the shortest distance Pixels Distribute to the corresponding soil layers;

[0020] Correspondingly, the method for creating the soil layer mask matrix is ​​as follows: a two-dimensional mask matrix is ​​created and stored for the position of each soil layer in the matrix. A fixed and unique number corresponding to the soil layer type is used to represent the position of the current soil layer in the entire image matrix, and all other positions are marked as 0.

[0021] In a further embodiment, the soil layer mask matrices of each soil layer are merged, and each soil layer is identified by a fixed and unique number.

[0022] Each soil layer is assigned a unique color based on its numerical value, and different soil layer locations are colored to complete the construction of a pseudo-color matrix.

[0023] In a further embodiment, the number of rings in the shield tunnel structure is defined as follows: ,right A pseudo-color matrix of a certain size is used for inter-ring region segmentation, where, The row number of the pseudo-color matrix. The column number is the number of columns in the pseudo-color matrix; the segmentation process of the inter-ring region is as follows:

[0024] The average number of columns in the pseudo-color matrix allocated to each ring is calculated using the following formula. and the number of remaining columns :

[0025] ; To round down, mod For modulo operation;

[0026] For the number is ring block, The ring block was determined through the following analysis. The number of columns in the corresponding geological distribution matrix The value of: If ,but ;on the contrary, ;

[0027] Based on the determined number of columns The pseudo-color matrix is continuously and non-overlappingly divided to obtain N geological distribution matrices: .

[0028] In a further embodiment, the obtaining method of the proportion queue of each ring is as follows:

[0029] Each column in the geological distribution matrix of each ring is sequentially counted from top to bottom to form N proportion statistical matrices: , is the soil layer distribution queue of the ring block ;

[0030] The storage format of the soil layer distribution queue is as follows:

[0031] , wherein represents a soil layer label, is a count value of the soil layer , is a height proportion of the soil layer , is a height value of the soil layer .

[0032] In a further embodiment, the construction process of the three-dimensional space matrix is as follows:

[0033] According to the proportion statistical matrix of each ring soil layer, combined with the plane form of the excavation face of the shield tunnel and the excavation diameter, the real soil layer distribution of each ring is sliced to show, and a real space distribution three-dimensional matrix of the soil layer distribution of each ring is established.

[0034] In a further embodiment, the calculation process of the distribution area is as follows:

[0035] A two-dimensional rectangular coordinate system is established with the center of the tunnel section as the origin, the Y-axis direction is the diameter direction of the circle, and the radius of the circle is determined according to the proportion queue of each ring d , so the radius range of the circle is ;

[0036] According to the soil layer label of each ring, N band regions are obtained, and the upper boundary and the lower boundary of the band region are determined by the following formula:

[0037] , ; in the formula, , ;

[0038] Therefore, the band region ​​The area calculation formula of the area of each belt area is as follows:

[0039] ;

[0040] The area calculation of each belt area is completed The area ratio calculation formula is as follows:

[0041] , The area ratio of the belt area .

[0042] The beneficial effects of the present application are as follows: the present application realizes the automatic identification of the soil layers in each tunneling ring in the longitudinal section drawing of shield tunneling by adopting the computer language programming and image recognition mode, effectively replaces the manual measurement operation, greatly reduces the labor cost, and solves the gross error problem caused by manual measurement.

[0043] The present application can accurately classify and identify the different soil layer corresponding dyeing in the geological map by introducing the technology of combining super pixel segmentation and clustering algorithm, avoids the problem of missing area recognition caused by color tolerance in the image, automatically groups and positions all color parts in the image, and guarantees the accuracy of soil layer distribution position recognition.

[0044] The present application reconstructs the three-dimensional spatial distribution characteristics of the soil layers of shield tunneling by identifying the soil layer distribution information of the two-dimensional longitudinal section drawing, and intuitively displays the spatial distribution form of the soil layers of each tunneling ring; the calculation method of the spatial ratio of different soil layers in each tunneling ring is proposed, which provides an effective support method for defining the comprehensive geotechnical physical and mechanical parameters in the tunneling ring. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 is the flow chart of the shield ring geological identification representation method based on the super pixel segmentation algorithm of embodiment 1.

[0046] Figure 2 is the different soil layer differential dyeing processing, interference information removal and tunnel top and bottom interface color filling of the geological longitudinal section drawing of embodiment 1.

[0047] Figure 3 is the grouping result of the picture pixel points of the super pixel segmentation algorithm of embodiment 1.

