Multi-layer substation building calculation method based on multi-view projection and hierarchical analysis

Abstract

This invention provides a method for measuring multi-layer substation buildings based on multi-view projection and hierarchical analysis, belonging to the field of substation building information measurement technology. The method includes: collecting point cloud data of the substation building and preprocessing the collected point cloud data; then performing hierarchical analysis of the point cloud data using gradient segmentation to obtain single-layer point cloud data; based on the obtained single-layer point cloud data, segmenting the single-layer point cloud data into a top plane and a side plane using normal vector analysis; projecting the segmented top plane and side plane to generate a multi-view projection image of the substation point cloud data; and based on the multi-view projection image, obtaining the polygonal boundaries of the substation building through contour extraction, measuring and integrating the dimensional information of each layer of the substation building. This invention achieves efficient extraction and accurate measurement of multi-layer substation buildings, solving problems such as incomplete noise removal, loss of details due to downsampling, and inaccurate height threshold segmentation in existing methods.

Description

Technical Field

[0001] This invention relates to the field of substation building information measurement technology, and in particular to a multi-level substation building measurement method based on multi-view projection and hierarchical analysis. Background Technology

[0002] With the continuous advancement of laser scanning technology, 3D point cloud data, as a high-precision spatial information representation method, has been widely used in various fields, especially in the measurement and analysis of buildings. Laser scanning can generate high-density, high-resolution 3D point cloud data, providing comprehensive spatial information, which lays the foundation for geometric modeling, dimensional measurement, and condition monitoring of buildings. However, when faced with complex building structures, especially large-scale, multi-layered buildings such as substations, traditional point cloud data processing methods often fall short of practical requirements in terms of efficiency and accuracy.

[0003] Existing technologies for processing building point cloud data typically employ conventional methods such as KD-tree-based denoising and height thresholding. These methods perform reasonably well when dealing with relatively simple single-story buildings or regular geometric structures, but they exhibit significant limitations when faced with complex multi-story buildings.

[0004] First, KD-tree-based denoising methods accelerate data retrieval by establishing a spatial index for the point cloud and identifying noise points through the statistical characteristics of local neighborhoods. While this method has some capability in handling noise within a small area, it often struggles to effectively distinguish noise points from useful data points in complex environments like substations with dense building elements. Non-building elements (such as cables, supports, and equipment) in the substation environment are highly intertwined with the structural features of the building. KD-tree-based denoising methods are affected by these complex elements, leading to incomplete noise removal or misinterpreting useful building data as noise, thus impacting subsequent building extraction and analysis.

[0005] Secondly, existing height threshold segmentation methods exhibit significant limitations in handling multi-story buildings. These methods typically rely on building height information, dividing the building into different levels by setting different height thresholds. However, in multi-story substation buildings, the height distribution of different floors is often irregular, sometimes even with insignificant height differences between floors. Because this method depends on height information, it is easily affected by local height variations within the building, leading to inaccurate floor segmentation. Furthermore, this method cannot handle complex internal and external geometric structures, such as the interweaving of exterior walls, interior walls, and staircases, resulting in unclear layering and incomplete extraction.

[0006] In summary, to address the complex cloud structure and dense noise at substation sites, it is necessary to introduce multi-angle projection to deconstruct the building structure. At the same time, the characteristics of multiple building layers and unclear boundaries also require point cloud hierarchical analysis to measure the precise geometric data of each layer of the plant. Summary of the Invention

[0007] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis. This method achieves efficient extraction and accurate measurement of multi-story substation buildings and solves problems such as incomplete noise removal, loss of details due to downsampling, and inaccurate height threshold segmentation in existing methods.

[0008] To achieve the above objectives, the present invention provides the following solution:

[0009] A method for measuring the dimensions of multi-story substation buildings based on multi-view projection and hierarchical analysis includes the following steps:

[0010] S1. Collect point cloud data of the substation building and preprocess the collected point cloud data;

[0011] S2. Perform layered analysis of the preprocessed point cloud data using the gradient segmentation method to obtain single-layer point cloud data;

[0012] S3. Based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis;

[0013] S4. Project the top plane and side plane obtained after segmentation to generate a multi-view projection image of the substation cloud data.

[0014] S5. Based on the multi-view projection image, obtain the polygonal boundary of the substation building through contour extraction, and measure and integrate the size information of each floor of the substation building.

[0015] Preferably, in step S1, the preprocessing of the collected point cloud data includes:

[0016] A top-down approach is used to denoise the point cloud data of the substation building.

