A method for reconstructing large-size part models based on local point cloud information

Abstract

This invention provides a method for reconstructing large-size part models based on local point cloud information, belonging to the field of digital twin technology. The method includes: acquiring a triangular mesh model of the large-size part to be modeled; acquiring a local point cloud data model of the large-size part; aligning the local point cloud data model with the triangular mesh model; moving points within the target region of the triangular mesh model so that these points coincide with matching points within the target region of the local point cloud data model, thus deforming the triangular mesh model; using the Laplace equation, based on the reference region and the target region, obtaining all point cloud information of the deformed triangular mesh model; performing reverse point cloud modeling based on the obtained point information to obtain a high-fidelity model; and using the high-fidelity model for assembly simulation. This method enables model reconstruction based on local point cloud information.

Description

Technical Field

[0001] This invention belongs to the field of digital twin technology, specifically relating to a method for reconstructing large-size part models based on local point cloud information. Background Technology

[0002] As the requirements for assembly performance continue to increase, traditional assembly simulation using triangular mesh models can no longer adequately meet practical needs. During simulation, discrepancies between the actual model and the triangular mesh model can render the simulation results unusable. Therefore, CAD models that more realistically reflect the actual condition of the object are required. Current technology uses a 3D laser scanner to scan the point cloud data of the actual object and performs reverse reconstruction of the point cloud, thereby constructing a twin model that more accurately reflects the actual object.

[0003] However, reverse modeling of point clouds requires a full scan of the object, which consumes a lot of time and storage resources for large parts. For example, large thin-walled skins on aircraft require a full scan of the entire structure, which is resource-intensive and time-consuming, making it difficult to reconstruct the model.

[0004] Currently, existing technologies for reconstructing high-fidelity models from point cloud data primarily employ full scanning to acquire point cloud data of all object surfaces and then using this complete point cloud data to construct the model. However, this approach has the following drawbacks: when dealing with the reverse reconstruction of large parts or objects with extremely complex surfaces, it becomes impossible to scan the entire surface of the object, thus hindering reverse modeling.

[0005] Meanwhile, to improve modeling efficiency, some researchers have proposed a rapid reconstruction method that involves scanning partial feature surface point cloud data, reconstructing the feature surfaces in reverse, and then replacing the corresponding feature surfaces. However, this method has limitations: the replacement of feature surfaces does not involve other non-feature surfaces, so when the model deformation is large, the model after surface replacement will be distorted. Furthermore, for skinning, there may be cases where the feature surfaces are very large, so the above method cannot effectively address these issues. Summary of the Invention

[0006] To overcome the shortcomings of the existing technology, the present invention provides a method for reconstructing large-size part models based on local point cloud information.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for reconstructing large-size part models based on local point cloud information includes:

[0009] Obtain the triangular mesh model of the large part to be modeled;

[0010] Discretize the triangular mesh model to form a triangular mesh model composed of nodes and the lines connecting the nodes.

[0011] Scan the key local features of large-sized parts to obtain local point cloud data models of large-sized parts;

[0012] The local point cloud data model is divided into a reference region and a target region;

[0013] The feature points in the reference region and the target region are matched with the corresponding points in the triangular mesh model. After matching, the pose transformation of the local point cloud data model is performed so that the local point cloud data model coincides with the triangular mesh model.

[0014] Search for the nearest point on the target region of the local point cloud data model and the nearest point on the triangular mesh model to obtain the nearest point index set Mp on the triangular mesh model;

[0015] Match the nearest point index set Mp with the corresponding points in the target region of the local point cloud data model;

[0016] The points located within the target area in the triangular mesh model are moved so that they coincide with the matching points within the target area of ​​the local point cloud data model, thereby deforming the target area of ​​the triangular mesh model.

[0017] Using the Laplace equation, we obtain all point cloud information M after the triangular mesh model is deformed;

[0018] Based on all the obtained point cloud information M, reverse modeling of the point cloud is performed to obtain a high-fidelity model, and the high-fidelity model is used to perform assembly simulation of the parts.

[0019] Furthermore, the local features of point i in the triangular mesh model are:

[0020]

[0021] Where, δ i Let (v) be the local feature matrix of point i in the triangular mesh model. i -v j Let be the vector formed by point i and its connected points.

