A complex surface reconstruction method for bridge digital twin models based on 3D point clouds

Abstract

The present invention discloses a method for reconstructing complex surfaces of a bridge digital twin model based on a three-dimensional point cloud, comprising three steps: meshing the bridge three-dimensional point cloud, extracting type value points based on the meshing, and constructing a complex surface of the bridge digital twin model based on the type value points. The meshing of the three-dimensional point cloud specifically determines the number of basic type value points based on the average density and average curvature of the point cloud, and then performs corresponding number encryption according to the change in the local curvature of the point cloud, and ensures the number of type value points required for spline surface interpolation. The type value point extraction based on the meshing can accurately capture the edges and corners of the point cloud, so that the selected type value points are both comprehensive and representative. The present invention can not only improve the accuracy of complex surface reconstruction in the bridge digital twin model, but also help improve the efficiency of the modeling process.

Description

Technical Field

[0001] The present invention relates to the fields of digitalization of the construction industry and point cloud processing, and in particular to a method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds. Background Art

[0002] The continuous advancement of laser scanning technology enables the acquisition of increasingly detailed and accurate 3D point cloud data. This point cloud data has widespread applications in bridge engineering and computer-aided design. It can not only be used to detect deformation and settlement of bridge structures, but also to construct geometric digital twin models of bridges. These models, leveraging the large data volume and high precision of point clouds, can reflect the service status of bridge structures in real time, effectively improving the accuracy of structural safety assessments and damage predictions.

[0003] In computer-aided design, spline interpolation technology is often used to create surfaces in geometric digital twin models. The points used for surface interpolation are called type points. These type points are carefully selected from the point cloud of the corresponding component's outer surface and must be able to represent the shape characteristics of the entire point cloud. Therefore, the selection of type points needs to take into account both comprehensiveness and representativeness. Comprehensiveness requires that the overall layout of the type points evenly cover every area of ​​the point cloud to ensure the integrity of the surface; representativeness requires that the type points must also fully capture the edges of the point cloud and be encrypted at locations where the curvature changes significantly to ensure the continuity and smoothness of the surface.

[0004] Although the existing methods for selecting type value points take into account the requirement of uniform distribution, they often ignore the fine processing of the edge of the point cloud and the area of ​​sudden change in curvature. For the arched components of long-span bridges in service, the complexity of the structural shape caused by stress fatigue is very common. If only the point cloud is uniformly sampled, the subtle local deformation of the component may be ignored, which will lead to inaccurate model. In order to improve the accuracy of the model, the number of type value points can only be increased globally, but this will significantly increase the computational burden. In summary, the traditional uniform sampling method has limitations in flexibility and is not suitable for high-precision geometric digital twin model reconstruction of complex-shaped bridge components. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds. The method first determines the number of basic type value points based on the average density and average curvature of the point cloud, and then encrypts the number accordingly based on the changes in the local curvature of the point cloud, thereby ensuring that the selected type value points are both comprehensive and representative.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A method for reconstructing complex curved surfaces of a bridge digital twin model based on three-dimensional point clouds includes the following steps:

[0008] Step 1: Meshing of the bridge 3D point cloud;

[0009] S101: Calculate the average density a_d and average curvature a_c of the bridge 3D point cloud, and align the three main directions of the 3D point cloud with the X, Y, and Z directions of the Cartesian coordinate system; obtain the maximum and minimum values ​​x of the X and Y coordinates of the aligned point cloud. max 、x min 、y max 、y min , and project all points of the 3D point cloud onto the XOY plane; the projected points retain their Z coordinate value and curvature value information in the 3D space;

[0010] S102: Calculate the side length l of the initial grid, generate a uniform grid G, and assign each point of the projected point cloud to the corresponding grid;

[0011] S103: traverse each grid in the grid G ​​and calculate the average curvature of all points in the current grid; if the average curvature obtained exceeds a set average curvature threshold, subdivide all grids in the same X direction or the same Y direction as the grid;

