Large-scale tetrahedron parallel deletion method based on quadratic error measurement

By adopting the parallel deletion method of quadratic error metric in large-scale tetrahedral mesh, and using parallel computing optimization algorithms and data structures, the problem of high computational complexity of large-scale mesh simplification in the existing technology is solved, and efficient Grid simplification and high geometric fidelity, suitable for fields such as computer graphics.

CN119027612APending Publication Date: 2024-11-26TIANJIN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411097088.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing mesh simplification method has high computational complexity and high resource consumption when dealing with large-scale three-dimensional mesh, making it difficult to effectively simplify large-scale tetrahedral mesh, resulting in low computing efficiency and poor simplification effect.

Method used

Using a large-scale tetrahedron parallel deletion method based on quadratic error metric, by defining the quadratic error matrix of non-boundary surfaces and boundary surface vertices, randomly extracting edges to minimize error shrinkage and delete related tetrahedrons, using parallel computing optimization algorithm and data structures, iteratively update the data structure until the preset simplified ratio or iteration number is reached.

Benefits of technology

It significantly improves the simplified efficiency of large-scale tetrahedral mesh, maintains high geometric fidelity and computational efficiency, and is suitable for fields that process large-scale three-dimensional data, such as computer graphics and scientific computing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027612A_ABST
    Figure CN119027612A_ABST
Patent Text Reader

Abstract

The invention discloses a large-scale tetrahedron parallel deletion method based on quadratic error measurement, and the method comprises the steps: S1, reading tetrahedron mesh data containing vertex coordinates, vertex information and vertex indexes of tetrahedrons, and completing the initialization of a data structure through the analysis of the tetrahedron mesh data; s2, defining a boundary surface, and respectively setting a quadratic error matrix Q of vertexes on a non-boundary surface of the tetrahedron and a quadratic error matrix Qb of vertexes on the boundary surface so as to evaluate the shrinkage error of each edge; s3, randomly extracting the edges of a plurality of tetrahedrons, and iteratively updating the data structure in a manner of shrinking the edges with the minimum error and deleting the tetrahedrons related to the shrunk edges until a preset simplification proportion or iteration times is reached; according to the large-scale tetrahedron parallel deletion method based on quadratic error measurement, a large-scale tetrahedron mesh can be efficiently simplified, and meanwhile it is guaranteed that the simplified mesh has high geometric fidelity and calculation efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional volume data compression, and in particular to a large-scale tetrahedron parallel deletion method based on quadratic error measurement. Background Art

[0002] With the rapid development of computer graphics and computational geometry, three-dimensional mesh models have been widely used in many application fields, such as computer-aided design (CAD), finite element analysis (FEA), computer animation, medical image processing, virtual reality (VR), etc. However, as the size and complexity of the model increase, the demand for computing and storage resources also increases dramatically. This demand not only affects the real-time rendering and interactive performance of the model, but also increases the burden of data transmission and storage. Therefore, how to reduce the number of vertices and faces of the model while ensuring the geometric accuracy and fidelity of the model has become an important research direction.

[0003] As an important means to solve the above problems, mesh simplification technology aims to reduce the number of vertices and facets in the mesh, thereby reducing computational complexity and storage requirements. Traditional mesh simplification methods mainly include edge collapse, vertex removal, and facet deletion. These methods are usually based on geometric error metrics and gradually simplify the mesh by locally optimizing the mesh structure. However, these methods often have problems such as low computational efficiency and poor simplification effect when processing large-scale mesh data.

[0004] The quadratic error metric (QEM) method has shown significant advantages in the field of mesh simplification. The QEM method can more accurately evaluate the geometric errors in the mesh simplification process by calculating the quadratic error matrix of the vertices, thereby achieving efficient mesh simplification while maintaining the geometric characteristics of the model. However, the QEM method still faces the challenges of high computational complexity and high resource consumption when processing large-scale three-dimensional meshes. Summary of the invention

[0005] The purpose of the present invention is to provide a large-scale tetrahedron parallel deletion method based on quadratic error metric that solves the above technical problems, fully utilizes parallel computing technology, and significantly improves mesh simplification by optimizing algorithms and data structures.

