A Method for Extracting and Detecting Geometric Information of Substation Building Pit Based on Laser Point Clouds

Abstract

This invention provides a method for extracting and detecting geometric information of substation foundation pits based on laser point clouds, belonging to the field of substation foundation pit information detection technology. The method includes: acquiring a complete substation foundation pit point cloud using a laser point cloud acquisition device; subsequently employing steps such as point cloud filtering, ground removal, multi-view projection, morphological operations, and polygon fitting to efficiently and accurately extract the foundation pit outline and calculate its geometric information, significantly improving the accuracy and reliability of foundation pit detection. Compared to existing technologies, this invention has strong feasibility in engineering projects, especially with complex point cloud data, while also exhibiting high stability and sufficient accuracy.

Description

Technical Field

[0001] This invention relates to the field of substation building foundation pit information detection technology, and in particular to a method for extracting and detecting geometric information of substation building foundation pits based on laser point clouds. Background Technology

[0002] Substations are critical facilities in power systems, and their construction quality directly impacts the security and reliability of power supply. As an integral part of substation infrastructure, the construction quality of the substation foundation pit is particularly important. Precise control is required during pit excavation to ensure the stability and safety of subsequent equipment installation. Traditional pit inspection methods rely primarily on manual measurement, which is time-consuming, inefficient, and has limited accuracy, failing to meet the high-efficiency and high-precision requirements of modern engineering. Therefore, employing laser point cloud technology for substation foundation pit inspection has become a necessary and effective solution.

[0003] Laser point cloud technology utilizes lidar equipment to collect three-dimensional spatial information, enabling rapid and accurate acquisition of detailed terrain and building data. Compared to traditional measurement methods, laser point cloud technology offers advantages such as high resolution, high precision, and high efficiency, allowing for comprehensive data acquisition even in complex environments. This is particularly important for the inspection of substation foundation pits, as these pits are typically located at construction sites with complex and rapidly changing environments. Laser point cloud technology can provide a complete three-dimensional model in a short time, offering reliable data support for subsequent analysis and decision-making. Furthermore, laser point cloud data can be compared with design drawings to promptly identify and correct construction deviations, ensuring construction quality.

[0004] However, the large volume of laser point cloud data from substation foundation pits, coupled with potential obstructions and noise, presents a major challenge in achieving real-time and accurate detection. Furthermore, the complex structures of various foundation pits, their irregular shapes, and inconsistent inclination levels create measurement difficulties, necessitating measurements from multiple angles to ensure stability. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method for extracting and detecting geometric information of substation building foundation pits based on laser point clouds. This method can efficiently and accurately extract the outline of the foundation pit and calculate its geometric information, significantly improving the accuracy and reliability of foundation pit detection. It also has strong feasibility in engineering projects, especially with complex point cloud data, exhibiting high stability and sufficient accuracy.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for extracting and detecting geometric information of substation foundation pits based on laser point clouds includes the following steps:

[0008] S1. Obtain complete point cloud data of the substation building foundation pit through laser point cloud acquisition equipment;

[0009] S2. Preprocess the point cloud data of the substation building foundation pit to obtain sparse point cloud data, and remove the ground point part of the sparse point cloud data to obtain the point cloud data of the bottom and wall of the substation building foundation pit.

[0010] S3. Based on the specific shape and characteristics of the substation building foundation pit, select three orthogonal directions to perform two-dimensional projection on the point cloud data of the bottom and wall of the substation building foundation pit, generate corresponding two-dimensional images, and convert the two-dimensional images into binary images.

[0011] S4. Apply morphological operations to the binary image to eliminate noise and small isolated regions in the binary image to obtain a smooth outline of the substation building pit. Then, perform maximum connected component detection on the processed binary image to identify and extract the pit edge in the maximum connected component.

[0012] S5. Perform polygon fitting on the extracted pit edge to form a geometric shape representing the pit outline, and calculate the geometric information of the fitted polygon to obtain the geometric information of the substation building pit.