[0012] S104: traverse the grid G ​​again, delete all empty grids that do not contain projection points, and subdivide the adjacent grids of the empty grids until no empty grids are left, thus completing the grid division of the three-dimensional point cloud;

[0013] Step 2: Extraction of type value points based on grid division;

[0014] If there are corner points of the projected point cloud in the grid, the corner points are directly selected as the shape value points;

[0015] If there are no corner points of the projected point cloud in the grid, but there are edge points, then the distance coordinates of all edge points are selected. The nearest point is the type value point; are the average values ​​of the x, y, and z coordinates of all points in the grid, respectively;

[0016] If there are neither corner points nor edge points in the grid, the distance coordinate is selected The nearest point is the type value point;

[0017] Step 3: Based on the type value points, construct the complex surface of the bridge digital twin model.

[0018] Furthermore, the three main directions of the three-dimensional point cloud of the bridge are obtained by principal component analysis, which are the first main direction, the second main direction, and the third main direction; the first main direction is aligned with the Y axis, the second main direction is aligned with the X axis, and the third main direction is aligned with the Z axis.

[0019] Furthermore, the initial side length of the grid is l = a_d / a_c, the number of grids in the X direction of the grid G ​​is a, the number of grids in the Y direction is b, and a = ceil((x max –x min ) / l), b=ceil((y max –y min ) / l), ceil() represents the rounding up function.

[0020] Furthermore, the coordinates of the point in the three-dimensional point cloud of the bridge are [x, y, z], and the corresponding grid after being projected onto the XOY plane is G{n x ,n y}, where n x =ceil((x–x min ) / l), n y =ceil((y–y min ) / l).

[0021] Furthermore, the step S103 specifically includes the following sub-steps:

[0022] S1031: Calculate the average curvature of all points in the current grid G{i,j}

[0023] S1032: Determine whether the >average curvature threshold thr; if yes, proceed to S1033, otherwise proceed to S1035;

[0024] S1033: Divide the Y-direction lengths of all grids located in the same X direction as the grid G{i,j} into two equal parts; Divide the X-direction lengths of all grids located in the same Y direction as the grid G{i,j} into two equal parts;

[0025] S1034: Update the size of the grid G ​​and set a=a+1, b=b+1;

[0026] S1035: Traverse the next grid.

[0027] Furthermore, in S103 and S104, the method for traversing the grid G{i,j} is as follows: starting from G{1,1}, the value of i is increased by 1 after each traversal, and j=1 remains unchanged until G{a,1}, completing the traversal of the first column of grids; then starting from G{1,2}, the value of i is continued to increase from 1 to a, and j=2 remains unchanged, completing the traversal of the second column of grids; and so on, until when i=a and j=b, the traversal ends at the grid G{a,b}.

[0028] Furthermore, in S103, the average curvature threshold is 1.1 to 1.4 times the average curvature a_c.

[0029] Furthermore, in S104, all empty grids that do not contain projection points are deleted, and the adjacent grids of the empty grids are subdivided, specifically:

[0030] After deleting a single grid or n consecutive grids in the same X direction, subdivide the adjacent grids according to the following rules:

[0031] If a single grid or n consecutive grids are at the beginning or end of all grids in that direction, the X-direction length of the first or last grid after deletion is divided into n+1 equal parts; otherwise, the X-direction length of the last grid in the first segment and the first grid in the second segment after deletion is divided into p equal parts and q equal parts respectively; if n is an odd number, then p = (n+3) / 2, q = (n+1) / 2; if n is an even number, then p = q = (n+2) / 2.

[0032] Furthermore, the calculation formulas for the average density a_d and average curvature ac of the three-dimensional point cloud of the bridge are as follows:

[0033] a_d=sum(D) / t

[0034] a_c=sum(C) / t

[0035] Where D is the distance from any point in the 3D point cloud of the bridge to its nearest point; C is the curvature value of any point in the 3D point cloud of the bridge, and t is the total number of points in the point cloud.

