A method and device for reconstructing the surface of CAD industrial parts based on approximate geodesic lines

Abstract

The present invention discloses a CAD industrial part surface reconstruction method based on approximate geodesics, and belongs to the technical field of industrial part structure design. The present invention obtains a rough surface by scanning the industrial part; then, after sampling the rough surface, a fine surface model of the industrial part for CAD modeling is generated, which is consistent with the part topology and similar to the part surface. Therefore, based on the fine surface model of the industrial part obtained by the present invention, since the quality of the model surface is improved in advance by the surface reconstruction technology, when the finite element method widely used in the industry is used to analyze the new surface, various more accurate parameter information of the part can be obtained more accurately, thereby making the simulation analysis results of the part more accurate and reliable, and providing support for the optimization of industrial part structure design.

Description

Technical Field

[0001] The invention belongs to the field of industrial part structure design, and in particular relates to a CAD industrial part surface reconstruction method and device based on approximate geodesics. Background Art

[0002] With the development of industrial technology, the design requirements for special parts are getting higher and higher. Using the finite element method to conduct simulation experiments in the computer to test the pressure resistance and heat resistance of parts can effectively reduce the scrap rate of experimental parts, thereby reducing costs and reducing the waste of resources. When conducting simulation experiments, traditional methods often only perform rough sampling of parts to obtain a low-precision model surface. Using the finite element method to analyze such a low-precision surface cannot obtain reliable results. Therefore, it is necessary to use surface reconstruction technology to improve the quality of the model surface, so as to make the analysis results of the parts more accurate and reliable. Summary of the invention

[0003] The purpose of the present invention is to solve the problems existing in the county-level technology and to propose a method and device for reconstructing the surface of a CAD industrial part based on approximate geodesics.

[0004] The specific technical solutions adopted by the present invention are as follows:

[0005] In a first aspect, the present invention provides a method for reconstructing a CAD industrial part surface based on an approximate geodesic, comprising:

[0006] S1. Obtain a rough part surface obtained by scanning an industrial part, and represent the rough part surface as a first mesh model in the form of a binary group consisting of vertices and triangles;

[0007] S2, sampling the rough part surface obtained by scanning more densely to obtain a lattice point set; inserting the lattice point set into the first mesh model and re-splitting the edges and faces through the Euler operation without changing the model topology to obtain a second mesh model; then subdividing the second mesh model to obtain a third mesh model;

[0008] S3, converting the third mesh model into an undirected weighted graph, and then generating a maximal disjoint path set in combination with the shortest path algorithm, wherein the maximal disjoint path set divides the fine part surface corresponding to the third mesh model into a series of generalized triangle areas, wherein the generalized triangle is a triangle whose edges are polylines;

[0009] S4. Straighten the three broken line edges of each generalized triangular area on the surface of the fine part and reconstruct them into a standard triangle with straight line segments as edges, so as to obtain an industrial part surface model for CAD modeling.

[0010] As a preferred embodiment of the first aspect, in S1, the rough part surface obtained by scanning is represented as a first grid model S in the form of a binary group.<V,T> , where V = {v′ i} represents the vertex set of the model, v′ i is the three-dimensional coordinate of the i-th vertex in the model, T = {t i} is the set of triangles that make up the model, where t i is the i-th triangle that makes up the model.

[0011] As a preferred embodiment of the first aspect, the specific implementation method of S2 is as follows:

[0012] S21. The rough surface of the scanned part is sampled more densely to obtain a lattice point set V s ;

[0013] S22, through the Euler operation without changing the model topology, V s Each lattice point in the first mesh model is iteratively inserted into S =<V,T> In each iteration, only V s A lattice point is extracted and inserted into the first mesh model S, and different update operations are performed on the triangle set according to its position after each round of lattice point insertion, where:

[0014] If the inserted lattice point is located at the vertex of the current latest first mesh model, no operation is required on the triangle set;

[0015] If the inserted lattice point is located on the common edge of two triangles of the current latest first mesh model, the two triangles are split into four triangles;

[0016] If the inserted lattice point is located inside a triangle of the current latest first mesh model, the triangle is split into three triangles;

[0017] Complete lattice point set V s After iterative insertion of all lattice points in , the second mesh model S′ is obtained r = <V∪V s ,T′ r >, where T′ r To insert the lattice point set V s The triangle set obtained after splitting and updating;

[0018] S23, using the mesh subdivision algorithm to refine the second mesh model S′ r Subdivide and obtain a fine mesh model S r = <V r ,T r >, where

[0019] As a preferred embodiment of the first aspect, the grid subdivision algorithm adopts Catmull-Clark subdivision algorithm.