[0017] Ground plane correction is performed on the point cloud data of the substation building after noise reduction;

[0018] After calibration, the point cloud data of the substation building is transformed into a coordinate system to facilitate height calculation.

[0019] Preferably, the top-down method for denoising the substation powerhouse point cloud data includes:

[0020] Calculate the height range of point cloud data:

[0021] z min =min(z) i );

[0022] z max =max(z) i );

[0023] Among them, z min For the minimum height, z max z is the maximum height. i Let be the height coordinates of the i-th point in the point cloud data;

[0024] Then, a height threshold z is set. threshold As a benchmark, points exceeding this threshold are marked as potential noise points:

[0025] z threshold =z max -λ(z max -z min );

[0026] Where λ is the adjustment coefficient, 0.01≤λ≤0.05;

[0027] Calculate the number N of point clouds in each layer of the substation powerhouse point cloud data. i And set the density threshold N threshold Points with a density lower than this value are considered noise points, as shown in the formula:

[0028]

[0029] Where n is the total number of point cloud data layers, α is an adjustment coefficient, 0.5≤α≤0.8; then low-density points are removed, i.e., N i <N threshold Points that are not in the noise category are marked as noise and removed.

[0030] Preferably, ground plane correction is performed on the denoised substation powerhouse point cloud data, including:

[0031] Ground points were extracted from the substation powerhouse point cloud data using RANSAC, and the equation of the ground plane was determined using a plane fitting method. The fitted plane equation was set as follows:

[0032] ax + by + cz + d = 0;

[0033] The plane parameters a, b, c, and d are solved using the least squares method. The solution process is as follows:

[0034] Select the ground point set {(x i ,y i ,z i)}, and construct matrix A and vector b;

[0035]

[0036] Solving for the plane parameters [a,b,d] using the least squares method:

[0037]

[0038] Where c = -1, the ground plane correction is performed on the point cloud data of the substation building.

[0039] Preferably, after correction, the point cloud data of the substation building is transformed into a coordinate system, including:

[0040] The ground plane normal vector [a, b, c] is rotated to the Z-axis direction, and the rotation axis is defined as the angle between the ground plane normal vector and the Z-axis. The rotation axis is calculated using the cross product, and the rotation angle θ is the angle to be rotated, calculated using the angle between the ground plane normal vector and the Z-axis.

[0041]

[0042] Subsequently, the point cloud data is rotated around the rotation axis to align the ground plane with the xy plane, thus completing the coordinate system transformation.

[0043] Preferably, in step S2, the preprocessed point cloud data is subjected to layered analysis using gradient segmentation to obtain single-layer point cloud data, including:

[0044] The gradient of the point density distribution in the height direction of the point cloud data is calculated using the following formula:

[0045]

[0046] Where G(z) is the gradient, dz is the change in height, and D(z) is the point density at height z, calculated by the following formula:

[0047]

[0048] By calculating the gradient G(z), regions with large gradient changes are identified, and a threshold T is set. g As a standard for floor division points; when |G(z)|>T g At this point, it is considered that a density abrupt change occurred at height z, which is defined as the hierarchical boundary of the substation building, and its height z is recorded. i As a floor dividing point;

[0049] Based on the obtained floor boundary point z i The point cloud data is cut along the Z-axis to obtain single-layer point cloud data.

[0050] Preferably, in step S3, based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis, including:

[0051] Based on the single-layer point cloud data, for each point in each layer, a neighborhood with radius r is selected. All point cloud data within this neighborhood are considered as the neighborhood point set of that point. Then, for each point p... i The neighborhood point set is P i :

[0052] P i ={p j |d(p j ,p i )≤r};

[0053] Where, p j For any point in the point cloud data, d(p) j ,p i Let p be a point. j Point P i distance;

[0054] For each point's neighborhood point set N i Calculate the covariance matrix C:

[0055]

[0056] Where μ is the neighborhood point set N of the point cloud. i The centroid is calculated using the following formula:

[0057]

[0058] Then, the covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and eigenvectors, where the eigenvector corresponding to the smallest eigenvalue is the normal vector.

[0059] The plane to which the calculation point belongs is determined by calculating the angle between the normal vector and the Z-axis. The normal vector n = [n...] x ,n y ,n z The angle θ between the Z-axis and the Z-axis:

[0060]

[0061] If the normal vector is less than a threshold, which is set to 10°, then the point belongs to the top surface; if the included angle θ is approximately 90°, then the point belongs to the side plane.

