A method and device for checking and correcting the appearance of a three-dimensional model of a substation equipment

Abstract

The present invention discloses a method and device for appearance verification and correction of a three-dimensional model of a substation equipment. The method utilizes a registration algorithm to align the model point cloud with the real-scene scanning point cloud posture, thereby aligning the three-dimensional model with the real-scene scanning point cloud posture, and then rendering the redundant triangular facets of the three-dimensional model and the missing data points in the model point cloud according to the apparent difference between the three-dimensional model after the posture alignment and the real-scene scanning point cloud; the appearance verification and correction method of the present invention is not a direct difference between heterogeneous point clouds, but renders the redundant triangular facets of the three-dimensional model and the missing data points in the model point cloud according to the apparent difference between the three-dimensional model after the posture alignment and the real-scene scanning point cloud, thereby avoiding the difference caused by the non-corresponding data points of the heterogeneous data, so the appearance verification and correction method of the present invention has high accuracy.

Description

Technical Field

[0001] The invention belongs to the field of digital handover of power grid projects, and in particular relates to a method and a device for checking and correcting the appearance of a three-dimensional model of a substation equipment. Background Art

[0002] The digital transfer of power grid project data across multiple links is an important measure to achieve horizontal integration of data from multiple departments and vertical integration of data across links. The three-dimensional data of power grid projects is mainly the design model in GIM format during the design stage. After entering the inspection stage, the three-dimensional scanned point cloud data is obtained by scanning by drones or ground lidar. There are certain apparent differences between the design model of the same substation equipment and the point cloud data after construction acceptance. Therefore, technicians are required to perform an apparent check on the three-dimensional model, aiming to check and correct the apparent consistency of the transferred three-dimensional model and the scanned point cloud data.

[0003] At present, the apparent inspection and correction work of power grid engineering infrastructure transfer and maintenance is mainly carried out based on the "Three-dimensional Design Model Interaction Specification for Power Transmission and Transformation Engineering" and "Three-dimensional Design Modeling Specification for Power Transmission and Transformation Engineering". However, it is still mainly based on manual verification and return correction, resulting in low efficiency of digital handover of power grid engineering data and insufficient support for business by digital handover results.

[0004] Some technicians use the three-dimensional comparison method between point clouds to check the apparent consistency of the three-dimensional model and the scanned point cloud data. Most of these three-dimensional comparison methods first discretize the three-dimensional model to obtain the model point cloud data, and then differentiate the model point cloud data from the scanned point cloud data to obtain the difference. However, the model point cloud data and the scanned point cloud data are heterogeneous point cloud data, in which the data points in the model point cloud data are evenly distributed, while the data points in the scanned point cloud data are usually non-uniform and usually accompanied by some noise points, which will cause the data points between the model point cloud data and the scanned point cloud data to not accurately correspond. If the two point cloud data are directly differentiated, the differential result will include both the difference between the true shape of the three-dimensional model and the scanned point cloud data, and the difference caused by the non-corresponding data points of the heterogeneous data. Therefore, this type of method will produce large errors when applied to the appearance verification of the three-dimensional model of substation equipment.

[0005] In addition, during the model design phase, technicians will mostly color the 3D model of the substation equipment according to the color of existing substation equipment of the same type. Therefore, the 3D model does not have real color texture, and most 3D models have large color differences from the actual substation equipment. In the verification and correction of the appearance consistency between the 3D model and the scanned point cloud data, it generally relies on manual comparison and modification in the later stage, which further reduces the efficiency of digital transfer of power grid project data. Summary of the invention

[0006] In order to solve the problems existing in the prior art, the present invention provides a method and device for checking and correcting the appearance of a three-dimensional model of a substation equipment.

[0007] The present invention adopts the following technical solution:

[0008] A method for checking and correcting the appearance of a three-dimensional model of a substation equipment, wherein the three-dimensional model is composed of a plurality of triangular facets, and the method for checking and correcting the appearance comprises:

[0009] Obtain a real-scene scanning point cloud of the target substation equipment, which is composed of a number of data points with color attributes;

[0010] Discretely sample the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points;

[0011] Using a registration algorithm to align the model point cloud with the real scene scanning point cloud posture, thereby aligning the three-dimensional model with the real scene scanning point cloud posture;

[0012] Rendering redundant triangular facets of the three-dimensional model and missing data points of the three-dimensional model according to the apparent difference between the three-dimensional model after posture alignment and the real scene scanning point cloud;

[0013] According to the color attributes of the data points in the real-scene scanning point cloud, the triangular facets of the three-dimensional model are colored to complete the texture correction.