[0020] As a preferred embodiment of the first aspect, the number of iteration rounds of the Catmull-Clark subdivision algorithm is preferably 3 rounds.

[0021] As a preferred embodiment of the first aspect, the specific implementation method of S3 is as follows:

[0022] S31, the third grid model S r = <V r ,T r >Convert to an undirected weighted graph G r = <V r ,E(S r )>, where E(S r ) is the third grid model S r The edge set of G r The edge weight in is the length of the corresponding edge; the path in an undirected weighted graph is defined as a sequence of vertices, where any two adjacent vertices in the path can be connected to form E(S r ) in the ; define the set A excluding the element b as A\{b};

[0023] S32, searching the undirected weighted graph by the shortest path algorithm, and generating a maximal disjoint path set P according to the following sub-steps S321 to S328:

[0024] S321: Initialize the path set P to be an empty set;

[0025] S322: For the lattice point set V s Every point v in i , use Dijkstra's algorithm to find a path with v i is the starting point and the end point is at V s \{v i}; lattice point set V s |V s | lattice points found |V s | shortest paths;

[0026] S323: |V s | shortest paths are put into a minimum heap q with path length as the keyword;

[0027] S324: If there is a path in the minimum heap q, take a path p from the top of the minimum heap and continue to execute the following steps S325 to S327, otherwise jump to S328;

[0028] S325: If the path p does not intersect any path in the path set P, then add the path p to the path set P; if and only if the two paths have no other common points except at most one endpoint, the two paths are considered to be disjoint;

[0029] S326: Let the starting point of path p be u, and use Dijkstra algorithm to search whether there is a path in the undirected weighted graph that starts at u and ends at V s The shortest path p′ in \{u} that does not intersect any path in set P. If such a path p′ exists, add the path p′ to the minimum heap q;

[0030] S327: Jump back to S324 and start execution;

[0031] S328: End the process, and use the latest path set P as the final maximal disjoint path set P;

[0032] S33, mapping all paths in the maximal disjoint path set P to the third grid model S r On the corresponding fine part surface, the fine part surface is divided into a series of generalized triangular areas.

[0033] As a preferred embodiment of the first aspect, in S4, when the generalized triangle is straightened to form a standard triangle, the three vertices of the generalized triangle area remain stationary, and the three broken line sides are straightened to form straight lines.

[0034] As a preferred embodiment of the first aspect, the rough part surface is obtained by performing three-dimensional scanning of the industrial part by a three-dimensional scanner.

[0035] In a second aspect, the present invention provides a CAD industrial part surface reconstruction device based on approximate geodesics, which includes a three-dimensional scanner and a reconstruction computing device;

[0036] The three-dimensional scanner is used to scan the target industrial part to obtain scanning data of the rough part surface;

[0037] The reconstruction computing device is used to reconstruct the rough part surface scanned by a three-dimensional scanner according to the CAD industrial part surface reconstruction method described in any scheme of the first aspect above, so as to obtain an industrial part surface model for CAD modeling.

[0038] As a preferred embodiment of the second aspect, a finite element analysis module is provided in the reconstruction computing device for performing finite element analysis on the industrial part surface model, thereby optimizing the design of the industrial part structure.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention obtains a rough surface by scanning an industrial part; then, after sampling the rough surface, a fine surface model of the industrial part for CAD modeling is generated, which is consistent with the part topology and similar to the part surface. Therefore, based on the fine surface model of the industrial part obtained by the present invention, since the quality of the model surface is improved by the surface reconstruction technology in advance, when the finite element method widely used in the industry is used to analyze the new surface, various more accurate parameter information of the part can be obtained more accurately, thereby making the simulation analysis results of the part more accurate and reliable, and providing support for the optimization of the structural design of industrial parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The flowchart of the surface reconstruction method of CAD industrial parts based on approximate geodesics.