[0022] Furthermore, the process of dividing the local point cloud data model into a reference region and a target region includes:

[0023] In Geomagic Studio, the scanned local point cloud data model is divided into regions, and the reference region is selected manually.

[0024] By using the reverse selection method, the remaining point cloud data model regions are used as target regions.

[0025] Furthermore, the step of matching feature points in the reference region and target region with corresponding points in the triangular mesh model, and then performing pose transformation on the local point cloud data model to make the local point cloud data model coincide with the triangular mesh model includes:

[0026] Using the feature-based matching function of Geomagic Studio, feature points are manually created. Three feature points are created in the corresponding regions of the reference region, the target region, and the triangular mesh model, respectively. The three feature points are matched to obtain the first transformation matrix.

[0027] The local point cloud data model is transformed into a pose based on the first transformation matrix to obtain the local point cloud data model after the transformation position.

[0028] The reference region, target region and corresponding region of triangular mesh model are matched by ICP matching algorithm to obtain the second transformation matrix;

[0029] The local point cloud data model is transformed in pose according to the second transformation matrix to obtain the local point cloud data model after transformation, thus achieving the overlap between the local point cloud data model and the triangular mesh model.

[0030] Furthermore, the Laplace equation is:

[0031] LM = δ

[0032] Where L is the connection matrix of the triangular mesh model, δ is the local feature matrix in the triangular mesh model, and M is the total point cloud information of the triangular mesh model.

[0033] Furthermore, the solution to the Laplace equation is as follows:

[0034] The coordinate information of the reference area points and the target area points of the deformed triangular mesh model are added to the connection matrix L of the triangular mesh model to obtain the updated connection matrix L'.

[0035] The target region point coordinates of the deformed triangular mesh model are added to the local feature matrix δ to obtain the updated local feature matrix δ'.

[0036] Substitute the updated connection matrix L' and local feature matrix δ' into the Laplace equation to calculate all point cloud information M, and then obtain all point cloud information after the part is deformed.

[0037] Substituting the connectivity matrix L' and the local characteristic matrix δ', the Laplace equation becomes:

[0038]

[0039] in, For the updated connection matrix L', Let δ' be the updated local feature matrix, ω be the weight of the points in the scanned region, and I be the local feature matrix. n×m M is a class identity matrix. n These are the three-dimensional coordinates of the points in the scanned area.

[0040] Furthermore, the complete point cloud information M is:

[0041] M = (L T L) -1 L T δ'

[0042] Among them, L T Let L be the transpose of the connectivity matrix L, and δ′ be the updated local feature matrix.

[0043] Furthermore, the step of performing point cloud inverse modeling based on the obtained point information to obtain a high-fidelity model includes:

[0044] Import all point cloud information M into Geomagic Studio in txt format;

[0045] All point cloud information M is encapsulated, and the mesh problems of self-intersection and sharp objects in the mesh model are removed to obtain the first triangular mesh model.

[0046] Using the Precise Surface module, after constructing the outline of the first triangular mesh model, a surface patch is constructed and then trimmed.

[0047] Construct a grid and fit the grid into a curved surface;

[0048] The surface is converted into a CAD model to achieve the construction of a high-fidelity model.

[0049] The present invention provides a method for reconstructing large-size part models based on local point cloud information, which has the following advantages:

[0050] Beneficial effects:

[0051] This invention proposes a method for reconstructing large-size part models using incomplete point cloud information. The method obtains a local point cloud data model by scanning the local key features of the large-size part, and then combines it with the original triangular mesh model of the large-size part to reconstruct the model. This method can scan only the local key feature positions, thereby greatly reducing the difficulty and time of 3D laser scanning, and also reducing the amount of data to be stored. Attached Figure Description

[0052] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a flowchart of a method for reconstructing a large-size part model based on local point cloud information according to an embodiment of the present invention;

[0054] Figure 2 This is a partial schematic diagram of the triangular mesh model according to an embodiment of the present invention;

[0055] Figure 3 This is a three-dimensional model of a barrel segment according to an embodiment of the present invention;

[0056] Figure 4 This is a schematic diagram of the discretization of the bucket segment model according to an embodiment of the present invention;

[0057] Figure 5 This is a schematic diagram illustrating the division of the reference region and the target region in an embodiment of the present invention;

[0058] Figure 6 This is a schematic diagram of a deformed discrete point cloud model according to an embodiment of the present invention. Detailed Implementation

[0059] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0060] Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. In the description of this invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "connected" or "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. In the description of this invention, unless otherwise stated, "a plurality of" means two or more, which will not be elaborated further here.