[0062] Preferably, in step S4, the top plane and side plane obtained after segmentation are projected to generate a multi-view projection image of the substation cloud data, including: projecting the top plane and side plane obtained after segmentation into the three-view direction and sampling the network to obtain the top view, front view and left view of the single-story substation building respectively.

[0063] Preferably, in step S5, based on the multi-view projection image, the polygonal boundary of the substation building is obtained through contour extraction, and the dimensional information of each floor of the substation building is measured and integrated, including:

[0064] Based on the top view, front view, and left view obtained in step S4, the front view and left view are measured using the improved Douglas-Peucker method. A point set P = {p1, p2, ..., p...} is defined on the plane. n}, identify and extract all approximate locations at y=y min Let P be the initial set of points, with two endpoints selected from it. base ;

[0065] For P, those that do not belong to P base For the other points, the Douglas-Peucker algorithm is applied to calculate the line p0p point by point. n The vertical distance is equal to the straight-line distance:

[0066]

[0067] Where × represents the cross product of vectors, and ||·|| represents the magnitude of the vectors. The point with the maximum distance is found; then the point p with the maximum distance is determined. max And set a threshold ∈, if the distance of the point is 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 process involves two parts, each of which is then recursively processed to obtain the final point set P. simplified ;

[0068] Finally, standard polygon edge extraction is performed on the top view of the factory building. Measurements are then taken based on the edges of the multiple views and cross-validated to obtain the final measurement results of the single-layer point cloud data and output them.

[0069] According to specific embodiments provided by the present invention, compared with the prior art, the present invention discloses the following technical effects:

[0070] (1) This invention has high precision and multi-level differentiation capabilities: Through multi-view projection technology, this invention can perform projection analysis on point cloud data from multiple perspectives, effectively solving the data confusion problem caused by complex structures and significantly improving the ability to differentiate between different floors. Combined with the hierarchical analysis method, this invention can also gradually extract the geometric information of the building at different scales and accurately identify the structural details of each floor. This layer-by-layer analysis and segmentation method, especially when dealing with multi-level buildings such as substation buildings, can avoid the problem of inaccurate layer division in traditional methods and achieve accurate measurement.

[0071] (2) The method provided by the present invention improves the efficiency of processing complex buildings and solves the problems of incomplete noise removal, loss of details due to downsampling, and inaccurate height threshold segmentation in the existing methods. By optimizing the processing steps, the present invention ensures efficient extraction and accurate measurement of multi-story substation buildings, and provides more reliable technical support for the fields of automated inspection, facility maintenance and engineering measurement of substation buildings. Attached Figure Description

[0072] 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.

[0073] Figure 1 This is a flowchart of a multi-level substation building measurement method based on multi-view projection and hierarchical analysis according to the present invention.

[0074] Figure 2 This is a schematic diagram provided for Embodiment 1 of the present invention;

[0075] Figure 3 This is a gradient layering analysis data map of substation powerhouse point cloud data provided in Embodiment 1 of the present invention;

[0076] Figure 4 This is a schematic diagram showing the separation of the roof plane and side plane of the substation building according to Embodiment 1 of the present invention;

[0077] Figure 5 This is a diagram showing the measurement results of the substation building dimensions provided in Embodiment 1 of the present invention; wherein, Figure 5 (a) is a front view of the substation building's dimensions. Figure 5 (b) is a top view of the substation building's dimensions. Figure 5 (c) is the dimension drawing of the substation building from the left view. Detailed Implementation

[0078] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0079] 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.

[0080] Example 1

[0081] like Figure 1 and Figure 2 As shown, this invention provides a method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis, including the following steps:

[0082] S1. Collect point cloud data of the substation building and preprocess the collected point cloud data;

[0083] S2. Perform layered analysis of the preprocessed point cloud data using the gradient segmentation method to obtain single-layer point cloud data;

[0084] S3. Based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis;

[0085] S4. Project the top plane and side plane obtained after segmentation to generate a multi-view projection image of the substation cloud data.

[0086] S5. Based on the multi-view projection image, obtain the polygonal boundary of the substation building through contour extraction, and measure and integrate the size information of each floor of the substation building.

[0087] In step S1, the preprocessing of the collected point cloud data includes:

[0088] A top-down approach is used to denoise the point cloud data of the substation building.