[0042] Figure 2 This is the rough surface obtained by scanning the part (universal joint) in the embodiment of the present invention.

[0043] Figure 3 Schematic diagram of edge splitting steps in an embodiment of the present invention.

[0044] Figure 4 Schematic diagram of the face splitting step in an embodiment of the present invention.

[0045] Figure 5 Schematic diagram of a fine grid model in an embodiment of the present invention.

[0046] Figure 6 A flow chart is generated for a set of disjoint paths in an embodiment of the present invention.

[0047] Figure 7 Schematic diagram of a set of disjoint paths in an embodiment of the present invention.

[0048] Figure 8 It is a schematic diagram of the final reconstruction result in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention is further described and illustrated below in conjunction with the accompanying drawings and specific embodiments. The technical features of each embodiment of the present invention can be combined accordingly without conflicting with each other.

[0050] In order to make the above-mentioned purpose, features and advantages of the present invention more obvious and easy to understand, the specific implementation mode of the present invention is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below. The technical features in each embodiment of the present invention can be combined accordingly without conflicting with each other.

[0051] In the description of the present invention, it should be understood that the terms "first" and "second" are only used for the purpose of distinguishing descriptions, and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features.

[0052] like Figure 1 As shown, in a preferred embodiment of the present invention, a method for reconstructing a CAD industrial part surface based on an approximate geodesic is provided, which comprises:

[0053] S1. Obtain a rough part surface obtained by scanning an industrial part, and represent the rough part surface as a first mesh model in the form of a binary group consisting of vertices and triangles;

[0054] As a preferred implementation of the embodiment of the present invention, in S1, the rough part surface obtained by scanning is represented as a first grid model S in the form of a binary group.<V,T> , where V = {v′ i} represents the vertex set of the model, v′ i is the three-dimensional coordinate of the i-th vertex in the model, T = {t i} is the set of triangles that make up the model, where t i is the i-th triangle that makes up the model.

[0055] As a preferred implementation of the embodiment of the present invention, the rough part surface can be obtained by performing three-dimensional scanning on the industrial part using a three-dimensional scanner.

[0056] S2, sampling the rough part surface obtained by scanning more densely to obtain a lattice point set; inserting the lattice point set into the first mesh model and re-splitting the edges and faces through the Euler operation without changing the model topology to obtain a second mesh model; then subdividing the second mesh model to obtain a third mesh model;

[0057] As a preferred implementation of the embodiment of the present invention, the specific implementation method of S2 is as follows:

[0058] S21. The rough surface of the scanned part is sampled more densely to obtain a lattice point set V s ;

[0059] S22, through the Euler operation without changing the model topology, V s Each lattice point in the first mesh model is iteratively inserted into S =<V,T> In each iteration, only V s A lattice point is extracted and inserted into the first mesh model S, and different update operations are performed on the triangle set according to its position after each round of lattice point insertion, where:

[0060] If the inserted lattice point is located at the vertex of the current latest first mesh model, no operation is required on the triangle set;

[0061] If the inserted lattice point is located on the common edge of two triangles of the current latest first mesh model, the two triangles are split into four triangles;

[0062] If the inserted lattice point is located inside a triangle of the current latest first mesh model, the triangle is split into three triangles;

[0063] Complete lattice point set V s After iterative insertion of all lattice points in , the second mesh model S′ is obtained r = <V∪V s ,T r ′>, where T r ′ is the inserted lattice point set V s The triangle set obtained after splitting and updating;

[0064] S23, using the mesh subdivision algorithm to refine the second mesh model S′ r Subdivide and obtain a fine mesh model S r = <V r ,T r >, where

[0065] The mesh subdivision algorithm used is preferably the Catmull-Clark subdivision algorithm, and the number of iterations of the Catmull-Clark subdivision algorithm is preferably 3.