[0048] Figure 4 is the two-dimensional mask matrix of each soil layer created by the super pixel segmentation algorithm of embodiment 1 and the combined gray scale and pseudo-color matrix result.

[0049] Figure 5 is the three-dimensional spatial distribution result of each ring soil layer of embodiment 1.

[0050] Figure 6 is the area of each zone layer calculated based on the relationship between the boundary line between different soil layers and the center of the tunneling section in Example 1. DETAILED DESCRIPTION

[0051] Example 1

[0052] As shown in Figure 1 , the present embodiment discloses a shield ring-in geological identification and characterization method based on a superpixel segmentation algorithm, which at least includes the following steps:

[0053] The geological section of the shield engineering project is preprocessed to generate a PNG format picture; the pixel of the PNG format picture is processed to determine the range of the soil layer penetrated by the shield excavation face;

[0054] The superpixel segmentation algorithm is used to identify and locate each soil layer in the range of the soil layer penetrated, and a corresponding soil layer mask matrix is created; the soil layer mask matrix is combined according to the arrangement position to obtain a pseudo-color matrix;

[0055] The corresponding number of shield tunnel structure rings is obtained, the geological section is segmented according to the inter-ring area based on the number of shield tunnel structure rings, and the pseudo-color matrix is formed into a geological distribution matrix of each ring;

[0056] The geological distribution matrix of each ring is scanned column by column, and the layer height in each ring is combined to obtain the proportion queue of each ring; according to the shape of the tunnel section, a three-dimensional space matrix of the soil layer distribution of each ring is constructed, and the three-dimensional space matrix is analyzed by slicing;

[0057] The distribution area and area proportion of each soil layer from high to low corresponding to the slice are calculated, and the same calculation is performed on the three-dimensional space matrix to complete the distribution characteristic analysis of the soil layer in the shield tunneling ring.

[0058] In further embodiments, the generation process of the PNG format picture is as follows:

[0059] The calibration elements in the geological section are deleted, only the layered soil layer information and the design interface of the tunnel top boundary and the tunnel bottom boundary are retained to obtain the processed geological section. For example, the calibration elements in the present embodiment can be geological drilling lines, elevation text annotations, legend interference lines, text content, etc.

[0060] The different soil layers of the processed geological section are differentially colored, the tunnel top boundary and the tunnel bottom boundary are specially colored, and are converted into a PNG format. In the present embodiment, as Figure 2As shown, there are four types of soil layers in the geological section, and the filling colors between each soil layer should have obvious distinction. Therefore, four colors, RGB(0,0,131), RGB(0,0,255), RGB(255,219,0), and RGB(128,0,0), are selected at equal intervals to complete the soil layer coloring difference filling operation.

[0061] After completing the soil layer coloring, special coloring is applied to the tunnel top and bottom interfaces, such as using RGB (255, 0, 0) to modify the line colors. After modification, the image is converted to a PNG image in a three-channel uint8 data storage format.

[0062] Furthermore, PNG format images are read to obtain a three-dimensional matrix of red, green, and blue color images. The interface positions of the tunnel top and bottom boundaries are obtained by picking fixed RGB (255, 0, 0) values. Based on the position of the tunnel top boundary in the three-dimensional matrix, the color image three-dimensional matrix is ​​segmented. Values ​​in the color image three-dimensional matrix outside the tunnel top boundary are filled with 0. Rows and columns not belonging to the tunnel top or bottom boundaries, or soil layers within the excavation section, are deleted, thus identifying and locating the specific soil layer range traversed by the shield excavation face.

[0063] Based on the above description, before identifying and locating each soil layer, it is also necessary to classify each layer within the excavation surface by color to avoid identification loss due to color differences in some areas.

[0064] Specifically, the image is first normalized by converting the RGB image to a Double type. Normalization to Scope, of which, Grayscale value These are the normalized pixel values.

[0065] Then, the following formula is used to create a mask that excludes regions with a pixel value of 0, and these regions are defined as valid pixels. :

[0066] ; Here are the R, G, and B values ​​corresponding to each pixel, and c represents the number of color channels. These are the row and column values ​​of each pixel in the image matrix.

[0067] Extract the RGB values ​​and coordinates of the effective pixels to form an initial RGB feature set. and initial coordinate feature set .

[0068] Furthermore, the initial coordinate feature set Represented as: , the initial RGB feature set is expressed as: , is a pixel point , the pixel value after normalization processing.

[0069] The spatial coordinates of each valid pixel point are normalized: , , X is a set of row values corresponding to each valid pixel point, Y is a set of column values corresponding to each valid pixel point.