[0036] A device for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds includes one or more processors for implementing a method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds.

[0037] The beneficial effects of the present invention are as follows:

[0038] 1. In the three-dimensional point cloud mesh division step of the method of the present invention, the point cloud is first evenly divided according to the average density and average curvature of the entire point cloud, so that the overall layout of the type value points can cover every area of ​​the point cloud, thereby ensuring the comprehensiveness of the type value point selection.

[0039] 2. The 3D point cloud meshing step of the present invention not only performs uniform division but also appropriately subdivides the point cloud based on its local curvature, allowing the densification of type value points in local locations where the point cloud shape changes significantly. Furthermore, the type value point extraction step of the present invention fully captures the edges and corners of the point cloud, ensuring the representativeness of the selected type value points.

[0040] 3. The method of the present invention ensures the requirement of a×b type value points for spline surface interpolation, and has grouped the type value points in the u direction (corresponding to the X direction of grid G) and the v direction (corresponding to the Y direction of grid G) of the spline surface through grid division, making the subsequent interpolation surface generation more convenient and improving the efficiency of the entire three-dimensional model reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of the complex surface reconstruction method of the bridge digital twin model based on three-dimensional point cloud of the present invention.

[0042] Figure 2 is a three-dimensional point cloud of one embodiment, wherein T1 is the first main direction, T2 is the second main direction, and T3 is the third main direction.

[0043] Figure 3 It is the projection of the 3D point cloud on the XOY plane, where L1, L2, L3, and L4 are the outer contours of the point cloud, and E1, E2, E3, and E4 are the corner points of the point cloud.

[0044] Figure 4 is the initial grid division result, where G is the grid, G{1,1} is the grid in the grid G, a is the number of grids in the X direction, b is the number of grids in the Y direction, and l is the side length of each grid.

[0045] Figure 5 It is the specific process of mesh subdivision in the mesh division step of three-dimensional point cloud.

[0046] Figure 6 This is the result of the subdivision of grid G. The grid shown in bold in the figure is divided into four equal parts.

[0047] Figure 7 This is the result of deleting the empty grids in grid G. The diagonal lines in the figure show the empty grids in grid G, and the bold ones show the subdivided grids.

[0048] Figure 8 This is the result of type value point selection based on grid G. The points displayed in bold in the figure are type value points.

[0049] Figure 9 It is a three-dimensional perspective of the type value point selection results. The points displayed in bold in the figure are type value points. DETAILED DESCRIPTION

[0050] The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments, and the purpose and effects of the present invention will become more apparent. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0051] like Figure 1 As shown, the complex surface reconstruction method of the bridge digital twin model based on three-dimensional point cloud of this embodiment includes three steps: meshing of the bridge three-dimensional point cloud, extraction of type value points based on meshing, and construction of the complex surface of the bridge digital twin model based on type value points.

[0052] The meshing of the bridge 3D point cloud in step 1 specifically includes the following sub-steps:

[0053] S101: Calculate the average density a_d and average curvature a_c of the bridge 3D point cloud, and align the three main directions of the 3D point cloud with the X, Y, and Z directions of the Cartesian coordinate system; obtain the maximum and minimum values ​​x of the X and Y coordinates of the aligned point cloud. max 、x min 、y max 、y min , and project all points of the 3D point cloud onto the XOY plane; the projected points retain their Z coordinate value and curvature value information in the 3D space;

[0054] S102: Calculate the side length l of the initial grid, generate a uniform grid G, and assign each point of the projected point cloud to the corresponding grid;

[0055] S103: traverse each grid in the grid G ​​and calculate the average curvature of all points in the current grid; if the average curvature obtained exceeds a set average curvature threshold, subdivide all grids in the same X direction or the same Y direction as the grid;

[0056] S104: traverse the grid G ​​again, delete all empty grids that do not contain projection points, and subdivide the adjacent grids of the empty grids until no empty grids appear, thus completing the grid division of the three-dimensional point cloud.