[0066] S3, converting the third mesh model into an undirected weighted graph, and then generating a maximal disjoint path set in combination with the shortest path algorithm, wherein the maximal disjoint path set divides the fine part surface corresponding to the third mesh model into a series of generalized triangle areas, wherein the generalized triangle is a triangle whose edges are polylines;

[0067] As a preferred implementation method of the embodiment of the present invention, the specific implementation method of S3 is as follows:

[0068] S31, the third grid model S r = <V r ,T r >Convert to an undirected weighted graph G r = <V r ,E(S r )>, where E(S r ) is the third grid model S r The edge set of G r The edge weight in is the length of the corresponding edge; the path in an undirected weighted graph is defined as a sequence of vertices, where any two adjacent vertices in the path can be connected to form E(S r ) in the ; define the set A excluding the element b as A\{b};

[0069] S32, searching the undirected weighted graph by the shortest path algorithm, and generating a maximal disjoint path set P according to the following sub-steps S321 to S328:

[0070] S321: Initialize the path set P to be an empty set;

[0071] S322: For the lattice point set V s Every point v in i , use Dijkstra's algorithm to find a path with v i is the starting point and the end point is at V s \{v i}; lattice point set V s |V s | lattice points found |V s | shortest paths;

[0072] S323: |V s | shortest paths are put into a minimum heap q with path length as the keyword;

[0073] S324: If there is a path in the minimum heap q, take a path p from the top of the minimum heap and continue to execute the following steps S325 to S327, otherwise jump to S328;

[0074] S325: If the path p does not intersect any path in the path set P, then add the path p to the path set P; if and only if the two paths have no other common points except at most one endpoint, the two paths are considered to be disjoint;

[0075] S326: Let the starting point of path p be u, and use Dijkstra algorithm to search whether there is a path in the undirected weighted graph that starts at u and ends at Vs The shortest path p′ in \{u} that does not intersect any path in set P. If such a path p′ exists, add the path p′ to the minimum heap q;

[0076] S327: Jump back to S324 and start execution;

[0077] S328: End the process, and use the latest path set P as the final maximal disjoint path set P;

[0078] S33, mapping all paths in the maximal disjoint path set P to the third grid model S r On the corresponding fine part surface, the fine part surface is divided into a series of generalized triangular regions. The strict definition of a generalized triangle is that it is homeomorphic to a disk, and there are exactly 3 points on the boundary belonging to V s area.

[0079] S4. Straighten the three broken line edges of each generalized triangular area on the surface of the fine part and reconstruct them into a standard triangle with straight line segments as edges, so as to obtain an industrial part surface model for CAD modeling.

[0080] As a preferred implementation method of the embodiment of the present invention, in S4, when the generalized triangle is straightened to form a standard triangle, the three vertices of the generalized triangle area remain stationary, and the three broken line sides are straightened to form straight lines.

[0081] In addition, based on the same inventive concept, the present invention also provides a CAD industrial part surface reconstruction device based on approximate geodesics, which includes a three-dimensional scanner and a reconstruction computing device;

[0082] The three-dimensional scanner is used to scan the target industrial part to obtain scanning data of the rough part surface;

[0083] The reconstruction computing device is used to reconstruct the rough part surface scanned by the three-dimensional scanner according to the aforementioned CAD industrial part surface reconstruction method to obtain an industrial part surface model for CAD modeling.

[0084] In addition, the above-mentioned reconstruction computing device may also be provided with a finite element analysis module for performing finite element analysis on the industrial part surface model, and then optimizing the design of the industrial part structure.

[0085] Based on the CAD industrial part surface reconstruction method described in S1 to S4 above, the following uses a specific example to demonstrate its implementation details and technical effects.

[0086] Example

[0087] In this embodiment, a universal joint is used as an industrial part to be processed, and its part surface is reconstructed according to the following process.