[0061] Example:

[0062] This invention provides a method for reconstructing large-size part models based on local point cloud information, specifically as follows: Figure 1The process includes: acquiring a triangular mesh model of the large-sized part to be modeled; discretizing the triangular mesh model to form a triangular mesh model composed of nodes and connections between nodes; scanning the local key features of the large-sized part to acquire a local point cloud data model; dividing the local point cloud data model into a reference region and a target region; matching the feature points in the reference region and the target region with the corresponding points in the triangular mesh model, and then performing pose transformation on the local point cloud data model to make the local point cloud data model coincide with the triangular mesh model; and searching for points in the target region of the local point cloud data model that match the triangular mesh model. The nearest point is obtained by finding the nearest point index set Mp on the triangular mesh model; the nearest point index set Mp is matched with the corresponding points in the target region of the local point cloud data model; the points in the triangular mesh model located in the target region are moved so that the points in the triangular mesh model located in the target region coincide with the matching points in the target region of the local point cloud data model, thereby deforming the target region of the triangular mesh model; using the Laplace equation, the complete point cloud information M after the triangular mesh model is deformed is obtained; point cloud inverse modeling is performed based on the obtained complete point cloud information M to obtain a high-fidelity model, and the high-fidelity model is used to perform assembly simulation of the parts.

[0063] The following are detailed descriptions of the implementation of this invention:

[0064] S1. Extract the surface geometric information of the theoretical design model, discretize the triangular mesh model, and form a triangular mesh model composed of nodes and the lines connecting the nodes.

[0065] Specifically, step S1 includes:

[0066] S101. Triangulation of a 3D triangular mesh model: Discretize the model surface into a triangular mesh and store the mesh point set M = {m1, m2, ..., m}. n The set of lines connecting points E = {e1, e2, ..., e} m Information from}

[0067] S102. Based on the obtained grid point set M and the set of lines connecting the points E, this invention uses the vector sum of a point i and its connected point j as the local feature δ of that point. i The local features of this point can be represented as:

[0068]

[0069] δ i =∮(v i -v)dl(v)#(2)

[0070] When the position of this point changes, the corresponding vector sum remains unchanged, and the local feature δ is trusted. i This property remains unchanged, such as Figure 2Based on the above, when there are enough points around a point and the distance between the point and the points around the point approaches zero, equation (1) can be transformed into equation (2), where l(v) is the perimeter of the neighborhood approaching point i.

[0071] S2. Divide the scanned local point cloud data into regions to obtain the reference region and the target region.

[0072] Specifically, step S2 includes:

[0073] S201. In Geomagic Studio, the scanned local point cloud data is divided into regions. By manually selecting, the reference region is selected and saved in the reference set C. The reference region C is used to match the triangular mesh model to determine that the point cloud and the model position coincide.

[0074] S202. By using the reverse selection method, the remaining data is used as the target area. At the same time, there can be multiple target areas, which are used to match with the triangular mesh model to determine the overlap between the point cloud and the model position. The multiple target area sets are stored in set B.

[0075] S3. Match the reference area to make the point cloud data coincide with the triangular mesh model.

[0076] Specifically, step S3 includes:

[0077] S301. Using feature-based matching in Geomagic Studio, perform a coarse match between the reference region and the triangular mesh model. Create three feature points manually in the corresponding regions of both the reference region and the triangular mesh model, ensuring that these three points are roughly identical.

[0078] S302. Based on the transformation matrix obtained in step S301, the scanned local point cloud data is transformed in pose according to this matrix to obtain the point cloud data after transformation.