[0013] Preferably, in step S2, the point cloud data of the substation building foundation pit is preprocessed to obtain sparse point cloud data, including:

[0014] S21. Remove noise and outliers from the point cloud data of the substation foundation pit using a statistical filtering algorithm, as shown in the following formula:

[0015]

[0016] In the formula, d i p is the average neighborhood distance of the i-th point, k is the number of neighboring points, and p is the average neighborhood distance of the i-th point. i and p j These are the coordinates of point i and its neighboring point j, respectively.

[0017] S22. Merge the processed point cloud data of the substation building foundation pit and convert it to a unified engineering coordinate system;

[0018] S23. Divide the point cloud data in a unified coordinate system into regular three-dimensional voxel regions, and select a representative point in each three-dimensional voxel region to achieve point cloud data downsampling. The formula is as follows:

[0019]

[0020] In the formula, v i It is the voxel index of point i, l x ly l z These are the dimensions of the voxel in the XYZ directions;

[0021] Based on the above-mentioned filtering, thinning, and fusion of the point cloud data of the substation building foundation pit, sparse point cloud data with noise removed is obtained.

[0022] Preferably, in step S2, the ground point portion of the sparse point cloud data is removed to obtain point cloud data of the bottom and walls of the substation building pit, including:

[0023] S24. Randomly select three points within the three-dimensional voxel region of the sparse point cloud data to obtain the initial ground plane equation: Ax + By + Cz + D = 0, that is, calculate the distance from all points to this plane:

[0024]

[0025] S25. Using the initial ground plane equation, set the iteration number k, the inlier threshold t, and the minimum inlier ratio d. If the distance from a point to the ground plane is less than the inlier threshold t, then that point is an inlier P in the plane. inliers :

[0026] P inliers ={p i |d i <t};

[0027] S26. If the number of interior points in step S25 is greater than the minimum interior point ratio d, then record the planar model and extract all interior points of the planar model to obtain ground point cloud data.

[0028] S27. Remove the ground point cloud data obtained in step S26, and repeat steps S24 to S26 to obtain point cloud data of the bottom and walls of the substation building pit.

[0029] Preferably, in step S3, based on the specific shape and characteristics of the substation building foundation pit, three orthogonal directions are selected to perform two-dimensional projection on the point cloud data of the bottom and walls of the substation building foundation pit, generating corresponding two-dimensional images, and converting the two-dimensional images into binary images, including:

[0030] S31. By performing principal component analysis on the point cloud data of the bottom and walls of the substation building foundation pit, three main directions of the substation building foundation pit are obtained. The main directions are the top view XOY plane, the front view XOZ plane, and the side view YOZ plane.

[0031] First, the point cloud data of the bottom and walls of the substation building pit are centralized, transforming the original point cloud matrix X into a centralized point cloud matrix X. c middle:

[0032] X c =X-μ;

[0033] In the formula, μ is the mean vector of the point cloud;

[0034] Subsequently, the centralized point cloud matrix X is calculated. c covariance matrix N is the number of points;

[0035] Finally, the covariance matrix is ​​decomposed into eigenvalues ​​CV = VΛ to obtain the eigenvalue diagonal matrix Λ and eigenvectors V. The eigenvectors corresponding to the diagonal elements in the eigenvalue diagonal matrix Λ are the orthogonal directions of the point cloud data.

[0036] S32. Based on the orthogonal direction of the point cloud data determined in step S31, perform coordinate transformation on the point cloud data so that the obtained orthogonal direction is aligned with the standard coordinate axis.

[0037] S33. After completing step S32, the point cloud data is projected onto the two-dimensional image in three orthogonal directions. The planar point cloud is projected onto the XOY plane, and the remaining pit wall point clouds are projected onto the XOZ and YOZ planes. Subsequently, the projected point cloud data is rasterized to generate a binary image.

[0038] Grid(i,j)={p k |p k ∈X proj and(x k ,y k )∈cell(i,j)};

[0039] In the formula, X proj Let cell(i,j) be the raster with row and column indices (i,j), and p be the point cloud to be projected. k Let i be any point in the grid cell(i,j).