[0089] Ground plane correction is performed on the point cloud data of the substation building after noise reduction;

[0090] After calibration, the point cloud data of the substation building is transformed into a coordinate system to facilitate height calculation.

[0091] In this embodiment, the top-down method for denoising the substation powerhouse point cloud data includes:

[0092] Calculate the height range of point cloud data:

[0093] z min =min(z) i );

[0094] z max =max(z) i );

[0095] Among them, z min For the minimum height, z max z is the maximum height. i Let be the height coordinates of the i-th point in the point cloud data;

[0096] Then, a height threshold z is set. threshold As a benchmark, points exceeding this threshold are marked as potential noise points:

[0097] z threshold =z max -λ(z max -z min );

[0098] Where λ is the adjustment coefficient, 0.01≤λ≤0.05;

[0099] Calculate the number N of point clouds in each layer of the substation powerhouse point cloud data. i And set the density threshold N threshold Points with a density lower than this value are considered noise points, as shown in the formula:

[0100]

[0101] Where n is the total number of point cloud data layers, α is an adjustment coefficient, 0.5≤α≤0.8; then low-density points are removed, i.e., N i <N threshold Points that are not in the noise category are marked as noise and removed.

[0102] Secondly, ground plane correction is performed on the denoised substation powerhouse point cloud data, including:

[0103] Ground points were extracted from the substation powerhouse point cloud data using RANSAC, and the equation of the ground plane was determined using a plane fitting method. The fitted plane equation was set as follows:

[0104] ax + by + cz + d = 0;

[0105] The plane parameters a, b, c, and d are solved using the least squares method. The solution process is as follows:

[0106] Select the ground point set {(x i ,y i ,z i )}, and construct matrix A and vector b;

[0107]

[0108] Solving for the plane parameters [a,b,d] using the least squares method:

[0109]

[0110] Where c = -1, the ground plane correction is performed on the point cloud data of the substation building.

[0111] In addition, after the calibration is completed, the coordinate system of the substation powerhouse point cloud data is transformed, including:

[0112] The ground plane normal vector [a, b, c] is rotated to the Z-axis direction, and the rotation axis is defined as the angle between the ground plane normal vector and the Z-axis. The rotation axis is calculated using the cross product, and the rotation angle θ is the angle to be rotated, calculated using the angle between the ground plane normal vector and the Z-axis.

[0113]

[0114] Subsequently, the point cloud data is rotated around the rotation axis to align the ground plane with the xy plane, thus completing the coordinate system transformation.

[0115] In step S2, the preprocessed point cloud data is subjected to layered analysis using the gradient segmentation method to obtain single-layer point cloud data, including:

[0116] The gradient of the point density distribution in the height direction of the point cloud data is calculated using the following formula:

[0117]

[0118] Where G(z) is the gradient, dz is the change in height, and D(z) is the point density at height z, calculated by the following formula:

[0119]

[0120] By calculating the gradient G(z), regions with large gradient changes are identified, and a threshold T is set. g As a standard for floor division points; when |G(z)|>T g At this point, it is considered that a density abrupt change occurred at height z, which is defined as the hierarchical boundary of the substation building, and its height z is recorded. i As a floor dividing point;

[0121] Based on the obtained floor boundary point z i The point cloud data is cut along the Z-axis to obtain single-layer point cloud data, and the result is as follows. Figure 3 As shown.

[0122] In step S3, based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis, including:

[0123] Based on the single-layer point cloud data, for each point in each layer, a neighborhood with radius r is selected. All point cloud data within this neighborhood are considered as the neighborhood point set of that point. Then, for each point p... i The neighborhood point set of P i :

[0124] P i ={p j |d(p j ,p i )≤r};

[0125] Where, p j For any point in the point cloud data, d(p) j ,p i Let p be a point. j Point P i The distance;

[0126] For each point's neighborhood point set N i Calculate the covariance matrix C:

[0127]

[0128] Where μ is the neighborhood point set N of the point cloud. i The centroid is calculated using the following formula:

[0129]

[0130] Then, the covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and eigenvectors, where the eigenvector corresponding to the smallest eigenvalue is the normal vector.

[0131] The plane to which the calculation point belongs is determined by calculating the angle between the normal vector and the Z-axis. The normal vector n = [n...] x ,n y ,n z The angle θ between the Z-axis and the Z-axis:

[0132]

[0133] If the normal vector is less than a threshold, which is set to 10°, then the point belongs to the top plane; if the included angle θ is approximately 90°, then the point belongs to the side plane, as shown in the following results. Figure 4 As shown.