[0079] S303. Achieve precise matching of the reference region using the ICP matching algorithm. Perform ICP matching between the reference region and the triangular mesh model to obtain the transformation matrix.

[0080] S304. Similar to step S302, based on the transformation matrix obtained in step S303, the local point cloud data is transformed to achieve the matching process between the point cloud and the triangular mesh model in the reference area.

[0081] S4. Based on the target area, perform coordinate transformation on the corresponding area of ​​the triangular mesh model to obtain the deformed position;

[0082] Specifically, step S4 includes:

[0083] S401, the same as steps S301 and S303, performs coarse matching and matching between the target area and the triangular mesh model, so that the position of the target area is transformed to the corresponding area of ​​the triangular mesh model.

[0084] S402. Based on the relatively uniform nature of point cloud data, search for the nearest point between the target area and the triangular mesh model after the position transformation, and obtain the nearest point index set Mp.

[0085] S403. Match the nearest point set obtained in step S402 with the target region to obtain the position information of the target region of the triangular mesh model.

[0086] S5. Based on the Laplace mesh transformation, preserve the local features of the part, and obtain all point information of the part after deformation according to the reference area and the target area.

[0087] Specifically, step S5 includes:

[0088] S501. Based on equation (1) in step S102, construct the Laplace equation as follows:

[0089] LM=δ#(3)

[0090] L is the connectivity matrix of the triangular mesh model, used to preserve the relationships between points, where the Laplace equation for the i-th point is as follows:

[0091] L i M = δ i #(4)

[0092] Equation (4) is equivalent to Equation (1), and the coefficients of the i-th row of the connection matrix L can be determined from this. The complete connection matrix L can be obtained based on the point set M and the line set E.

[0093] Specifically:

[0094] by Figure 2 For example, the following matrix is Figure 2 The connection matrix L is such that v2 is connected to v1 and v3. According to equation (1), the coefficient of v2 is 1, while the coefficients of v1 and v3 are both 1. The information corresponding to all points is obtained in sequence, and the connection relationship matrix L is constructed.

[0095]

[0096] S502. The present invention assumes that the local features of the triangular mesh model and the measured model do not change, the local feature matrix δ does not change, and the connection relationship matrix L of the model does not change. Therefore, the deformed point cloud data M also satisfies equation (3).

[0097] S503. The connection matrix L is a singular matrix, therefore it cannot solve for the deformed point cloud data. To make equation (3) solvable, rows need to be added to the connection matrix to satisfy the requirement of full row rank. The local point cloud information scanned above is added to matrix L.

[0098] Specifically, the reference region information and target region information are added to matrix L.

[0099] by Figure 2 For example, if scanning the point cloud reveals a change in v1, then corresponding information needs to be added to L to map the actual coordinates of that point to the local feature matrix δ. Therefore, row information needs to be added to L, which includes the corresponding point index position coefficient of 1 and the rest of the positions of 0, as shown in the matrix below:

[0100]

[0101] Simultaneously, the original local feature matrix δ is supplemented by adding the deformed data of the scanned points to the local feature matrix. The reference region points are the original coordinate values ​​of the triangular mesh model, and the target region points are the deformed values ​​obtained after scanning, resulting in the added local feature matrix δ', whose expression is as follows:

[0102]

[0103] Where M c and M b It is an n×3 matrix containing information about the corresponding points after deformation.

[0104] Substituting the connectivity matrix L' and the local characteristic matrix δ', the Laplace equation becomes:

[0105]

[0106] in, For the updated connection matrix L', Let δ' be the updated local feature matrix, ω be the weight of the points in the scanned region, and I be the local feature matrix. n×m M is a class identity matrix. n These are the three-dimensional coordinates of the points in the scanned area.

[0107] S504. Add rows to the connection relation matrix L through step S503. The resulting matrix L is a column full rank matrix. The system of equations generally becomes an overdetermined system of equations. In order to solve equation (3), both sides of the system of equations need to be multiplied by the transpose of the connection relation matrix L to obtain equation (6), and finally obtain equation (7) of M.

[0108] (L T L)M=L T δ′#(6)

[0109] M = (LT L) -1 L T δ'#(7)

[0110] S6. Perform reverse modeling of the point cloud based on the obtained point information to obtain a high-fidelity model.