[0006] To this end, the technical solution of the present invention is as follows:

[0007] A large-scale tetrahedron parallel deletion method based on quadratic error metric, the steps are:

[0008] S1, read the tetrahedron mesh data including vertex coordinates, vertex information and vertex index of the tetrahedron, and complete the data structure initialization by parsing the tetrahedron mesh data;

[0009] S2. Define the boundary surface and set the quadratic error matrix Q of the vertices on the non-boundary surface of the tetrahedron and the quadratic error matrix Q of the vertices on the boundary surface respectively. b , used to evaluate the shrinkage error of each edge;

[0010] S3, randomly extract multiple edges of tetrahedrons, adopt the method of shrinking the edges in a way that minimizes the error, and delete the tetrahedrons related to the shrunk edges, and iteratively update the data structure until a preset simplification ratio or number of iterations is reached; wherein, the method of shrinking the edges in a way that minimizes the error, and deletes the tetrahedrons related to the shrunk edges is as follows: 1) randomly extract multiple tetrahedrons, and randomly extract an edge from each tetrahedron; 2) for each edge, respectively calculate the quadratic error matrix Q of the vertices at both ends of the edge, and obtain a new quadratic error matrix Q' by merging the quadratic error matrix Q of the vertices at both ends of each edge; 3) according to the calculation result of step S301, select the edge with the minimum quadratic error matrix Q' from the multiple randomly extracted edges as the shrunk edge; 4) based on the quadratic error matrix Q' of the shrunk edge, obtain the vertex position after the edge is shrunk; 5) update the coordinates in the vertex information of the tetrahedron that originally shared the same vertex with the shrunk edge; 6) mark the tetrahedron whose vertices are shrunk to the same position as the tetrahedron to be deleted and remove it in the subsequent data structure update.

[0011] Furthermore, in step S1, parsing and completing the initialization of the data structure includes: parsing the vertex coordinates into a floating-point array, parsing the vertex index of the tetrahedron into an integer, creating a vertex list for storing the coordinate information of all vertices, creating a tetrahedron list for storing the vertex index of the tetrahedron, creating an edge list for storing the vertex index information of each edge in the tetrahedron, and initializing data structures related to tetrahedral mesh processing, including the quadratic error matrix of the edge and the error metric of the tetrahedron.

[0012] Furthermore, in the created vertex list, all vertex coordinates and vertex-related information obtained by parsing are stored one by one in the vertex list, and each vertex has a unique index in the list. Vertex-related information includes physical quantities and colors; in the created tetrahedron list, the vertex indexes of all tetrahedrons are stored one by one in the table; in the created edge list, the vertex index information of each edge in the tetrahedron is stored one by one in the table.

[0013] Further, in step S2, the boundary surface is defined as: a surface formed by coordinates of three points constituting the plane that are not two completely identical; wherein,

[0014] (1) For the vertices on the non-boundary surfaces of the tetrahedron, the quadratic error matrix Q is defined as:

[0015] Suppose there is a point p and a neighboring point x in a high-dimensional space, and we need to calculate the error from x to several planes;

[0016] Set the basis vectors [e1,e2,…,e d ] forms the tangent plane where p is located, [e d+1 ,…,e n ] is perpendicular to the plane where p lies;

[0017] Then the error function is defined, that is, the square of the distance from point x to the plane is expressed as:

[0018]

[0019] Choose a set of basis vectors [e i ] so that the tangent plane [e i ] remains unchanged, and we get:

[0020]

[0021] definition: but,

[0022] The expression of the quadratic error matrix Q for each vertex of each tetrahedron is:

[0023] Q p (x) = x T Ax-2x T Ap+p T Ap,

[0024] In the above two formulas, b=-Ap,c=p T Ap;