[0134] In step S4, the top plane and side plane obtained after segmentation are projected to generate a multi-view projection image of the substation cloud data, including: projecting the top plane and side plane obtained after segmentation into the three-view direction and sampling the network to obtain the top view, front view and left view of the single-story substation building respectively.

[0135] According to step S4, in step S5, based on the multi-view projection image, the polygonal boundary of the substation building is obtained through contour extraction, and the dimensional information of each floor of the substation building is measured and integrated, including:

[0136] Based on the top view, front view, and left view obtained in step S4, the front view and left view are measured using the improved Douglas-Peucker method. A point set P = {p1, p2, ..., p...} is defined on the plane. n}, identify and extract all approximate locations at y = y min Let P be the initial set of points, with two endpoints selected from it. base ;

[0137] For P, those that do not belong to P base For the other points, the Douglas-Peucker algorithm is applied to calculate the line p0p point by point. n The vertical distance is equal to the straight-line distance:

[0138]

[0139] Where × represents the cross product of vectors, and ||·|| represents the magnitude of the vectors. The point with the maximum distance is found; then the point p with the maximum distance is determined. max And set a threshold ∈, if the distance of the point is 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 process involves two parts, each of which is then recursively processed to obtain the final point set P. simplified ;

[0140] Finally, standard polygon edge extraction was performed on the top view of the factory building. Measurements were then taken based on the edges of the obtained multi-view data, and cross-validated to obtain the final measurement results of the single-layer point cloud data, which were then output. The obtained dimensional information results are as follows: Figure 5 As shown. (Refer to...) Figure 5 (a) The dimensions of the substation building in the front view are 24.336m in length and 10.65m in height, referring to... Figure 5 (b) The dimensions of the substation building in the top view are 24.461m long and 3.223m wide, according to reference. Figure 5(c) The dimensions of the substation building in the left view are 10.897m high and 3.041m wide. According to the final results, the dimensions are not much different, thus achieving accurate measurement of the multi-story substation building.

[0141] Therefore, the above-mentioned method for measuring multi-level substation buildings based on multi-view projection and hierarchical analysis achieves efficient extraction and accurate measurement of multi-level substation buildings, solving the problems of incomplete noise removal, loss of details due to downsampling, and inaccurate height threshold segmentation in existing methods.

[0142] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis, characterized in that, Includes the following steps: S1. Collect point cloud data of the substation building and preprocess the collected point cloud data; S2. The preprocessed point cloud data is subjected to layered analysis using the gradient segmentation method to obtain single-layer point cloud data; including: The gradient of the point density distribution in the height direction of the point cloud data is calculated using the following formula: Where G(z) is the gradient, dz is the change in height, and D(z) is the point density at height z, calculated by the following formula: By calculating the gradient G(z), regions with large gradient changes are identified, and a threshold T is set. g As a standard for floor division points; when |G(z)|>T g At this point, it is considered that a density abrupt change occurred at height z, which is defined as the hierarchical boundary of the substation building, and its height z is recorded. i As a floor dividing point; Based on the obtained floor boundary point z i The point cloud data is cut along the Z-axis to obtain single-layer point cloud data; S3. Based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis; S4. Project the top and side planes obtained after segmentation to generate a multi-view projection image of the substation cloud data. S5. Based on the multi-view projection image, obtain the polygonal boundary of the substation building through contour extraction, and measure and integrate the size information of each floor of the substation building.

2. The method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis according to claim 1, characterized in that, In step S1, the preprocessing of the collected point cloud data includes: A top-down approach is used to denoise the point cloud data of the substation building. Ground plane correction is performed on the point cloud data of the substation building after noise reduction; After calibration, the point cloud data of the substation building is transformed into a coordinate system to facilitate height calculation.

3. The method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis according to claim 2, characterized in that, The top-down approach to denoising the substation powerhouse point cloud data includes: Calculate the height range of point cloud data: With min =min(z i ); With max =max(z i ); Among them, z min For the minimum height, z max z is the maximum height. i Let be the height coordinates of the i-th point in the point cloud data; Then, a height threshold z is set. threshold As a benchmark, points exceeding this threshold are marked as potential noise points: With threshold =z max -λ(z max -With min ); Where λ is the adjustment coefficient, 0.01≤λ≤0.05; Calculate the number of point clouds N in each layer of the substation powerhouse point cloud data. i And set the density threshold N threshold Points with a density lower than this value are considered noise points, as shown in the formula: Where n is the total number of point cloud data layers, α is an adjustment coefficient, 0.5≤α≤0.8; then low-density points are removed, i.e., N i <N threshold Points that are not in the noise category are marked as noise and removed.