[0111] Specifically, step S6 includes:

[0112] S601. Transfer the deformed point cloud data obtained in S504 into Geomagic Studio in txt format, encapsulate the point cloud data, and perform mesh diagnosis to remove mesh problems such as self-intersection and sharp objects from the mesh model to obtain a triangular mesh model.

[0113] S602. Enter the Precise Surface module, construct the outline, then construct the surface patch and trim the surface patch, then construct the grid, and finally fit the grid into a surface.

[0114] S603. Convert the surface obtained from reverse modeling into a CAD model to reconstruct the model with incomplete point cloud information.

[0115] The following are embodiments of the present invention:

[0116] Step 1: Identify the key feature areas of large, easily deformable parts, such as fixed parts and mating parts. Scan the corresponding areas with a 3D laser scanner to obtain point cloud files in .txt format.

[0117] Step 2: Discretize the model. In CATIA V5R20, use the meshing tool in CATIA's STL rapid prototyping module to set the spacing between points, such as... Figure 3 The three-dimensional model of the bucket segment is generated, and the STL model is exported to realize the discretization of the triangular mesh model.

[0118] Step 3: Read the binary STL model in MATLAB and store the point and edge information of the triangular mesh model in the point set M and the set of connections between points E. The point set M is an n×3 matrix, where n is the number of points, and each row contains the coordinate information of a point; the set of connections between points E is an m×2 matrix, where m is the number of edges between all points, and each row contains the endpoint indices of an edge.

[0119] Step 4: Divide the scanned point cloud data into reference and target regions. Use the reverse engineering software Geomagic Studio to divide the point cloud. Manually store the point cloud data from different regions into their respective regions to obtain... Figure 5 The black area represents the target area for both regions, while the red area represents the reference area.

[0120] Step 5: In Geomagic Studio software, create three-point features for the reference region and the corresponding region of the triangular mesh model. Use feature-based matching to roughly align the reference region point cloud with the corresponding region of the triangular mesh model, achieving a coarse match for the reference region. Then, use the ICP matching algorithm to perform a fine match between the two parts, obtaining the transformation matrix, and then move the entire point cloud data.

[0121] Step 6: Match the target region with the triangular mesh model using the ICP algorithm, and solve for the set of closest points between the matched target region and the theoretical triangular mesh model. Then, perform ICP matching between the obtained set of closest points and the target region to realize the coordinate changes of the triangular mesh model scanning the target region.

[0122] Step 7: Construct the connection relationship matrix L based on the point set M and the set of connecting lines between points E, and obtain the local feature information δ of the model according to equation (1) based on the theoretical triangular mesh model.

[0123] Step 8: Obtain the data information of the corresponding points in the triangular mesh model after deformation through steps 5 and 6, and add the corresponding information to the connection relationship matrix L and the local feature information δ.

[0124] Step 9: Substitute the connection matrix L and local feature information δ obtained in Step 8 into Equation (3), and multiply both ends by L on the left. T Then multiply by (L) on the left. T L) -1 The solution formula (7) for the deformed point cloud data is obtained, and the transformed point cloud dataset is finally obtained, as shown in the figure. Figure 6 Blue dotted clouds.

[0125] Step 10: In Geomagic Studio software, the point cloud data is encapsulated into a triangular mesh model, and the mesh doctor tool is used to remove mesh problems such as self-intersection and sharp points; enter the precise surface module, construct the contour line, then construct the surface patch and trim the surface patch, then construct the grid, and finally fit the grid into a surface; convert the reverse-modeled surface into a CAD model to realize the model reconstruction of local point cloud information.

[0126] The above-described embodiments are merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited thereto. Any simple changes or equivalent substitutions of the technical solutions that can be obviously obtained by those skilled in the art within the scope of the technology disclosed in the present invention shall fall within the scope of protection of the present invention.