[0014] Furthermore, the discretization sampling is performed using a Poisson-Disk distribution grid sampling method, including the following specific steps:

[0015] 2.1) Using a random uniform sampling algorithm to obtain initial sample points on each triangular face of the 3D model;

[0016] 2.2) Use the minimum radius threshold to compare the initial sample points on each triangular patch, select the final sample points on each triangular patch, and use them as data points of the model point cloud to form a model point cloud.

[0017] Furthermore, the specific steps of using the minimum radius threshold to compare the initial sample points on each triangular facet and screen out the final sample points on each triangular facet include:

[0018] 2.2.1) Select an initial sample point and calculate the distance between the selected initial sample point and other initial sample points. If the distance between the selected initial sample point and other initial sample points is greater than the set minimum radius threshold, the selected initial sample point is recorded as the final sample point;

[0019] The distance between the selected initial sample point and other initial sample points adopts the geodesic distance of the veneer surface. For the initial sample point p i and other sample points p j The geodesic distance d between G The specific calculation formula is as follows:

[0020]

[0021]

[0022] Where, d E is the Euclidean distance between two points; is the initial sample point p i To other sample points p j The unit direction vector of ; is the normal vector of the sample point;

[0023] 2.2.2) Repeat step 2.2.1) and determine whether each initial sample point is a final sample point.

[0024] Furthermore, the specific step of aligning the model point cloud with the real scene scanning point cloud using the registration algorithm includes:

[0025] 3.1) Performing model registration on the model point cloud and the real scene scanning point cloud using the Teaser++ algorithm to obtain a posture transformation matrix between the model point cloud and the real scene scanning point cloud;

[0026] 3.2) Using the posture transformation matrix, the posture of the model point cloud is aligned to the coordinate system of the real scene scanning point cloud.

[0027] Furthermore, the specific steps of rendering redundant triangular facets of the three-dimensional model and missing data points in the model point cloud according to the apparent difference between the three-dimensional model after the posture alignment and the real scene scanning point cloud include:

[0028] 4.1) Select any triangular patch in the three-dimensional model and randomly obtain N sampling points in the triangular patch;

[0029] 4.2) Perform a field radius search based on the corresponding position of each sampling point in the 3D model in the real scene scan point cloud. If the number of sampling points that can find neighboring points is greater than αN, α∈[0,1], the selected triangular face is not a redundant triangular face; if the number of sampling points that can find neighboring points is less than or equal to αN, the selected triangular face is a redundant triangular face;

[0030] 4.3) Repeating steps 4.1) to 4.2), determining whether each triangular face in the three-dimensional model is redundant, and rendering the redundant triangular face;

[0031] 4.4) Perform a field radius search at the corresponding position of each data point of the real scene scanning point cloud in the three-dimensional model. If no neighboring point is found, the data point is considered to be a missing data point in the three-dimensional model, and the corresponding data point is rendered in the real scene scanning point cloud.

[0032] Furthermore, the specific steps of performing color correction processing on the triangular facets of the three-dimensional model according to the color attributes of the data points in the real scene scanning point cloud include:

[0033] 5.1) Select any triangular patch in the three-dimensional model, obtain five sampling points on the angular bisector of the triangular patch, perform a field radius search based on the corresponding position of each sampling point in the three-dimensional model in the real scene scanning point cloud, obtain 10 neighboring points corresponding to each sampling point, and use the obtained 50 neighboring points as elements in the set to form a neighbor point set E i ;

[0034] 5.2) The neighbor point set E i All neighboring points in are projected onto the plane where the selected triangle patch is located, and the projection points within the triangle patch are screened out to form a projection point set F i ;

[0035] 5.3) Calculate the projection point set F i The average value of the color attributes of the neighboring points corresponding to each projection point in is used as the color attribute of the selected triangle patch.

[0036] 5.4) Repeat steps 5.1) to 5.3) to color each triangular face in the three-dimensional model to complete texture correction.

[0037] Furthermore, the real-scene scanning point cloud with color attributes of the target substation equipment is acquired through a three-dimensional laser radar.