[0040] Preferably, S4 includes:

[0041] S41. Perform n convolutional erosion operations with a 5*5 kernel on the binary image to remove noise and make the pit outline more obvious:

[0042]

[0043] In the formula, I is a binary image, and B is the convolution kernel. 5*5 It is a 5x5 matrix of all ones;

[0044] S42. After completing step S41, perform 2n+2m dilation operations with a 3*3 kernel on the binary image to expand the white area in the binary image and fill the small holes in the pit outline:

[0045]

[0046] In the formula, the convolution kernel B 3*3 It is a 3x3 matrix of all ones;

[0047] S43. After completing step S42, perform m convolutional erosion operations with a 5*5 kernel on the binary image, and obtain the pit outline image after restoration and dilation.

[0048] S44. Based on the foundation pit outline image, find the largest connected region in the foundation pit outline image, and extract the foundation pit edge in the largest connected region.

[0049] Preferably, S5 includes:

[0050] S51. For the XOY plane projection, the Douglas-Peucker algorithm is used to fit the polygon. All edge points of the XOY plane projection are considered as vertices of the polygon, and the first and last two points p0 and p1 are selected. n Starting from the point, calculate point by point to the line p0p. n The vertical distance is equal to the straight-line distance:

[0051]

[0052] In the formula, × represents the cross product of vectors, and ||·|| represents the magnitude of the vector. The process involves finding the point with the maximum distance, and then determining the maximum distance point d. max If d max If the value is greater than the set threshold ∈, then the polygon edge is segmented into p0p. max p max p n The problem is divided into two parts, and each part is processed recursively.

[0053] S52. After completing step S51, obtain the set of all edge points P. Project the coordinates of the set of edge points on the XOY plane in the opposite direction to the three-dimensional engineering coordinate system to obtain the polygons at the top and bottom of the foundation pit, the three-dimensional coordinates of the polygon corner points, and the side lengths of all sides of the polygon.

[0054] S53. Perform steps S51 to S52 on the XOZ and YOZ planes to perform polygon fitting, calculate the side length of the foundation pit, and perform cross-validation. Calculate the height of the foundation pit by using the height of the side projection of the foundation pit, and determine the geometric information of the substation building foundation pit.

[0055] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0056] This invention provides a method for extracting and detecting geometric information of substation building foundation pits based on laser point clouds, comprising: acquiring complete point cloud data of a substation building foundation pit using a laser point cloud acquisition device; preprocessing the point cloud data of the substation building foundation pit to obtain sparse point cloud data, and removing the ground point portion of the sparse point cloud data to obtain point cloud data of the bottom and walls of the substation building foundation pit; based on the specific shape and characteristics of the substation building foundation pit, selecting three orthogonal directions to perform two-dimensional projection on the point cloud data of the bottom and walls of the substation building foundation pit to generate corresponding two-dimensional images, and converting the two-dimensional images into binary images; applying morphological operations to the binary images to eliminate noise and small isolated regions in the binary images to obtain a smooth outline of the substation building foundation pit, and then performing maximum connected component detection on the processed binary images to identify and extract the foundation pit edges in the maximum connected component; performing polygon fitting on the extracted foundation pit edges to form a geometric shape representing the foundation pit outline, and calculating the geometric information of the fitted polygons to obtain the geometric information of the substation building foundation pit. This invention can efficiently and accurately extract the outline of the foundation pit and calculate its geometric information, which significantly improves the accuracy and reliability of foundation pit detection. At the same time, it has strong feasibility in engineering projects, especially with complex point cloud data, and has obvious advantages, with high stability and sufficient accuracy. Attached Figure Description

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in 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 paying any creative work.