[0025] (2) For the vertices on the boundary surface of the tetrahedron, the quadratic error matrix Q b Defined as:

[0026] Assume (v1, v2, v3) are the adjacent boundary vertices of boundary vertex v1, and assume that the plane normal vectors are n1 and n2, then define the k function by the cross product of the normal vectors of adjacent boundary surfaces:

[0027]

[0028] Furthermore, the quadratic error matrix is ​​given by Q b The expression is:

[0029] Q b (v1) = Q (v1) + W b *k(v1),

[0030] Q b (v2) = Q (v2) + W b*k(v2),

[0031] In the formula, Q b (v1) and Q b (v2) is the quadratic error function at the boundary surface vertices v1 and v2; W b is a user-defined weight factor; k(v1) and k(v2) are the cross products of the boundary normal vectors at vertices v1 and v2.

[0032] Furthermore, the specific implementation steps of step S3 are:

[0033] S301, before each edge contraction operation starts, initialize related data structures, such as a vertex list, a tetrahedron list, and an edge list; set a counter for recording the number of deleted tetrahedrons, and initialize the simplification ratio and the number of iterations;

[0034] S302, performing edge shrinkage and tetrahedron deletion operations by adopting a method of shrinking the edge in a way that minimizes the error and deleting the tetrahedrons associated with the shrunk edge:

[0035] S303, data structure update, including updating the vertex list, updating the edge list and updating the tetrahedron list;

[0036] S304, calculate the simplification ratio of the current grid after this iteration, and determine whether step S3 is completed: 1) If the preset simplification ratio or number of iterations is reached, step S404 is completed and the final updated data structure is output; 2) If the preset simplification ratio or number of iterations is not reached, steps S302 to S304 are repeated.

[0037] Furthermore, step S3 is performed using a parallel computing method through a parallel computing library, and the steps are:

[0038] S301, assigning edges to threads: assigning all edges to be processed to different threads according to a set strategy, ensuring that each thread processes the same or similar number of edges;

[0039] S302, executing the shrinking operation in parallel: each thread independently and in parallel executes the edge shrinking operation, calculates the new vertex position, and updates the vertex and edge information;

[0040] S303, synchronization operation: after each round of parallel shrinking operation is completed, a synchronization operation is performed to ensure that the results of all threads are consistent and to prepare for the next round of edge shrinking operation.

[0041] Compared with the existing technology, this large-scale tetrahedron parallel deletion method based on quadratic error metric can efficiently simplify large-scale tetrahedral meshes while ensuring that the simplified meshes have high geometric fidelity and computational efficiency; this method is particularly suitable for fields that need to process large-scale three-dimensional data, such as computer graphics, scientific computing, and engineering simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of the large-scale tetrahedron parallel deletion method based on quadratic error metric of the present invention;

[0043] Figure 2 It is a comparison diagram of the original image of 1 million grid data in an embodiment of the present invention, the VTK method commonly used in the industry, and the method of the present invention;

[0044] Figure 3 It is a comparison diagram of the original image of 100,000 grid data in the embodiment of the present invention, the VTK method commonly used in the industry, and the method of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but the following embodiments are by no means intended to limit the present invention in any way.

[0046] See also Figure 1 The specific implementation steps of the large-scale tetrahedron parallel deletion method based on quadratic error metric are as follows:

[0047] S1, data preprocessing, including reading tetrahedral mesh data, parsing tetrahedral mesh data, and initializing data structure;

[0048] The specific implementation steps of step S1 are:

[0049] S101, reading tetrahedral mesh data from an input file or a data source;

[0050] The purpose of this step is to obtain the original tetrahedral mesh data from various data sources; these data sources can be files, databases or other storage media; the tetrahedral mesh data read includes vertex coordinates, vertex information and vertex index of the tetrahedron; this information is the basis for subsequent mesh simplification and processing;