[0088] Step 1: Get a rough surface:

[0089] First, a 3D scanner is used to scan the part to obtain a simple data format representing the surface of a 3D geometric figure. This embodiment uses a binary group S=<V,T> To represent the surface data format obtained by scanning. Where V = {v′ i} represents the vertex set of the model, v′ i is the three-dimensional coordinate of the i-th point in the model. i} are all the triangles that make up the model, where t i is the i-th triangle that constitutes the model. Figure 2 As shown, the rough surface obtained in this step has very poor quality. Using a rough surface for finite element analysis will result in large errors, which will reduce the reliability of the simulation experiment and increase the scrap rate. Therefore, this embodiment obtains a new surface through the following steps.

[0090] Step 2: Obtain high-resolution surface: This embodiment densely samples on the rough surface to obtain a lattice point set V s . Get the rough surface S =<V,T> With the lattice sampling point set V s After that, this embodiment uses a series of Euler operations such as edge splitting and face splitting that do not change the topological structure of the model to convert V s Insert mesh model S=<V,T> In the grid model S′ t = <V∪V s ,T′ r >. This is a step-by-step iterative process. In each iteration, this embodiment only adds V s A point in is inserted into S. In this embodiment, M0=S. The mesh model obtained after inserting the i-th point is defined as M i = <V i ,T i >. From the definition, we know that V i =V i-1 ∪{v i}. When V s Insert the points in into S, and we get Next, we will discuss v i Already M i-1 Vertex, at M i-1 On the edge or in M i-1 How to insert the point by modifying the topology when it is inside the triangle.

[0091] In each iteration, if v i Already M i-1 No operation is required for the vertices.

[0092] like Figure 3 As shown, if v i In M i-1 On the common side ab of Δabc and Δabd, then v i Add to edge ab and split the 2 related triangles to form 4 triangles.

[0093] T i =(T i-1 \{Δabc, Δabd})∪{Δacv i , Δbcv i , Δadv i , Δbdv i}

[0094] like Figure 4 As shown, if v i In M i-1 Inside the triangle Δabc, then v i Add it to Δabc and split Δabc into 3 triangles.

[0095] T i =(T i-1 \{Δabc})∪{Δabv i , Δbcv i , Δcav i}

[0096] So far, V has been split through edge splitting, face splitting and other Euler operations. s Insert into S, and get S′ r = <V∪V s ,T′ r >.

[0097] The fineness of the mesh directly affects the error degree of the approximate geodesic. The edges added by Loop subdivision are parallel to the original edges, so it is difficult to approximate the geodesic in certain specific directions. Sqrt3 subdivision and Modified Butterfly subdivision cannot maintain features due to the existence of edge turning operations. After experiments, the Catmull-Clark subdivision algorithm can maintain features and has a small error in approximating geodesics in all directions. Therefore, this embodiment uses the Catmull-Clark subdivision algorithm to approximate the mesh model S′ r Subdivide and obtain a fine mesh model S r = <V r ,T r >, where In this embodiment, the number of Catmull-Clark iterations can be set empirically. For most models, only three iterations are usually performed. Figure 5Shown

[0098] At this point, the fine grid S is obtained. r = <V r ,T r >, meet Theoretically, the finer the mesh, the closer the shortest path along the edge is to the geodesic. When the mesh is infinitely fine and the vertices can cover the entire surface area, the shortest path along the edge can be considered to be the geodesic. r If we use Dijkstra's algorithm to calculate the shortest path between any two points on the grid, we can use it to approximate the geodesic with a smaller error using the shortest path, which can improve the quality of the new grid's cells.

[0099] Step 3: Generate the maximum disjoint path set P:

[0100] In this embodiment, the fine surface S obtained in the previous step r = <V r ,T r >As a data structure in computer science - undirected weighted graph G r = <V r ,E(S r )>, where E(S r ) is the fine surface S r The edge set of G. r The edge weight in is S r The length of the corresponding side in (the Euclidean distance between two points).

[0101] This embodiment defines a path as a vertex sequence:

[0102] p=(u1,u2,...,u n >

[0103]

[0104] This embodiment obtains a path set P through the following sub-steps 3.1 to 3.8, and its flow chart is as follows: Figure 6 shown.