[0058] Figure 1 A flowchart of the method provided by the present invention;

[0059] Figure 2 This is a schematic diagram of ground point cloud data provided in Embodiment 1 of the present invention;

[0060] Figure 3 This is a schematic diagram of the planar point cloud data of the foundation pit provided in Embodiment 1 of the present invention;

[0061] Figure 4 This is a schematic diagram of point cloud data provided in Embodiment 1 of the present invention; wherein, Figure 4 (a) and Figure 4 (b) are schematic diagrams of the original point cloud data;

[0062] Figure 5 This is a schematic diagram of point cloud data projection provided in Embodiment 1 of the present invention; wherein, Figure 5 (a) is a schematic diagram of point cloud data projection in the XOY direction. Figure 5 (b) is a schematic diagram of the point cloud data projection in the XOZ direction. Figure 5 (c) is a schematic diagram of point cloud data projection in the YOZ direction. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] To make the objectives, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] Example 1

[0066] like Figure 1 As shown, this invention provides a method for extracting and detecting geometric information of substation building foundation pits based on laser point clouds, including the following steps:

[0067] S1. Obtain complete point cloud data of the substation building foundation pit through laser point cloud acquisition equipment;

[0068] S2. Preprocess the point cloud data of the substation building foundation pit to obtain sparse point cloud data, and remove the ground point part of the sparse point cloud data to obtain the point cloud data of the bottom and wall of the substation building foundation pit.

[0069] S3. Based on the specific shape and characteristics of the substation building foundation pit, select three orthogonal directions to perform two-dimensional projection on the point cloud data of the bottom and wall of the substation building foundation pit, generate corresponding two-dimensional images, and convert the two-dimensional images into binary images.

[0070] S4. Apply morphological operations to the binary image to eliminate noise and small isolated regions in the binary image to obtain a smooth outline of the substation building pit. Then, perform maximum connected component detection on the processed binary image to identify and extract the pit edge in the maximum connected component.

[0071] S5. Perform polygon fitting on the extracted pit edge to form a geometric shape representing the pit outline, and calculate the geometric information of the fitted polygon to obtain the geometric information of the substation building pit.

[0072] In this embodiment, in step S2, the point cloud data of the substation building foundation pit is preprocessed to obtain sparse point cloud data, including:

[0073] S21. Remove noise and outliers from the point cloud data of the substation foundation pit using a statistical filtering algorithm, as shown in the following formula:

[0074]

[0075] In the formula, d i p is the average neighborhood distance of the i-th point, k is the number of neighboring points, and p is the average neighborhood distance of the i-th point. i and p j These are the coordinates of point i and its neighboring point j, respectively.

[0076] S22. Merge the processed point cloud data of the substation building foundation pit and convert it to a unified engineering coordinate system;

[0077] S23. Divide the point cloud data in a unified coordinate system into regular three-dimensional voxel regions, and select a representative point in each three-dimensional voxel region to achieve point cloud data downsampling. The formula is as follows:

[0078]

[0079] In the formula, v i It is the voxel index of point i, l x l y l z These are the dimensions of the voxel in the XYZ directions;

[0080] Based on the above-mentioned filtering, thinning, and fusion of the point cloud data of the substation building foundation pit, sparse point cloud data with noise removed is obtained.

[0081] In this embodiment, in step S2, the ground point portion of the sparse point cloud data is removed to obtain point cloud data of the bottom and walls of the substation building pit, including:

[0082] S24. Randomly select three points within the three-dimensional voxel region of the sparse point cloud data to obtain the initial ground plane equation: Ax + By + Cz + D = 0, that is, calculate the distance from all points to this plane:

[0083]

[0084] S25. Using the initial ground plane equation, set the iteration number k, the inlier threshold t, and the minimum inlier ratio d. If the distance from a point to the ground plane is less than the inlier threshold t, then that point is an inlier P in the plane. inliers :

[0085] P inliers ={p i |d i <t};

[0086] S26, such as Figure 2 As shown, if the number of interior points in step S25 is greater than the minimum interior point ratio d, then the planar model is recorded, and all interior points of the planar model are extracted to obtain ground point cloud data.