[0057] like Figure 2As shown in the figure, the preferred three-dimensional point cloud data of this embodiment is listed. In this embodiment, the average density a_d of the three-dimensional point cloud is used to characterize the overall density of the point cloud, in millimeters, and is calculated by the formula a_d=sum(D) / t. In the formula, D is the distance from any point in the point cloud to its nearest point, and t is the total number of points in the point cloud. The larger the value of the average density a_d, the sparser the point cloud, and vice versa. The average curvature a_c of the three-dimensional point cloud is used to characterize the degree of curvature of the overall shape of the point cloud, and is dimensionless, and is calculated by the formula a_c=sum(C) / t. In the formula, C is the curvature value of any point in the point cloud, and t is the total number of points in the point cloud. The point curvature value is used to characterize the local shape of the point cloud near a certain point in it, and is calculated by the principal curvature estimation class (PrincipalCurvaturesEstimation) in the point cloud library (PCL). The larger the value of the average curvature a_c, the more curved the shape of the point cloud, and vice versa.

[0058] In this embodiment, the three main directions (T1, T2, T3) of the three-dimensional point cloud are obtained by principal component analysis (PCA) to characterize the density of the point cloud in a single direction. Figure 2 As shown, the point cloud distribution in the first principal direction T1 is the sparsest, resulting in the most elongated shape. The point cloud distribution in the third principal direction T3 is the densest, resulting in the narrowest shape. The second principal direction T2 is perpendicular to the first and third principal directions T1 and T3, and the first and third principal directions T1 are also perpendicular to each other. In this embodiment, by rotating the point cloud, the first principal direction T1 is aligned with the Y-axis, the second principal direction T2 is aligned with the X-axis, and the third principal direction T3 is aligned with the Z-axis.

[0059] See Figure 3 , this figure lists the projection of the three-dimensional point cloud of this embodiment on the XOY plane. For any point in the three-dimensional point cloud, assuming that its coordinates are [x, y, z], the coordinates of the projection of the point on the XOY plane are [x, y]. It should be noted that the projected point retains its Z coordinate value and curvature value information in the three-dimensional space. As shown in the figure, the curves L1, L2, L3, and L4 constitute the outer contour of the three-dimensional point cloud. The points located on the outer contour lines L1, L2, L3, and L4 are marked as edge points of the point cloud (shown in bold in the figure), and the intersection points E1, E2, E3, and E4 of the four outer contour lines are marked as corner points of the three-dimensional point cloud. In this embodiment, the edge points and corner points of the three-dimensional point cloud are identified and marked by the edge estimation class (BoundaryEstimation) in the PCL library.

[0060] In this embodiment, the total number of points in the 3D point cloud is 16705, the number of edge points is 304, and the edge points include 4 corner points of the point cloud. The calculated result of the average density a_d is 0.3621, and the calculated result of the average curvature a_c is 0.0232. The maximum and minimum values ​​of the X coordinate x max 、x min The maximum and minimum values ​​of the Y coordinate are 58.59 and 2.29 respectively. max 、y min They are 108.73 and 1.41 respectively, in millimeters.

[0061] See Figure 4 The figure shows the uniform grid G ​​of this embodiment. The number of grid cells in the grid G ​​is a in the X direction and b in the Y direction. The size of each grid cell is l×l. Where a=ceil((x max –x min ) / l), b=ceil((y max –y min ) / l), l = a_d / a_c, ceil() represents the rounding function. In this embodiment, the side length of each grid is l = 15.6 mm, so the initial number of grid cells in the grid G ​​is 4×7. For any point projected onto the XOY plane, assuming its coordinates are [x, y], the point will be included in the grid G{n x ,n y}, where n x =ceil((x–x min ) / l), n y =ceil((y–y min ) / l). All points on the XOY plane are summarized to complete the initial meshing of the 3D point cloud.