4. The method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis according to claim 3, characterized in that, Ground plane correction is performed on the denoised substation powerhouse point cloud data, including: Ground points were extracted from the substation powerhouse point cloud data using RANSAC, and the equation of the ground plane was determined using a plane fitting method. The fitted plane equation was set as follows: ax + by + cz + d = 0; The plane parameters a, b, c, and d are solved using the least squares method. The solution process is as follows: Select the ground point set {(x i ,y i ,z i )}, and construct matrix A and vector b; Solving for the plane parameters [a,b,d] using the least squares method: Where c = -1, the ground plane correction is performed on the point cloud data of the substation building.

5. The method for measuring multi-story substation buildings based on multi-view projection and hierarchical analysis according to claim 4, characterized in that, After calibration, the coordinate system of the substation powerhouse point cloud data is transformed, including: The ground plane normal vector [a, b, c] is rotated to the Z-axis direction, and the rotation axis is defined as the angle between the ground plane normal vector and the Z-axis. The rotation axis is calculated using the cross product, and the rotation angle θ is the angle to be rotated, calculated using the angle between the ground plane normal vector and the Z-axis. Subsequently, the point cloud data is rotated around the rotation axis to align the ground plane with the xy plane, thus completing the coordinate system transformation.

6. The method for measuring multi-level substation buildings based on multi-view projection and hierarchical analysis according to claim 1, characterized in that, In step S3, based on the obtained single-layer point cloud data, the single-layer point cloud data is segmented into a top plane and a side plane through normal vector analysis, including: Based on the single-layer point cloud data, for each point in each layer, a neighborhood with radius r is selected. All point cloud data within this neighborhood are considered as the neighborhood point set of that point. Then, for each point p... i The neighborhood point set is P i : P i ={p j |d(p j ,p i )≤r}; Where, p j For any point in the point cloud data, d(p) j ,p i Let p be a point. j Point P i The distance; For each point's neighborhood point set N i Calculate the covariance matrix C: Where μ is the neighborhood point set N of the point cloud. i The centroid is calculated using the following formula: Then, the covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and eigenvectors, where the eigenvector corresponding to the smallest eigenvalue is the normal vector. The plane to which the calculation point belongs is determined by calculating the angle between the normal vector and the Z-axis. The normal vector n = [n...] x ,n y ,n z The angle θ between the Z-axis and the Z-axis: If the normal vector is less than a threshold, which is set to 10°, then the point belongs to the top surface; if the included angle θ is approximately 90°, then the point belongs to the side plane.

7. The method for measuring multi-level substation buildings based on multi-view projection and hierarchical analysis according to claim 1, characterized in that, In step S4, the top plane and side plane obtained after segmentation are projected to generate a multi-view projection image of the substation cloud data, including: projecting the top plane and side plane obtained after segmentation into the three-view direction and sampling the network to obtain the top view, front view and left view of the single-story substation building respectively.

8. The method for measuring multi-level substation buildings based on multi-view projection and hierarchical analysis according to claim 7, characterized in that, In step S5, based on the multi-view projection image, the polygonal boundary of the substation building is obtained through contour extraction, and the dimensional information of each floor of the substation building is measured and integrated, including: Based on the top view, front view, and left view obtained in step S4, the front view and left view are measured using the improved Douglas-Peucker method. A point set P = {p1, p2, ..., p...} is defined on the plane. n }, identify and extract all approximate locations at y=y min Let P be the initial set of points, with two endpoints selected from it. base ; For P, those that do not belong to P base For the other points, the Douglas-Peucker algorithm is applied to calculate the line p0p point by point. n The vertical distance is equal to the straight-line distance: Where × represents the cross product of vectors, and ||·|| represents the magnitude of the vectors. The point with the maximum distance is found; then the point p with the maximum distance is determined. max And set a threshold ∈, if the distance of the point is 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 process involves two parts, each of which is then recursively processed to obtain the final point set P. simplified ; Finally, standard polygon edge extraction is performed on the top view of the factory building. Measurements are then taken based on the edges of the multiple views and cross-validated to obtain the final measurement results of the single-layer point cloud data and output them.

Images