[0087] S27, such as Figure 3 As shown, the ground point cloud data obtained in step S26 is removed, and steps S24 to S26 are repeated to obtain point cloud data of the bottom and walls of the substation building pit.

[0088] In this embodiment, in step S3, based on the specific shape and characteristics of the substation building foundation pit, three orthogonal directions are selected to perform two-dimensional projection on the point cloud data of the bottom and walls of the substation building foundation pit, generating corresponding two-dimensional images, and converting the two-dimensional images into binary images, including:

[0089] S31. By performing principal component analysis on the point cloud data of the bottom and walls of the substation building foundation pit, three main directions of the substation building foundation pit are obtained. The main directions are the top view XOY plane, the front view XOZ plane, and the side view YOZ plane.

[0090] First, the point cloud data of the bottom and walls of the substation building pit are centralized, transforming the original point cloud matrix x into a centralized point cloud matrix X. c middle:

[0091] x c =X-μ;

[0092] In the formula, μ is the mean vector of the point cloud;

[0093] Subsequently, the centralized point cloud matrix X is calculated. c covariance matrix N is the number of points;

[0094] Finally, the covariance matrix is ​​decomposed into eigenvalues ​​CV = VΛ to obtain the eigenvalue diagonal matrix Λ and eigenvectors V. The eigenvectors corresponding to the diagonal elements in the eigenvalue diagonal matrix Λ are the orthogonal directions of the point cloud data.

[0095] S32. Based on the orthogonal direction of the point cloud data determined in step S31, perform coordinate transformation on the point cloud data so that the obtained orthogonal direction is aligned with the standard coordinate axis.

[0096] S33. After completing step S32, the point cloud data is projected onto the two-dimensional image in three orthogonal directions. The planar point cloud is projected onto the XOY plane, and the remaining pit wall point clouds are projected onto the XOZ and YOZ planes. Subsequently, the projected point cloud data is rasterized to generate a binary image.

[0097] Grid(i,j)={p k |pk ∈X proj and(x k ,y k )∈cell(i,j)};

[0098] In the formula, X proj Let cell(i,j) be the raster with row and column indices (i,j), and p be the point cloud to be projected. k For any point in the raster cell (i,j), here is a schematic diagram of the original point cloud data, as shown in the figure. Figure 4 (a) and Figure 4 (b) shows a schematic diagram of the projected point cloud data, and the result is as follows: Figure 5 (a) Figure 5 (b) and Figure 5 As shown in (c).

[0099] In this embodiment, step S4 includes:

[0100] S41. Perform n convolutional erosion operations with a 5*5 kernel on the binary image to remove noise and make the pit outline more obvious:

[0101]

[0102] In the formula, I is a binary image, and B is the convolution kernel. 5*5 It is a 5x5 matrix of all ones;

[0103] S42. After completing step S41, perform 2n+2m dilation operations with a 3*3 kernel on the binary image to expand the white area in the binary image and fill the small holes in the pit outline:

[0104]

[0105] In the formula, the convolution kernel B 3*3 It is a 3x3 matrix of all ones;

[0106] S43. After completing step S42, perform m convolutional erosion operations with a 5*5 kernel on the binary image, and obtain the pit outline image after restoration and dilation.

[0107] S44. Based on the foundation pit outline image, find the largest connected region in the foundation pit outline image, and extract the foundation pit edge in the largest connected region.

[0108] In this embodiment, step S5 includes:

[0109] S51. For the XOY plane projection, the Douglas-Peucker algorithm is used to fit the polygon. All edge points of the XOY plane projection are considered as vertices of the polygon, and the first and last two points p0 and p1 are selected. nStarting from the point, calculate point by point to the line p0p. n The vertical distance is equal to the straight-line distance:

[0110]

[0111] In the formula, × represents the cross product of vectors, and ||·|| represents the magnitude of the vector. The process involves finding the point with the maximum distance, and then determining the maximum distance point d. max If d max If the value is greater than the set threshold ∈, then the polygon edge is segmented into o0p. max p max p n The problem is divided into two parts, and each part is processed recursively.