[0062] See Figure 5 , S103 in the three-dimensional point cloud mesh division specifically includes the following sub-steps:

[0063] S1031: Calculate the average curvature of all points in the current grid G{i,j}

[0064] S1032: Determine whether the >average curvature threshold thr; if yes, proceed to S1033, otherwise proceed to S1035;

[0065] S1033: Halve the Y lengths of all grids in the same X direction as grid G{i,j}. Halve the X lengths of all grids in the same Y direction as grid G{i,j}. Halving the X length means dividing the original l×l grid into two l / 2×l grids, and halving the Y length means dividing the original l×l grid into two l×l / 2 grids. The current grid G{i,j} is then divided into four equal parts.

[0066] S1034: Update the size of the grid G ​​and set a=a+1, b=b+1;

[0067] S1035: Traverse the next grid.

[0068] In this embodiment, the average curvature The calculation formula is Where C{i,j} represents the curvature value of any point in the current grid G{i,j}, and t{i,j} represents the total number of points in the current grid G{i,j}. The average curvature threshold thr is 1.1 to 1.4 times the average curvature a_c. Preferably, in this embodiment, the threshold thr is set to 1.3 and a_c is set to 0.0302. G{i,:} represents all grids in the same Y direction as G{i,j}, and G{:,j} represents all grids in the same X direction as G{i,j}.

[0069] The specific method for traversing the grid G{i,j} in S103 and S104 is: starting from G{1,1}, the value of i is increased by 1 after each traversal, and j=1 remains unchanged until G{a,1}, completing the traversal of the first column of grids; then starting from G{1,2}, the value of i is continued to increase from 1 to a, and j=2 remains unchanged, completing the traversal of the second column of grids; and so on, until when i=a and j=b, the traversal ends at the grid G{a,b}.

[0070] See Figure 6 , which lists the subdivision of the uniform grid G ​​in this embodiment. In this embodiment, when traversing to the grid G{3,4} (shown in bold in the figure), the average curvature of all points in the grid is Exceeds the threshold thr. At this point, the X-direction lengths of all grids in the same Y direction are halved, dividing the original l×l grid into two l / 2×l grids. Simultaneously, the Y-direction lengths of grids G{1,4}, G{2,4}, G{3,4}, and G{4,4} in the same X direction are halved, dividing the original l×l grid into two l×l / 2 grids. After this division, the current grid G{3,4} is divided into four equal parts, and the size of grid G ​​is updated. During the subsequent traversal, the average curvature of the points within grids G{4,5} and G{3,7} also exceeds the threshold thr, and the same subdivision operation is performed. After traversing and subdividing all grids, the number of grid cells in grid G ​​increases from 4×7 to 7×10.

[0071] See Figure 7 , the figure shows the exclusion of empty grids that do not contain projection points in the grid G ​​of this implementation. For the grids G{:,j} in the same X direction, after deleting a single grid or n consecutive grids, the adjacent grids are subdivided according to the following rules:

[0072] (1) If a single grid or n consecutive grids are at the beginning or end of G{:,j}, the X-direction length of the first or last grid after deletion is divided into n+1 equal parts;

[0073] (2) If a single grid or n consecutive grids are in the middle of G{:,j}, the X-direction length of the last grid in the first segment and the first grid in the second segment after deletion is divided into p equal parts and q equal parts respectively. If n is an odd number, then p = (n+3) / 2, q = (n+1) / 2; if n is an even number, then p = q = (n+2) / 2.

[0074] Repeat the above steps until there are no more empty grids in G{:,j}, and then traverse the next column of grids.

[0075] In this embodiment, the grids G{7,1}, G{1,3}, G{1,4}, G{1,5}, G{1,6}, G{1,7}, and G{7,10} do not contain any projection points and are therefore used as empty grids ( Figure 7 The grid G{1,3} is at the beginning of G{:,3} and G{7,1} is at the end of G{:,1}, so the X-axis lengths of the adjacent grids G{2,3} and G{6,1} (shown in bold in the figure) are halved. The remaining empty grids are processed similarly. After all empty grids are eliminated, the size of grid G ​​remains unchanged.