Claims

1. A method for reconstructing a large-size part model based on local point cloud information, characterized in that, The method comprises the following steps: obtaining a triangular mesh model of a large-size part to be modeled; discretely meshing the triangular mesh model to form a triangular mesh model composed of nodes and connecting lines between the nodes; scanning local key features of the large-size part to obtain a local point cloud data model of the large-size part; dividing the local point cloud data model into a reference region and a target region; matching feature points in the reference region and the target region with corresponding points in the triangular mesh model, and performing pose transformation on the local point cloud data model after matching so that the local point cloud data model coincides with the triangular mesh model; searching for points on the target region of the local point cloud data model and the nearest points on the triangular mesh model to obtain a nearest point index set Mp on the triangular mesh model; matching the nearest point index set Mp with corresponding points of the target region of the local point cloud data model; moving points in the triangular mesh model located in the target region so that the points in the triangular mesh model located in the target region coincide with matching points in the target region of the local point cloud data model, and deforming the target region of the triangular mesh model; obtaining all point cloud information M of the deformed triangular mesh model by using Laplace equation; performing point cloud reverse modeling according to the obtained all point cloud information M to obtain a high-fidelity model, and using the high-fidelity model to perform assembly simulation on the large-size part.

2. The method according to claim 1, wherein, Local features of a point i in the triangular mesh model are: where δ i is the local feature matrix of point i in the triangular mesh model, (v i -v j ) is the vector formed by point i and its connected points.

3. The method of claim 1, wherein, The division of the local point cloud data model into the reference region and the target region comprises: In Geomagic Studio, the scanned local point cloud data model is divided into regions, and the reference region is selected by manual selection; the remaining point cloud data model regions are taken as the target region by reverse selection.

4. The method of claim 1, wherein, The matching of feature points in the reference region and the target region with corresponding points in the triangular mesh model, and the coincidence of the local point cloud data model with the triangular mesh model after matching comprise: manual establishment of feature points by using the feature matching function of Geomagic Studio, creation of three feature points in corresponding regions of the reference region, the target region and the triangular mesh model, matching of the three feature points to obtain a first conversion matrix; pose transformation of the local point cloud data model according to the first conversion matrix to obtain a local point cloud data model after transformation; ICP matching of the reference region, the target region and the corresponding region of the triangular mesh model by using the ICP matching algorithm to obtain a second conversion matrix; pose transformation of the local point cloud data model according to the second conversion matrix to obtain a local point cloud data model after transformation, and coincidence of the local point cloud data model with the triangular mesh model.

5. The method of claim 1, wherein, The Laplace equation is: LM = δ wherein, L is a connection relationship matrix of the triangular mesh model, δ is a local feature matrix in the triangular mesh model, and M is all point cloud information of the triangular mesh model.

6. The method of claim 5, wherein, The solution of the Laplace equation is: the reference region point coordinate information and the target region point coordinate information of the deformed triangular mesh model are added to the connection relationship matrix L of the triangular mesh model to obtain an updated connection relationship matrix L'. The target area point coordinate information of the deformed triangular mesh model is added to the local feature matrix δ to obtain an updated local feature matrix δ'; The updated connection relationship matrix L' and the local feature matrix δ' are substituted into the Laplace equation to calculate all point cloud information M, and then all point cloud information of the deformed part is obtained; The Laplace equation after substituting the connection relationship matrix L' and the local feature matrix δ' is: wherein, is the updated connection relationship matrix L', is the updated local feature matrix δ', ω is the weight of the scanning area point, I n×m is the class unit matrix, M n is the three-dimensional coordinates of the scanning area point.

7. The method of claim 6, wherein, The all point cloud information M is: M = (L T L) -1 L T δ' where L T is the transpose of the connection matrix L, and δ' is the updated local feature matrix.

8. The method of claim 1, wherein, The point cloud reverse modeling is performed according to the obtained point information to obtain a high-fidelity model, which comprises: The all point cloud information M is transmitted into Geomagic Studio in a txt format; The all point cloud information M is encapsulated to remove the self-intersection and the grid problem of the mesh model, and a first triangular mesh model is obtained; After the contour line of the first triangular mesh model is constructed by using an accurate surface module, a surface patch is constructed and the surface patch is trimmed; A grid is constructed, and the grid is fitted into a surface; The surface is converted into a CAD model to realize the construction of the high-fidelity model.

Images