[0070] The normalized spatial coordinates and the RGB features are integrated to obtain a feature vector : , wherein is a spatial weight factor, and the value in the embodiment is 0.2.

[0071] The number K of superpixels is adaptively determined according to the image size:

[0072] , wherein is the maximum superpixel number, generally taking the value of the number of soil layer types, if the current soil layer type is four, then the value of K is 4. represents the total number of all valid pixels in the image. is the expected pixel value of each superpixel, generally set to 100.

[0073] Based on the determination of the number of superpixels, the steps of identifying and positioning each soil layer are as follows: a five-dimensional feature vector is used to represent each pixel between the tunnel top boundary and the tunnel bottom boundary , and K superpixels are preset, the K superpixels correspond to different soil layers;

[0074] The distance between the pixel and each superpixel is calculated , wherein represents the five-dimensional feature vector of the pixel , and is the five-dimensional feature vector of the i-th superpixel, . Further, the five-dimensional feature vector contains three-dimensional color information and two-dimensional spatial information; , wherein represents the RGB color value of the pixel , represents the spatial coordinates of the pixel , This is a spatial weighting factor used to balance the importance of color similarity and spatial proximity. Correspondingly, the distance... The calculation formula is as follows:

[0075] ;

[0076] In the formula, , representing the Euclidean distance in the color space. , where is the Euclidean distance of the spatial location.

[0077] Combination Figure 3 Clustering objective function E; minimizing clustering objective function E determines the pixel. shortest distance According to the shortest distance Pixels Assigned to the corresponding soil layers. The clustering objective function E is expressed as:

[0078] Therefore, .

[0079] The steps for identifying and locating each soil layer are as follows: Each pixel between the tunnel top and bottom boundaries is represented by a five-dimensional feature vector. K superpixels are pre-defined, and the K superpixels correspond to different soil layers;

[0080] Calculate pixels Distance to each superpixel ,in, Represents pixels The five-dimensional eigenvectors, It is the first The five-dimensional feature vector of each superpixel ;

[0081] Create a clustering objective function E, and minimize the clustering objective function E to determine the pixels. shortest distance According to the shortest distance Pixels Distribute to the corresponding soil layers;

[0082] Correspondingly, the method for creating the soil layer mask matrix is ​​as follows: a two-dimensional mask matrix is ​​created and stored for the position of each soil layer in the matrix. A fixed and unique number corresponding to the soil layer type is used to represent the position of the current soil layer in the entire image matrix, and all other positions are marked as 0.

[0083] In a further embodiment, the soil layer mask matrices of each soil layer are merged, and each soil layer is identified by a fixed and unique number.

[0084] According to the unique color given to each soil layer number, the different soil layer positions are dyed to complete the pseudo-color matrix construction. For example, if there are four types of soil layers in this embodiment, each type of soil layer is fixedly and uniquely identified by 1, 2, 3 and 4, and four different colors are used to dye different soil layer positions to complete the pseudo-color soil layer distribution two-dimensional matrix construction, as shown in the following figure. Figure 4

[0085] The shield tunnel structure ring number is defined as , the pseudo-color matrix of size is segmented by ring area, wherein is the row number of the pseudo-color matrix, is the column number of the pseudo-color matrix. If the current geological section has 140 ring tunneling segments distributed, then . The segmentation process of the ring area is as follows:

[0086] The following formula is used to calculate the average number of columns allocated to each ring in the pseudo-color matrix and the remaining column number :

[0087] ; is the floor value, mod is the modulo operation;

[0088] For the ring block numbered , , the column number of the geological distribution matrix corresponding to the ring block is determined by the following analysis: if , then ; otherwise, ;

[0089] According to the determined column number , the pseudo-color matrix is continuously and non-overlappingly divided to obtain N geological distribution matrices: .

[0090] The method for obtaining the proportion queue of each ring is as follows:

[0091] Each column in the geological distribution matrix of each ring is traversed from top to bottom to form N proportion statistical matrices: , is the soil layer distribution queue of the ring block ;

[0092] The storage format of the soil layer distribution queue is as follows:

[0093] ​wherein, represents a soil layer label, is a count value of a soil layer , is a height ratio of a soil layer , is a height value of a soil layer . Further, the height ratio of a soil layer is obtained by dividing the count value of the soil layer t in the corresponding column by the total count value of the column, and the height value of the soil layer is obtained by dividing the height ratio of the soil layer by the diameter of the excavation face of the ring. For example, if there are three soil layers in the soil layer distribution queue , namely, soil layer one, soil layer two and soil layer three, the storage format of the soil layer distribution queue is as follows:

[0094] .