[0076] See Figure 8 and Figure 9, Figure 2 lists the results of type value point selection of the three-dimensional point cloud. Step 2 of this embodiment is the extraction of type value points based on grid division, which specifically includes:

[0077] If there are corner points of the projected point cloud in the grid, the corner points are directly selected as the shape value points;

[0078] If there are no corner points of the projected point cloud in the grid, but there are edge points, then the distance coordinates of all edge points are selected. The nearest point is the type value point; are the average values ​​of the x, y, and z coordinates of all points in the grid, respectively;

[0079] If there are neither corner points nor edge points in the grid, the distance coordinate is selected The nearest point is the type value point.

[0080] in, are the average values ​​of the x, y, and z coordinates of all points in the grid, respectively.

[0081] In the point value extraction method, edge points are points on the outer contour of the point cloud, while corner points are the intersection of two or more outer contour lines of the point cloud. They are a special type of edge point. Both are identified using the BoundaryEstimation class in the PCL library.

[0082] In this example, the 3D point cloud contains 70 shape points (shown in bold in the figure), including 4 corner points and 26 edge points. The resulting shape points are grouped using a grid G: first, grouping them in the x-direction, grouping all shape points within G{:,j} into a single group for interpolation in the u-direction of the spline surface; then, grouping them in the y-direction, grouping all shape points within G{i,:} into a single group for interpolation in the v-direction of the spline surface.

[0083] Corresponding to the aforementioned embodiment of the method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point cloud, the present invention also provides an embodiment of a device for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point cloud.

[0084] An embodiment of the present invention provides a device for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds, comprising one or more processors for implementing the method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point clouds in the above-mentioned embodiment.

[0085] The embodiment of the complex surface reconstruction device of the digital twin model of a bridge based on three-dimensional point cloud of the present invention can be applied to any device with data processing capabilities, and the any device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the memory for execution. From the hardware level, in addition to the processor, memory, network interface, and non-volatile memory, the any device with data processing capabilities in which the device in the embodiment is located can also include other hardware according to the actual function of the any device with data processing capabilities, which will not be described in detail.

[0086] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0087] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present invention. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0088] An embodiment of the present invention also provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the method for reconstructing complex surfaces of a bridge digital twin model based on three-dimensional point cloud in the above embodiment is implemented.

[0089] The computer-readable storage medium may be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium may also be an external storage device, such as a plug-in hard disk, a smart memory card (SmartMedia card, SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of any device with data processing capabilities and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and may also be used to temporarily store data that has been output or is to be output.

[0090] Those skilled in the art will understand that the foregoing descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art will still be able to modify the technical solutions described in the foregoing examples or substitute equivalents for some of the technical features therein. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the invention shall be included within the scope of protection of the invention.

Claims

1. A complex surface reconstruction method for a bridge digital twin model based on three-dimensional point cloud, characterized in that: The steps include: Step 1: Meshing of the bridge 3D point cloud; S101: Calculate the average density a_d and average curvature a_c of the three-dimensional point cloud of the bridge, and align the three main directions of the three-dimensional point cloud with the X, Y, and Z directions of the Cartesian coordinate system; Get the maximum and minimum values ​​of the X and Y coordinates of the aligned point cloud max 、x min 、y max 、y min , and project all points of the 3D point cloud onto the XOY plane; the projected points retain their Z coordinate value and curvature value information in the 3D space; S102: Calculate the side length l of the initial grid, generate a uniform grid G, and assign each point of the projected point cloud to the corresponding grid; S103: traverse each grid in the grid G ​​and calculate the average curvature of all points in the current grid; if the average curvature obtained exceeds a set average curvature threshold, subdivide all grids in the same X direction or the same Y direction as the grid; S104: traverse the grid G ​​again, delete all empty grids that do not contain projection points, and subdivide the adjacent grids of the empty grids until no empty grids are left, thus completing the grid division of the three-dimensional point cloud; Step 2: Extraction of type value points based on grid division; If there are corner points of the projected point cloud in the grid, the corner points are directly selected as the shape value points; If there are no corner points of the projected point cloud in the grid, but there are edge points, then the distance coordinates of all edge points are selected. The nearest point is the type value point; are the average values ​​of the x, y, and z coordinates of all points in the grid, respectively; If there are neither corner points nor edge points in the grid, the distance coordinate is selected The nearest point is the type value point; Step 3: Based on the type value points, construct the complex surface of the bridge digital twin model.

2. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: The three main directions of the three-dimensional point cloud of the bridge are obtained by principal component analysis, namely the first main direction, the second main direction, and the third main direction; the first main direction is aligned with the Y axis, the second main direction is aligned with the X axis, and the third main direction is aligned with the Z axis.

3. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: The initial side length of the grid is l = a_d / a_c. The number of grid cells in the X direction of the grid G ​​is a, and the number of grid cells in the Y direction is b. a = ceil((x max –x min ) / l), b=ceil((y max –y min ) / l), ceil() represents the rounding up function.

4. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: The coordinates of the point in the three-dimensional point cloud of the bridge are [x, y, z], and the corresponding grid after being projected onto the XOY plane is G{n x ,n y }, where n x =ceil((x–x min ) / l), n y =ceil((y–y min ) / l).

5. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: The step S103 specifically includes the following sub-steps: S1031: Calculate the average curvature of all points in the current grid G{i,j} S1032: Determine whether the If yes, proceed to S1033, otherwise proceed to S1035; S1033: Divide the Y-direction lengths of all grids located in the same X direction as the grid G{i,j} into two equal parts; Divide the X-direction lengths of all grids located in the same Y direction as the grid G{i,j} into two equal parts; S1034: Update the size of the grid G ​​and set a=a+1, b=b+1; S1035: Traverse the next grid.

6. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: In S103 and S104, the method for traversing the grid G{i,j} is as follows: starting from G{1,1}, the value of i is increased by 1 after each traversal, and j=1 remains unchanged until G{a,1} is reached, completing the traversal of the first column of grids; then starting from G{1,2}, the value of i is continued to increase from 1 to a, and j=2 remains unchanged, completing the traversal of the second column of grids; and so on, until i=a and j=b, the traversal ends at the grid G{a,b}.

7. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: In the above-mentioned S103, the average curvature threshold is 1.1 to 1.4 times the average curvature a_c.

8. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: In S104, all empty grids that do not contain projection points are deleted, and the adjacent grids of the empty grids are subdivided, specifically: After deleting a single grid or n consecutive grids in the same X direction, subdivide the adjacent grids according to the following rules: If a single grid or n consecutive grids are at the beginning or end of all grids in that direction, the X-direction length of the first or last grid after deletion is divided into n+1 equal parts; otherwise, the X-direction length of the last grid in the first segment and the first grid in the second segment after deletion is divided into p equal parts and q equal parts respectively; if n is an odd number, then p = (n+3) / 2, q = (n+1) / 2; if n is an even number, then p = q = (n+2) / 2.

9. The complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to claim 1 is characterized in that: The calculation formulas for the average density a_d and average curvature a_c of the three-dimensional point cloud of the bridge are as follows: a_d=sum(D) / t a_c=sum(C) / t Where D is the distance from any point in the 3D point cloud of the bridge to its nearest point; C is the curvature value of any point in the 3D point cloud of the bridge; and t is the total number of points in the point cloud.

10. A complex surface reconstruction device for a bridge digital twin model based on three-dimensional point cloud, characterized in that: The method comprises one or more processors for implementing the complex surface reconstruction method of a bridge digital twin model based on three-dimensional point cloud according to any one of claims 1 to 9.

Images