[0038] A device for checking and correcting the appearance of a three-dimensional model of a substation equipment, wherein the three-dimensional model is composed of a plurality of triangular facets, and the device for checking and correcting the appearance comprises:

[0039] A real-scene scanning point cloud acquisition module is used to acquire a real-scene scanning point cloud composed of a number of data points with color attributes for a target substation;

[0040] A discrete sampling module is used to discretize the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points;

[0041] A posture alignment module, used to align the model point cloud with the real scene scanning point cloud using a registration algorithm, thereby aligning the three-dimensional model with the real scene scanning point cloud;

[0042] An appearance difference rendering module is used to render redundant triangular facets of the three-dimensional model and missing data points in the three-dimensional model according to the appearance difference between the three-dimensional model after posture alignment and the real scene scanning point cloud;

[0043] The texture correction module is used to color the triangular facets of the three-dimensional model according to the color attributes of the data points in the real-scene scanning point cloud to complete the texture correction.

[0044] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the above method.

[0045] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The present invention provides a method and device for appearance verification and correction of a three-dimensional model of a substation equipment. The method utilizes a registration algorithm to align the model point cloud with the real-scene scanning point cloud posture, thereby aligning the three-dimensional model with the real-scene scanning point cloud posture, and then rendering the redundant triangular facets of the three-dimensional model and the missing data points in the model point cloud according to the apparent difference between the three-dimensional model after the posture alignment and the real-scene scanning point cloud; the appearance verification and correction method of the present invention is not a direct difference between heterogeneous point clouds, but renders the redundant triangular facets of the three-dimensional model and the missing data points in the model point cloud according to the apparent difference between the three-dimensional model after the posture alignment and the real-scene scanning point cloud, thereby avoiding the difference caused by the non-corresponding data points of the heterogeneous data, so the appearance verification and correction method of the present invention has high accuracy.

[0048] The appearance verification and correction method of the present invention also colors the triangular facets of the three-dimensional model according to the color attributes of the data points in the real-scene scanning point cloud to complete texture correction. Compared with manual comparison and correction, the appearance verification and correction method of the present invention greatly reduces the workload and improves the efficiency of digital transfer of power grid engineering data. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of the method for checking and correcting the appearance of the three-dimensional model of the substation equipment in the first embodiment;

[0050] FIG2( a ) is a schematic diagram of a three-dimensional model of a lightning arrester in an application embodiment;

[0051] FIG2( b ) is a schematic diagram of a point cloud of a real-scene scan of a lightning arrester in an application embodiment;

[0052] FIG2( c ) is a schematic diagram of the alignment of the three-dimensional model of the arrester and the point cloud of the real scene scan in the application embodiment;

[0053] FIG3( a ) is a schematic diagram of a difference result of a lightning arrester model in an application embodiment, from a first perspective;

[0054] FIG3( b ) is a second perspective diagram of a differential result of a lightning arrester model in an application embodiment;

[0055] FIG4( a ) is a schematic diagram of a color scan point cloud of a lightning arrester in an application embodiment;

[0056] FIG4( b ) is a schematic diagram of a texture-free three-dimensional model of a lightning arrester in an application embodiment;

[0057] FIG4( c ) is a coloring diagram of the three-dimensional model of the arrester in the application embodiment, from perspective 1;

[0058] FIG4( d ) is a second perspective diagram of the coloring diagram of the three-dimensional model of the arrester in the application embodiment. DETAILED DESCRIPTION

[0059] Embodiment 1:

[0060] A method for checking and correcting the appearance of a three-dimensional model of a substation equipment, wherein the three-dimensional model is composed of a plurality of triangular facets, and the method for checking and correcting the appearance includes:

[0061] A real-scene scanning point cloud composed of a number of data points with color attributes of a target substation is obtained; the real-scene scanning point cloud can be obtained through a three-dimensional laser radar.

[0062] Discretely sample the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points;

[0063] Use the registration algorithm to align the model point cloud with the real scene scanning point cloud posture, and then align the 3D model with the real scene scanning point cloud posture;

[0064] According to the apparent difference between the 3D model after posture alignment and the real scene scanning point cloud, the redundant triangle patches of the 3D model and the missing data points of the 3D model are rendered;

[0065] According to the color attributes of the data points in the real-scene scanning point cloud, the triangular patches of the 3D model are colored to complete the texture correction.