[0112] S52. After completing step S51, obtain the set of all edge points P. Project the coordinates of the set of edge points on the XOY plane in the opposite direction to the three-dimensional engineering coordinate system to obtain the polygons at the top and bottom of the foundation pit, the three-dimensional coordinates of the polygon corner points, and the side lengths of all sides of the polygon.

[0113] S53. Perform steps S51 to S52 on the XOZ and YOZ planes to perform polygon fitting, calculate the side length of the foundation pit, and perform cross-validation. Calculate the height of the foundation pit by using the height of the side projection of the foundation pit, and determine the geometric information of the substation building foundation pit.

[0114] Therefore, this invention employs the aforementioned method for extracting and detecting geometric information of substation foundation pits based on laser point clouds. It acquires a complete substation foundation pit point cloud using a laser point cloud acquisition device, and then utilizes steps such as point cloud filtering, ground removal, multi-view projection, morphological operations, and polygon fitting to efficiently and accurately extract the foundation pit outline and calculate its geometric information, significantly improving the accuracy and reliability of foundation pit detection. Compared to existing technologies, this invention has strong feasibility in engineering projects, especially with complex point cloud data, while also exhibiting high stability and sufficient accuracy.

[0115] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for extracting and detecting geometric information of substation foundation pits based on laser point clouds, characterized in that, Includes the following steps: S1. Obtain complete point cloud data of the substation building foundation pit through laser point cloud acquisition equipment; S2. Preprocess the point cloud data of the substation building foundation pit to obtain sparse point cloud data, and remove the ground point part of the sparse point cloud data to obtain the point cloud data of the bottom and wall of the substation building foundation pit. S3. Based on the specific shape and characteristics of the substation building foundation pit, select three orthogonal directions to perform two-dimensional projection on the point cloud data of the bottom and wall of the substation building foundation pit, generate corresponding two-dimensional images, and convert the two-dimensional images into binary images. S4. Apply morphological operations to the binary image to eliminate noise and small isolated regions in the binary image to obtain a smooth outline of the substation building pit. Then, perform maximum connected component detection on the processed binary image to identify and extract the pit edge in the maximum connected component. S5. Perform polygon fitting on the extracted pit edges to form a geometric shape representing the pit outline, and calculate the geometric information of the fitted polygon to obtain the geometric information of the substation building pit; including: S51. For the XOY plane projection, the Douglas-Peucker algorithm is used to fit the polygon. All edge points of the XOY plane projection are considered as vertices of the polygon, and the first and last two points are selected. and Starting from the point, calculate point by point to the line. The vertical distance is equal to the straight-line distance: ； In the formula, Represents the cross product of vectors. The vector's magnitude is used to find the point with the maximum distance, and then the point with the maximum distance is determined. ;like Greater than the set threshold Then the polygon edges will be divided into , The problem is divided into two parts, and each part is processed recursively. S52. After completing step S51, obtain the set of all edge points. The coordinates of the edge point set on the XOY plane are projected in reverse to the three-dimensional engineering coordinate system to obtain the polygons at the top and bottom of the foundation pit, the three-dimensional coordinates of the polygon corner points, and the side lengths of all sides of the polygon. S53. Perform steps S51~S52 on the XOZ and YOZ planes to perform polygon fitting, calculate the side length of the foundation pit, and perform cross-validation. Calculate the height of the foundation pit by using the height of the side projection of the foundation pit, and determine the geometric information of the substation building foundation pit.

2. The method for extracting and detecting geometric information of substation foundation pits based on laser point clouds according to claim 1, characterized in that, In step S2, the point cloud data of the substation building foundation pit is preprocessed to obtain sparse point cloud data, including: S21. Remove noise and outliers from the point cloud data of the substation foundation pit using a statistical filtering algorithm, as shown in the following formula: ； In the formula, It is The average neighborhood distance of each point It is the number of neighboring points. and They are points and neighboring points The coordinates; S22. Merge the processed point cloud data of the substation building foundation pit and convert it to a unified engineering coordinate system; S23. Divide the point cloud data in a unified coordinate system into regular three-dimensional voxel regions, and select a representative point in each three-dimensional voxel region to achieve point cloud data downsampling. The formula is as follows: ； In the formula, Yes voxel index, These are the dimensions of the voxel in the XYZ directions; Based on the above-mentioned filtering, thinning, and fusion of the point cloud data of the substation building foundation pit, sparse point cloud data with noise removed is obtained.