[0051] S102, parsing the read tetrahedral mesh data and converting it into a data structure suitable for algorithm processing;

[0052] Since the read data is usually stored in a certain format, in order for the data to be effectively processed by the algorithm, it needs to be parsed and converted into a suitable internal data structure; specifically,

[0053] S1021, parsing the vertex coordinates into a floating point array; this parsing step is because the geometric position of the vertex is usually represented by a floating point number, and by parsing the vertex coordinates into a floating point array, geometric calculations and operations can be easily performed;

[0054] S1022, parsing the vertex index of the tetrahedron into an integer; specifically, a tetrahedron is composed of a number of vertices, each of which has a unique index, and by parsing these indexes into integers, it is convenient to perform subsequent association and operations between the vertices and the tetrahedron;

[0055] S103, creating a vertex list for storing coordinate information of all vertices;

[0056] In order to manage and operate vertices later, this step creates a vertex list to store the coordinate information of all vertices; the data structure of the vertex list can be an array, a linked list, etc. The specific steps are:

[0057] S1031. Create a vertex list data structure;

[0058] S1032, storing all vertex coordinates obtained by parsing into a vertex list one by one, and each vertex has a unique index in the list for identifying the vertex; the vertex list not only stores the geometric position of the vertex, but also may store other vertex-related information, such as physical quantity, color, etc.;

[0059] S104, creating a tetrahedron list, and storing the vertex indexes of all tetrahedrons obtained by parsing into the tetrahedron list one by one;

[0060] S105 , creating an edge list, and storing the extracted vertex index information of each edge in each tetrahedron into the edge list.

[0061] S2. Define the boundary surface and set the quadratic error matrix Q of the vertices on the non-boundary surface of the tetrahedron and the quadratic error matrix Q of the vertices on the boundary surface respectively. b , used to evaluate the shrinkage error of each edge;

[0062] In step S2, the boundary surface is defined as: a surface formed by coordinates of three points constituting the plane that are not two completely identical;

[0063] Suppose there is a point p and a neighboring point x in a high-dimensional space, and we need to calculate the error from x to several planes;

[0064] Conventional basis vectors [e1,e2,…,e d ] forms the tangent plane where p is located, [e d+1 ,…,e n ] is perpendicular to the plane where p lies;

[0065] Define the error function, that is, the square of the distance from point x to the plane as shown in formula (2-1):

[0066]

[0067] Its expanded square sum is shown by formula (2-2):

[0068]

[0069]

[0070] Then, we select a set of basis vectors [e i ] so that the tangent plane [e i ] remains unchanged, and according to formula (2-3):

[0071]

[0072] Further conversion gives formula (2-4):

[0073]

[0074] in, The above formula can be rewritten as: p (x)=(xp) T A(xp),

[0075] Further expansion can obtain formula (2-5):

[0076]

[0077] Therefore, we can get formula (2-6):

[0078]

[0079] For tetrahedral meshes, d = 3, then formula (2-6) is further converted to formula (2-7):

[0080]

[0081] S202: For the vertices on the boundary surface, a preservation strategy is adopted for the boundary. Therefore, based on the quadratic error matrix Q in step S201, formula (2-8) is obtained:

[0082]

[0083] The boundary surface is defined as a surface where there are no two surfaces with exactly the same coordinates of the three points that make up the plane; the k function is defined by taking the cross product of the normal vectors of adjacent boundary surfaces:

[0084] Assume (v1, v2, v3) are the adjacent boundary vertices of boundary vertex v1, and assume that the plane normal vectors are n1 and n2, then it can be shown as formula (2-9):

[0085] n1=(v2-v1)×(v3-v1) (2-9),

[0086] According to the cross product formula, the normal vector n1 can be expressed as formula (2-10):

[0087]

[0088] Similarly, another normal vector n2 is also expressed in the same way as above;

[0089] Furthermore, k is defined by formula (2-11):

[0090]