[0095] The construction process of the three-dimensional space matrix is as follows:

[0096] According to the proportion matrix of each ring soil layer, combined with the plane form of the excavation face of the shield tunnel and the excavation diameter, the real soil layer distribution of each ring is sliced and displayed, and a real space distribution three-dimensional matrix of the soil layer distribution of each ring is established, as shown in Figure 5 . Among them Figure 5 (a) is a three-dimensional space distribution result graph of each ring soil layer generated based on the proportion of each ring soil layer, which is generated from a single ring geological slice. Figure 5 (b) is a three-dimensional space distribution result graph of each ring soil layer generated based on the plane form of the excavation face, which is generated from a single ring geological three-dimensional space distribution matrix.

[0097] Combined with Figure 6 , the calculation process of the distribution area is as follows:

[0098] A two-dimensional rectangular coordinate system is established with the center of the tunnel section as the origin, the Y-axis direction as the diameter direction of the circle, and the radius of the circle is determined according to the proportion queue of each ring d , so the radius range of the circle is ;

[0099] According to the soil layer label of each ring, a band area is obtained , and the upper boundary and the lower boundary of the band area are determined by the following formula:

[0100] , ; in the formula, , ;

[0101] Therefore, the band area​ The area calculation formula of the region is as follows:

[0102] ;

[0103] The area calculation of each belt region is completed The area calculation formula of the area ratio is as follows:

[0104] , The area ratio of the belt region

[0105] In this embodiment, the radius of the circle The calculation formula of the radius is as follows:

[0106] .

[0107] Based on the above description, the same calculation is performed on the entire three-dimensional matrix and is weighted, so that the total area ratio of each soil layer in each tunneling ring is obtained, and the distribution characteristics of the soil layer in the shield tunneling ring are analyzed. In the example, a certain slice of the 20th ring is distributed with three soil layers, and the soil layer ratios obtained by the algorithm from top to bottom are 25.2921%, 50.6197% and 24.0883% respectively.

[0108] The application provides a shield ring-in-geology intelligent identification and representation method based on an improved superpixel segmentation algorithm. By using computer language programming and pixel point identification, the automatic identification of the soil layer in each tunneling ring in a shield construction tunneling longitudinal section diagram is realized, manual measurement work is effectively replaced, human cost is greatly reduced, and the gross error problem caused by manual measurement is solved. By introducing the intelligent algorithm of superpixel segmentation and clustering fusion, the dyeing corresponding to different soil layers in the geological diagram can be accurately classified and identified, the problem of missing identification of some areas in the image caused by color tolerance is avoided, all color parts in the image are automatically grouped and positioned, and the accuracy of soil layer distribution position identification is ensured. By identifying the soil layer distribution information of the two-dimensional longitudinal section diagram, the three-dimensional spatial distribution characteristics of the soil layer of the shield tunneling are reconstructed, the spatial distribution form of the soil layer of each tunneling ring is intuitively displayed, and a calculation method of the spatial area ratio of different soil layers in each tunneling ring is proposed, which provides an effective support method for defining the comprehensive geotechnical physical and mechanical parameters in the tunneling ring.

[0109] Although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to part of the technical features, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application shall be included in the protection scope of the application.​

Claims

1. A method for geological identification and characterization in a shield annulus based on a superpixel segmentation algorithm, characterized in that, At least comprising the steps of: Preprocessing the geological section of the shield engineering project to generate a PNG format picture; Processing the pixels of the PNG format picture to determine the range of the soil layers penetrated by the shield excavation face; Using a superpixel segmentation algorithm to identify and locate each soil layer in the range of the soil layers penetrated, and creating a corresponding soil layer mask matrix; Merging the soil layer mask matrices according to the arrangement positions to obtain a pseudo-color matrix; The corresponding shield tunnel structure ring number is acquired, the inter-ring area of the geological section is segmented by using the shield tunnel structure ring number, and the geological distribution matrix of each ring is formed based on the pseudo-color matrix. , the pseudo-color matrix of size is used to segment the inter-ring area, wherein, is the row number of the pseudo-color matrix, is the column number of the pseudo-color matrix; the segmentation process of the inter-ring area is as follows: The average number of columns in the pseudo-color matrix assigned to each ring pair is calculated using the following equation and the remaining number of columns : ; is rounded down, mod is a modulo operation; For the ring block numbered , , the value of the column number of the corresponding geological distribution matrix is determined by the following analysis: if , , then ; ; otherwise, ; According to the determined column number The pseudo-color matrix is continuously and non-overlappingly divided to obtain N geological distribution matrices: ; Scanning the geological distribution matrix of each ring column by column, combining the layer height in each ring to obtain the proportion queue of each ring; according to the shape of the tunnel section, a three-dimensional space matrix of the soil layer distribution of each ring is constructed, and the three-dimensional space matrix is analyzed by slicing; The distribution area and area proportion of each soil layer from high to low corresponding to the slice are calculated, and the same calculation is performed on the three-dimensional space matrix to complete the distribution characteristic analysis of the soil layer in the shield tunneling ring.