[0066] Embodiment 2:

[0067] This embodiment is further designed on the basis of the first embodiment in that: in this embodiment, the three-dimensional model of the target substation equipment is discretized and sampled using the Poisson-Disk distribution grid sampling method, including the following specific steps:

[0068] 2.1) Use the random uniform sampling algorithm to obtain the initial sample points on each triangular patch of the 3D model; for 3D models with triangular patches of different sizes and shapes, it is necessary to adaptively control the number of sample points generated in each triangular patch area and set it to be proportional to the area of ​​the triangular patch. For the number of initial sample points num on triangular patch i i Satisfies the following formula

[0069]

[0070]

[0071] In the formula, is the area of ​​triangle patch i; S total is the total area of ​​all triangular patches; a i , b i 、c i are the side lengths of triangle patch i; num total is the total number of model samples.

[0072] A sufficient number of initial sample points can satisfy the unbiased sampling requirements and maximize sampling of Poisson-Disk distribution grid sampling.

[0073] 2.2) Use the minimum radius threshold to compare the initial sample points on each triangular patch, select the final sample points on each triangular patch, and use them as data points of the model point cloud to form a model point cloud.

[0074] Embodiment three:

[0075] This embodiment is further designed on the basis of the first embodiment in that: this embodiment adopts an idea similar to the octree grid division to reduce the time complexity of the final sample point screening, and establishes a voxel grid for the three-dimensional model. The side length of the voxel grid is equal to the minimum threshold radius r. The initial sample points that do not meet the threshold radius condition can only be distributed in this grid or adjacent grids, which greatly narrows the screening range. To determine whether the initial sampling point in a grid is a reserved point (final sample point) of the Poisson-Disk distribution, it is only necessary to query itself and the (initial sampling points) in the adjacent grids. If the point distance is greater than the minimum threshold radius, this point is considered to be the final sample point and can be retained. Regarding the definition of the distance between two points, due to the existence of a curved surface on the surface of the three-dimensional model, the shortest distance between two points is the geodesic distance close to the surface. Using the simplest Euclidean distance for calculation will cause errors. Therefore, the present invention adopts an approximate fast calculation method. The specific steps include:

[0076] Select any initial sample point, and calculate the distance between the selected initial sample point and other initial sample points. If the distance between the selected initial sample point and other initial sample points is greater than the set minimum radius threshold, the selected initial sample point is recorded as the final sample point.

[0077] The distance between the selected initial sample point and other initial sample points adopts the geodesic distance of the veneer surface. For the initial sample point p i and other sample points p j The geodesic distance d between G The specific calculation formula is as follows:

[0078]

[0079]

[0080] Where, d E is the Euclidean distance between two points; is the initial sample point p i To other sample points p j The unit direction vector of ; is the normal vector of the sample point; p i represents the coordinates corresponding to the initial sample points; p j Indicates the corresponding coordinates of other sample points;

[0081] Repeat the above steps to determine whether each initial sample point is a final sample point.

[0082] Embodiment 4:

[0083] This embodiment is further designed on the basis of the first embodiment in that: in this embodiment, the specific steps of aligning the model point cloud with the real scene scanning point cloud posture by using the registration algorithm include:

[0084] 3.1) The Teaser++ algorithm is used to align the model point cloud and the real scene scanning point cloud to obtain the posture transformation matrix between the model point cloud and the real scene scanning point cloud;

[0085] 3.2) Use the posture transformation matrix to align the model point cloud posture to the coordinate system of the real scene scanning point cloud.

[0086] Embodiment five:

[0087] This embodiment is further designed on the basis of the first embodiment, the second embodiment, the third embodiment or the fourth embodiment in that: in this embodiment, according to the apparent difference between the three-dimensional model after posture alignment and the real scene scanning point cloud, the specific steps of rendering the redundant triangular facets of the three-dimensional model and the missing data points in the model point cloud include:

[0088] 4.1) Select any triangular patch in the 3D model and randomly obtain N sampling points in the triangular patch;

[0089] 4.2) Perform a field radius search based on the corresponding position of each sampling point in the 3D model in the real scene scan point cloud. If the number of sampling points that can find neighboring points is greater than αN, α∈[0,1], the selected triangular face is not a redundant triangular face; if the number of sampling points that can find neighboring points is less than or equal to αN, the selected triangular face is a redundant triangular face;

[0090] 4.3) Repeat steps 4.1) to 4.2) to determine whether each triangular face in the three-dimensional model is redundant, and render the redundant triangular face;

[0091] 4.4) Perform a field radius search at the corresponding position of each data point of the real scene scanning point cloud in the three-dimensional model. If no neighboring point is found, the data point is considered to be a missing data point in the three-dimensional model, and the corresponding data point is rendered in the real scene scanning point cloud.