3. The method for extracting and detecting geometric information of substation foundation pits based on laser point clouds according to claim 2, characterized in that, In step S2, the ground point portion of the sparse point cloud data is removed to obtain point cloud data of the bottom and walls of the substation building pit, including: S24. Randomly select three points within the three-dimensional voxel region of the sparse point cloud data to obtain the initial ground plane equation: That is, to calculate the distance from all points to the ground plane: ； S25. Using the initial ground plane equation, set the number of iterations. Interior point threshold and minimum interior point ratio If the distance from a point to the ground plane is less than the interior point threshold... Then the point is a point in the ground plane. : ； S26. If the number of interior points in step S25 is greater than the minimum interior point ratio Then, the planar model is recorded, and all interior points of the planar model are extracted to obtain ground point cloud data; S27. Remove the ground point cloud data obtained in step S26, and repeat steps S24 to S26 to obtain point cloud data of the bottom and walls of the substation building pit.

4. The method for extracting and detecting geometric information of substation foundation pits based on laser point clouds according to claim 3, characterized in that, In step S3, based on the specific shape and characteristics of the substation foundation pit, three orthogonal directions are selected to perform two-dimensional projection on the point cloud data of the bottom and walls of the substation foundation pit, generating corresponding two-dimensional images. These two-dimensional images are then converted into binary images, including: S31. By performing principal component analysis on the point cloud data of the bottom and walls of the substation building foundation pit, three main directions of the substation building foundation pit are obtained. The main directions are the top view XOY plane, the front view XOZ plane, and the side view YOZ plane. First, the point cloud data of the bottom and walls of the substation building pit are centered, and the original point cloud matrix is... Transformation to a centralized point cloud matrix middle: ； In the formula, is the mean vector of the point cloud; Subsequently, the centralized point cloud matrix is ​​calculated. covariance matrix , It is the number of points; Finally, eigenvalue decomposition is performed on the covariance matrix. The eigenvalue diagonal matrix is ​​obtained. and eigenvectors eigenvalue diagonal matrix The feature vectors corresponding to the diagonal elements are the orthogonal directions of the point cloud data. S32. Based on the orthogonal direction of the point cloud data determined in step S31, perform coordinate transformation on the point cloud data so that the obtained orthogonal direction is aligned with the standard coordinate axis. S33. After completing step S32, the point cloud data is projected onto the two-dimensional image in three orthogonal directions. The planar point cloud is projected onto the XOY plane, and the remaining pit wall point clouds are projected onto the XOZ and YOZ planes. Subsequently, the projected point cloud data is rasterized to generate a binary image. ； In the formula, For the point cloud to be projected, Row and column indexes The grid, For grid Any point.

5. The method for extracting and detecting geometric information of substation foundation pits based on laser point clouds according to claim 4, characterized in that, The S4 includes: S41. Perform n convolutions on the binary image using the following kernel: The erosion operation removes noise from the binary image, making the pit outline more obvious: ； In the formula, For binary images, convolution kernel for A matrix of all ones; S42. After completing step S41, perform 2n+2m convolutions on the binary image with the following kernels: The dilation operation expands the white area in the binary image, filling in the small holes in the pit outline: ； In the formula, the convolution kernel for A matrix of all ones; S43. After completing step S42, perform m convolutions on the binary image with the following kernel: The erosion operation is performed, and the expansion is restored to obtain the outline image of the foundation pit; S44. Based on the foundation pit outline image, find the largest connected region in the foundation pit outline image, and extract the foundation pit edge in the largest connected region.

Images