[0105] Step 3.1: Initialize the path set P to be an empty set.

[0106] Step 3.2: Sample the lattice point set V s Every point v in i , use Dijkstra's algorithm to find a path with v i As the starting point, V s \{v} is the shortest path with a point as the end point. s Find a shortest path for each point in |V s | shortest paths.

[0107] Step 3.3: Put this |V s | paths are put into a minimum heap q with the path length as the key.

[0108] Step 3.4: If there is a path in the minimum heap q, take a path p from the top of the minimum heap and continue with the following steps, otherwise jump to step 3.8.

[0109] Step 3.5: If path p does not intersect any path in path set P, then add path p to path set P. Two paths are disjoint if and only if they have no other common points except at most one endpoint.

[0110] Step 3.6: Let the starting point of p be u. Taking u as the starting point, V s A point in \{u} is the end point, and Dijkstra's algorithm is used to search for a shortest path that does not intersect any path in the set P. If such a path p' exists, p' is added to the minimum heap q.

[0111] Step 3.7: Skip to step 3.4.

[0112] Step 3.8: End the process.

[0113] like Figure 7 As shown, the path set P obtained through the above steps will be the surface S r Divided into multiple generalized triangular regions.

[0114] Here, the generalized triangle refers to a triangle whose sides are polylines. The strict definition of a generalized triangle is that it is homeomorphic to the disk and there are exactly 3 points on the boundary that belong to V s area.

[0115] Step 4: Get the new model surface:

[0116] like Figure 7 As shown, the path set P obtained in the previous step will transform the refined surface S r The generalized triangle region is divided into multiple generalized triangle regions. The boundaries of the generalized triangle region are three broken lines (i.e., paths) in the path set P. For each generalized triangle, this embodiment straightens the three broken lines to construct a triangle unit t. Figure 8 As shown, all the new triangular units constitute the new surface mesh model S of the universal joint. m = <V s ,T m >.

[0117] Step 5: New surface mesh model S based on the universal joint mConduct finite element analysis and simulation tests according to the simulation optimization plan. For example, conduct pressure resistance and heat resistance tests on universal joint parts to obtain more accurate parameter information and provide a basis for the optimal design of universal joint parts.

[0118] The above-described embodiment is only a preferred solution of the present invention, but it is not intended to limit the present invention. A person skilled in the relevant technical field may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, any technical solution obtained by equivalent replacement or equivalent transformation falls within the protection scope of the present invention.

Claims

1. A CAD industrial part surface reconstruction method based on approximate geodesics, characterized in that: include: S1. Obtain a rough part surface obtained by scanning an industrial part, and represent the rough part surface as a first mesh model in the form of a binary group consisting of vertices and triangles; S2, sampling the rough part surface obtained by scanning more densely to obtain a lattice point set; inserting the lattice point set into the first mesh model and re-splitting the edges and faces through the Euler operation without changing the model topology to obtain a second mesh model; then subdividing the second mesh model to obtain a third mesh model; S3, converting the third mesh model into an undirected weighted graph, and then generating a maximal disjoint path set in combination with the shortest path algorithm, wherein the maximal disjoint path set divides the fine part surface corresponding to the third mesh model into a series of generalized triangle areas, wherein the generalized triangle is a triangle whose edges are polylines; The maximal disjoint path set P is generated by searching the undirected weighted graph using the shortest path algorithm according to the following sub-steps S321 to S328: S321: Initialize the path set P to be an empty set; S322: For the lattice point set V s Every point v in i , use Dijkstra's algorithm to find a path with v i is the starting point and the end point is at V s \{v i }; lattice point set V s |V s | lattice points found |V s | shortest paths; S323: |V s | shortest paths are put into a minimum heap q with path length as the keyword; S324: If there is a path in the minimum heap q, take a path p from the top of the minimum heap and continue to execute the following steps S325 to S327, otherwise jump to S328; S325: If the path p does not intersect any path in the path set P, then add the path p to the path set P; if and only if the two paths have no other common points except at most one endpoint, the two paths are considered to be disjoint; S326: Let the starting point of path p be u, and use Dijkstra algorithm to search whether there is a path in the undirected weighted graph that starts at u and ends at V s The shortest path p′ in \{u} that does not intersect any path in set P. If such a path p′ exists, add the path p′ to the minimum heap q; S327: Jump back to S324 and start execution; S328: End the process, and use the latest path set P as the final maximal disjoint path set P; S4. Straighten the three broken line edges of each generalized triangular area on the surface of the fine part and reconstruct them into a standard triangle with straight line segments as edges, so as to obtain an industrial part surface model for CAD modeling.

2. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 1, characterized in that: In S1, the rough part surface obtained by scanning is represented as a first grid model S in the form of a binary group.<V,T> , where V = {v′ i } represents the vertex set of the model, v′ i is the three-dimensional coordinate of the i-th vertex in the model, T = {t i } is the set of triangles that make up the model, where t i is the i-th triangle that makes up the model.

3. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 1, characterized in that: The specific implementation method of S2 is as follows: S21. The rough surface of the scanned part is sampled more densely to obtain a lattice point set V s ; S22, through the Euler operation without changing the model topology, V s Each lattice point in the first mesh model is iteratively inserted into S =<V,T> In each iteration, only V s A lattice point is extracted and inserted into the first mesh model S, and different update operations are performed on the triangle set according to its position after each round of lattice point insertion, where: If the inserted lattice point is located at the vertex of the current latest first mesh model, no operation is required on the triangle set; If the inserted lattice point is located on the common edge of two triangles of the current latest first mesh model, the two triangles are split into four triangles; If the inserted lattice point is located inside a triangle of the current latest first mesh model, the triangle is split into three triangles; Complete lattice point set V s After iterative insertion of all lattice points in , the second mesh model S′ is obtained r = <V∪V s ,T′ r >, where T′ r To insert the lattice point set V s The triangle set obtained after splitting and updating; S23, using the mesh subdivision algorithm to refine the second mesh model S′ r Subdivide and obtain a fine mesh model S r = <V r ,T r >, where 4. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 3, characterized in that: The mesh subdivision algorithm adopts the Catmull-Clark subdivision algorithm.

5. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 4, characterized in that: The number of iterations of the Catmull-Clark subdivision algorithm is preferably 3.

6. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 1, characterized in that: The specific implementation method of S3 is as follows: S31, the third grid model S r = <V r ,T r >Convert to an undirected weighted graph G r = <V r ,E(S r )>, where E(S r ) is the third grid model S r The edge set of G r The edge weight in is the length of the corresponding edge; the path in an undirected weighted graph is defined as a sequence of vertices, where any two adjacent vertices in the path can be connected to form E(S r ) in the ; define the set A excluding the element b as A\{b}; S32, searching the undirected weighted graph by the shortest path algorithm to generate a maximum disjoint path set P; S33, mapping all paths in the maximal disjoint path set P to the third grid model S r On the corresponding fine part surface, the fine part surface is divided into a series of generalized triangular areas.

7. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 1, characterized in that: In S4, when the generalized triangle is straightened to form a standard triangle, the three vertices of the generalized triangle region remain stationary, and the three sides in the form of broken lines are straightened to form straight lines.

8. The CAD industrial part surface reconstruction method based on approximate geodesics according to claim 1, characterized in that: The rough part surface is obtained by performing three-dimensional scanning on the industrial part by a three-dimensional scanner.

9. A CAD industrial part surface reconstruction device based on approximate geodesic, characterized in that: Includes 3D scanners and reconstruction computing equipment; The three-dimensional scanner is used to scan the target industrial part to obtain scanning data of the rough part surface; The reconstruction computing device is used to reconstruct the rough part surface scanned by a three-dimensional scanner according to the CAD industrial part surface reconstruction method according to any one of claims 1 to 7 to obtain an industrial part surface model for CAD modeling.

10. The CAD industrial part surface reconstruction device according to claim 9, characterized in that: The reconstruction computing device is provided with a finite element analysis module for performing finite element analysis on the industrial part surface model, thereby optimizing the design of the industrial part structure.

Images