[0092] Embodiment six:

[0093] This embodiment is further designed on the basis of the fifth embodiment in that: in this embodiment, the specific steps of coloring and correcting the triangular facets of the three-dimensional model according to the color attributes of the data points in the real scene scanning point cloud include:

[0094] 5.1) Select any triangular patch in the 3D model, obtain five sampling points on the angular bisector of the triangular patch, perform a field radius search based on the corresponding position of each sampling point in the real scene scanning point cloud in the 3D model, obtain 10 neighboring points corresponding to each sampling point, and use the obtained 50 neighboring points as elements in the set to form a neighbor point set E. i; The specific steps for obtaining five sampling points on the angle bisector of the triangular patch are as follows:

[0095] First, calculate the coordinates of the incenter of the triangle according to the coordinates of the triangle vertex, then select the longest line among the three lines connecting the incenter and the vertex, and divide the line into five equal parts. The calculation formula of the coordinates of the incenter of the triangle is as follows:

[0096]

[0097] Where O is the incenter coordinate of the triangle patch; a, b, c are the vertex coordinates of the triangle patch; l a , l b , l c are the side lengths corresponding to vertices a, b, and c respectively;

[0098] 5.2) The neighbor point set E i All neighboring points in are projected onto the plane where the selected triangle patch is located, and the projection points within the triangle patch are screened out to form a projection point set F i ;

[0099] The specific steps for screening out the projection points within the triangle face are as follows:

[0100] Get the neighboring point set E i The corresponding projection points of all neighboring points on the plane where the selected triangle facet is located Calculate the intermediate parameters u and v for the obtained projection points respectively. If the calculated intermediate parameters u and v satisfy 0≤u≤1, 0≤v≤1, u+v≤1 at the same time, the projection point is a projection point inside the triangle patch; otherwise, the projection point is not a projection point inside the triangle patch.

[0101]

[0102]

[0103] In the formula, a, b, and c are the three vertices of the selected triangle patch; They are vector ab, vector ac, and vector

[0104] 5.3) Calculate the projection point set F i The average value of the color attributes of the neighboring points corresponding to each projection point in is used as the color attribute of the selected triangle patch.

[0105] 5.4) Repeat steps 5.1) to 5.3) to color each triangular face in the three-dimensional model to complete texture correction.

[0106] Embodiment seven:

[0107] The present invention discloses an appearance checking and correcting device for a three-dimensional model of a substation equipment. The three-dimensional model is composed of a plurality of triangular facets. The appearance checking and correcting device comprises:

[0108] A real-scene scanning point cloud acquisition module is used to acquire a real-scene scanning point cloud composed of a number of data points with color attributes for a target substation;

[0109] A discrete sampling module is used to discretize the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points;

[0110] The posture alignment module is used to align the model point cloud with the real scene scanning point cloud posture by using the registration algorithm, and then align the three-dimensional model with the real scene scanning point cloud posture;

[0111] An appearance difference rendering module is used to render redundant triangular facets of the 3D model and missing data points in the 3D model according to the appearance difference between the 3D model after posture alignment and the real scene scanning point cloud;

[0112] The texture correction module is used to color the triangular facets of the three-dimensional model according to the color attributes of the data points in the real-scene scanning point cloud to complete the texture correction.

[0113] Embodiment eight:

[0114] An electronic device includes a memory and a processor. The memory stores a computer program. The processor is used to call and run the computer program stored in the memory to execute the above method.

[0115] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0116] Application examples:

[0117] In this example, the appearance check and correction method of the three-dimensional model of the substation equipment of the present invention is used to perform appearance check and correction on the arrester in a substation. The schematic diagram of the three-dimensional model structure of the arrester is shown in Figure 2(a). The appearance check and correction method specifically includes the following steps:

[0118] The real-scene scanning point cloud of the lightning arrester consisting of several data points with color attributes is obtained through the three-dimensional laser radar; in this example, the real-scene scanning point cloud of the entire substation is obtained, and then the real-scene scanning point cloud of the lightning arrester is segmented. The schematic diagram of the segmented real-scene scanning point cloud of the lightning arrester is shown in Figure 2(b).