2. The method according to claim 1, wherein, The generation process of the PNG format picture is as follows: Delete the calibration elements in the geological section, and only keep the layered soil layer information and the design interface of the tunnel top boundary and the tunnel bottom boundary to obtain the processed geological section; Different soil layers of the processed geological section are differentially colored, the tunnel top boundary and the tunnel bottom boundary are specially colored, and are converted into PNG format.

3. The method of claim 1, wherein the method is characterized by, The determination method of the range of the soil layers penetrated is as follows: Obtain the three-dimensional matrix of the red, green and blue color images of the PNG format picture, obtain the interface position of the tunnel top boundary and the tunnel bottom boundary by picking up the fixed RGB (255, 0, 0) value, and segment the color image three-dimensional matrix according to the position of the tunnel top boundary in the three-dimensional matrix. Fill the corresponding values of the color image three-dimensional matrix outside the tunnel top boundary with 0, and delete the rows and columns that do not belong to the tunnel top boundary, the tunnel bottom boundary and the soil layer in the tunneling section.

4. The method of claim 1, wherein the method is characterized by, The steps of identifying and locating each soil layer are as follows: representing each pixel between the tunnel top boundary and the tunnel bottom boundary with a five-dimensional feature vector and presetting K superpixels, the K superpixels corresponding to different soil layers; Computing the pixels The distance between each superpixel wherein, denotes a pixel of the five-dimensional feature vector, is the five-dimensional feature vector of the th superpixel, ; creating a clustering objective function E, minimizing the clustering objective function E determines the shortest distance of the pixels , assigning the pixels to the corresponding soil layer according to the shortest distance ; Correspondingly, the creation method of the soil layer mask matrix is as follows: a two-dimensional mask matrix is created for the position of each soil layer in the matrix for storage, and a fixed unique number corresponding to the soil layer type is used to represent the position of the current soil layer in the entire picture matrix. The rest of the positions are marked as 0.

5. The method of claim 1, wherein the method is based on a superpixel segmentation algorithm. Merge the soil layer mask matrices of each soil layer, and identify each soil layer with a fixed and unique number; According to the number of each soil layer, a unique color is assigned to the soil layer position, and the pseudo-color matrix is constructed.

6. The method of claim 1, wherein the method is based on a superpixel segmentation algorithm. The method for obtaining the proportion queue of each ring is as follows: Count each column in the geological distribution matrix of each ring from top to bottom in turn to form N proportion statistical matrixes: , is the soil layer distribution queue of the ring block . The soil layer distribution queue The storage format is: wherein, represents a soil layer label, is a count value of a soil layer is a height value of a soil layer is a height ratio of a soil layer is a height value of a soil layer is a height value of a soil layer is a height value of a soil layer 7. The method according to claim 6, wherein, The construction process of the three-dimensional space matrix is as follows: According to the proportion queue of each ring, the plane form of the shield tunnel excavation face and the excavation diameter, the real soil layer distribution of each ring is displayed by slicing, and a real space distribution three-dimensional matrix of the soil layer distribution of each ring is established.

8. The method of claim 6, wherein the method is characterized by, The calculation process of the distribution area is as follows: A two-dimensional rectangular coordinate system is established with the center of the tunnel section as the origin, the Y-axis direction is the diameter direction of the circle, and the radius of the circle is determined according to the proportion queue of each ring d Therefore, the radius range of the circle is ; According to the soil layer label of each ring, the following formula is used to determine the belt area of the upper boundary and the lower boundary of the belt area : , ; wherein , ; Then, the area of the band region The area of the region is calculated as follows: ； The area of each band region is calculated The area ratio is calculated according to the following formula: , is the area ratio of the band region .

Images