[0119] Discrete sampling is performed on the three-dimensional model of the arrester to obtain a model point cloud composed of a number of data points;

[0120] The registration algorithm is used to align the model point cloud with the real scene scanning point cloud, and then the arrester 3D model is aligned with the real scene scanning point cloud. The posture alignment result is shown in Figure 2(c).

[0121] According to the apparent difference between the 3D model of the arrester after posture alignment and the real scene scanning point cloud, model differentiation is performed to render the redundant triangular faces of the 3D model of the arrester and the missing data points of the 3D model. Figure 3(a) and Figure 3(b) are the model differential rendering results under two perspectives.

[0122] According to the color attributes of the data points in the real-scene scanning point cloud, the triangular facets of the arrester 3D model are colored to complete the texture correction. The schematic diagram of the color scanning point cloud of the arrester is shown in Figure 4(a), and the schematic diagram of the textureless 3D model of the arrester is shown in Figure 4(b). Figures 4(c) and 4(d) are the coloring results under two viewing angles.

Claims

1. A method for checking and correcting the appearance of a three-dimensional model of a substation equipment, wherein the three-dimensional model is composed of a plurality of triangular facets, characterized in that: The appearance verification and correction method comprises: Obtain a real-scene scanning point cloud of the target substation equipment, which is composed of a number of data points with color attributes; Discretely sample the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points; Using a registration algorithm to align the model point cloud with the real scene scanning point cloud posture, thereby aligning the three-dimensional model with the real scene scanning point cloud posture; Rendering redundant triangular facets of the three-dimensional model and missing data points of the three-dimensional model according to the apparent difference between the three-dimensional model after posture alignment and the real scene scanning point cloud; According to the color attributes of the data points in the real scene scanning point cloud, the triangular facets of the three-dimensional model are colored to complete the texture correction; The specific steps of rendering redundant triangular facets of the 3D model and missing data points of the 3D model according to the apparent difference between the 3D model after posture alignment and the real scene scanning point cloud include: 4.1) Select any triangular patch in the three-dimensional model and randomly obtain N sampling points in the triangular patch; 4.2) Perform a field radius search based on the corresponding position of each sampling point in the 3D model in the real scene scan point cloud. If the number of sampling points that can find neighboring points is greater than αN, α∈[0,1], the selected triangular face is not a redundant triangular face; if the number of sampling points that can find neighboring points is less than or equal to αN, the selected triangular face is a redundant triangular face; 4.3) Repeating steps 4.1) to 4.2), determining whether each triangular face in the three-dimensional model is redundant, and rendering the redundant triangular face; 4.4) Perform a field radius search based on the corresponding position of each data point of the real scene scanning point cloud in the 3D model. If no neighboring point is found, the data point is considered to be a missing data point in the 3D model, and the corresponding data point is rendered in the real scene scanning point cloud; The specific steps of performing color correction processing on the triangular facets of the three-dimensional model according to the color attributes of the data points in the real scene scanning point cloud include: 5.1) Select any triangular patch in the three-dimensional model, obtain five sampling points on the angular bisector of the triangular patch, perform a field radius search based on the corresponding position of each sampling point in the three-dimensional model in the real scene scanning point cloud, obtain 10 neighboring points corresponding to each sampling point, and use the obtained 50 neighboring points as elements in the set to form a neighbor point set E i ; 5.2) The neighbor point set E i All neighboring points in are projected onto the plane where the selected triangle patch is located, and the projection points within the triangle patch are screened out to form a projection point set F i ; 5.3) Calculate the projection point set F i The average value of the color attributes of the neighboring points corresponding to each projection point in the image is used as the color attribute of the selected triangle patch; 5.4) Repeat steps 5.1) to 5.3) to color each triangular face in the three-dimensional model to complete texture correction.

2. The method for checking and correcting the appearance of the three-dimensional model of a substation equipment according to claim 1 is characterized in that: The discretization sampling is performed using the Poisson-Disk distribution grid sampling method, including the following specific steps: 2.1) Using a random uniform sampling algorithm to obtain initial sample points on each triangular face of the 3D model; 2.2) Use the minimum radius threshold to compare the initial sample points on each triangular patch, select the final sample points on each triangular patch, and use them as data points of the model point cloud to form a model point cloud.

3. The method for checking and correcting the appearance of the three-dimensional model of a substation equipment according to claim 2 is characterized in that: The specific steps of using the minimum radius threshold to compare the initial sample points on each triangular facet and screen out the final sample points on each triangular facet include: 2.2.1) Select an initial sample point and calculate the distance between the selected initial sample point and other initial sample points. If the distance between the selected initial sample point and other initial sample points is greater than the set minimum radius threshold, the selected initial sample point is recorded as the final sample point; The distance between the selected initial sample point and other initial sample points adopts the geodesic distance of the veneer surface. For the initial sample point p i and other sample points p j The geodesic distance d between G The specific calculation formula is as follows: Where, d E is the Euclidean distance between two points; is the initial sample point p i To other sample points p j The unit direction vector of ; is the normal vector of the sample point; 2.2.2) Repeat step 2.2.1) and determine whether each initial sample point is a final sample point.

4. The method for checking and correcting the appearance of the three-dimensional model of a substation equipment according to claim 1 is characterized in that: The specific steps of using the registration algorithm to align the model point cloud with the real scene scanning point cloud posture include: 3.1) Performing model registration on the model point cloud and the real scene scanning point cloud using the Teaser++ algorithm to obtain a posture transformation matrix between the model point cloud and the real scene scanning point cloud; 3.2) Using the posture transformation matrix, the posture of the model point cloud is aligned to the coordinate system of the real scene scanning point cloud.

5. The method for checking and correcting the appearance of the three-dimensional model of a substation equipment according to claim 1 is characterized in that: The real-scene scanning point cloud with color attributes of the target substation equipment is acquired through a three-dimensional laser radar.

6. A device for checking and correcting the appearance of a three-dimensional model of a substation equipment, wherein the three-dimensional model is composed of a plurality of triangular facets, characterized in that: The appearance checking and correction device comprises: A real-scene scanning point cloud acquisition module is used to acquire a real-scene scanning point cloud composed of a number of data points with color attributes for a target substation; A discrete sampling module is used to discretize the three-dimensional model of the target substation equipment to obtain a model point cloud composed of a number of data points; A posture alignment module, used to align the model point cloud with the real scene scanning point cloud using a registration algorithm, thereby aligning the three-dimensional model with the real scene scanning point cloud; An appearance difference rendering module is used to render redundant triangular facets of the three-dimensional model and missing data points in the three-dimensional model according to the appearance difference between the three-dimensional model after posture alignment and the real scene scanning point cloud; The texture correction module is used to color the triangular facets of the three-dimensional model according to the color attributes of the data points in the real scene scanning point cloud to complete the texture correction; the specific steps of rendering the redundant triangular facets of the three-dimensional model and the missing data points of the three-dimensional model according to the apparent difference between the three-dimensional model after posture alignment and the real scene scanning point cloud include: 4.1) Select any triangular patch in the three-dimensional model and randomly obtain N sampling points in the triangular patch; 4.2) Perform a field radius search based on the corresponding position of each sampling point in the 3D model in the real scene scan point cloud. If the number of sampling points that can find neighboring points is greater than αN, α∈[0,1], the selected triangular face is not a redundant triangular face; if the number of sampling points that can find neighboring points is less than or equal to αN, the selected triangular face is a redundant triangular face; 4.3) Repeating steps 4.1) to 4.2), determining whether each triangular face in the three-dimensional model is redundant, and rendering the redundant triangular face; 4.4) Perform a field radius search based on the corresponding position of each data point of the real scene scanning point cloud in the 3D model. If no neighboring point is found, the data point is considered to be a missing data point in the 3D model, and the corresponding data point is rendered in the real scene scanning point cloud; The specific steps of performing color correction processing on the triangular facets of the three-dimensional model according to the color attributes of the data points in the real scene scanning point cloud include: 5.1) Select any triangular patch in the three-dimensional model, obtain five sampling points on the angular bisector of the triangular patch, perform a field radius search based on the corresponding position of each sampling point in the three-dimensional model in the real scene scanning point cloud, obtain 10 neighboring points corresponding to each sampling point, and use the obtained 50 neighboring points as elements in the set to form a neighbor point set E i ; 5.2) The neighbor point set E i All neighboring points in are projected onto the plane where the selected triangle patch is located, and the projection points within the triangle patch are screened out to form a projection point set F i ; 5.3) Calculate the projection point set F i The average value of the color attributes of the neighboring points corresponding to each projection point in the image is used as the color attribute of the selected triangle patch; 5.4) Repeat steps 5.1) to 5.3) to color each triangular face in the three-dimensional model to complete texture correction.

7. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images