[0091] Its modulus is shown in formula (2-12):

[0092]

[0093] The modulus lengths of normal vectors n1 and n2 are shown in formula (2-13):

[0094]

[0095] Finally, substitute these values ​​into formula (2-11) to obtain formula (2-14):

[0096]

[0097] That is the calculation formula of k;

[0098] Then for the vertices on the tetrahedron boundary surface, the quadratic error matrix is ​​calculated according to Q b Function formula (2-15) calculation:

[0099]

[0100] Among them, Q b (v1) and Q b (v2) is the quadratic error function at the boundary surface vertices v1 and v2; W b is a user-defined weight factor; k(v1) and k(v2) are the cross products of the boundary normal vectors at vertices v1 and v2.

[0101] S3, updating the data structure by deleting tetrahedrons related to the contraction edge, and iteratively performing edge contraction and tetrahedron deletion operations until a preset simplification ratio or number of iterations is reached;

[0102] The specific implementation steps of step S3 are:

[0103] S301, initialization deletion and update operations:

[0104] Before each edge contraction operation begins, initialize related data structures, such as vertex lists, tetrahedron lists, and edge lists; set a counter for recording the number of deleted tetrahedrons, and initialize the error threshold or simplification ratio;

[0105] The purpose of this step is to ensure that each edge contraction and tetrahedron deletion operation has a clean starting point to facilitate the accurate execution of subsequent operations.

[0106] S302, perform edge contraction and tetrahedron deletion operations:

[0107] S3021, randomly selecting multiple tetrahedrons, and randomly selecting an edge from each tetrahedron; in this embodiment, the number of randomly selected tetrahedrons is 8;

[0108] S3022, for each edge, respectively calculate the quadratic error matrix Q of the vertices at both ends of the edge, and obtain a new quadratic error matrix Q' by merging the quadratic error matrices Q of the vertices at both ends of each edge; the new quadratic error matrix Q' is used to represent the error after shrinking the edge;

[0109] S3023, according to the calculation result of step S301, select the edge with the minimum quadratic error matrix Q' from the multiple randomly selected edges as the contraction edge; in this determination step, the smaller the contraction error value of the edge, the closer the geometric shape after the contraction edge is to the original shape; and the larger the error, the more the geometric shape after the contraction edge deviates from the original shape;

[0110] S3024, based on the quadratic error matrix Q' of the contracted edge, obtain the vertex position after the edge is contracted;

[0111] S3025, finding all tetrahedrons that originally shared the same vertex with the contracted edge, and updating the coordinates in the vertex information of all tetrahedrons;

[0112] S3026, marking tetrahedrons whose vertices are contracted to the same position as tetrahedrons to be deleted and removing them in subsequent data structure updates; this is because tetrahedrons whose vertices are contracted to the same position have a volume that becomes zero or a geometric shape that is degenerate, and therefore need to be marked for deletion;

[0113] S303, data structure update, including:

[0114] S3031, updating the vertex list, that is, removing all vertices that no longer exist in any tetrahedron from the vertex list to ensure that the vertex list contains only valid vertices;

[0115] S3032, updating the edge list, that is, removing all edges that no longer exist in any tetrahedron, to ensure that each edge in the edge list is valid;

[0116] S3033, updating the tetrahedron list, that is, removing the vertex index of the tetrahedron marked as to be deleted;

[0117] S304, calculate the simplification ratio of the current grid after this iteration, and determine whether step S3 is completed:

[0118] 1) If the preset simplification ratio or number of iterations is reached, step S404 ends and the final updated data structure is output;

[0119] 2) If the preset simplification ratio or number of iterations is not reached, steps S302 to S304 are repeated;

[0120] In step S304, the simplification ratio of the mesh is defined as: the ratio of the cumulative number of tetrahedrons marked for deletion to the total number of initial tetrahedrons during the continuous iteration process; the number of iterations is set to be ≥1000.

[0121] As a preferred technical solution of this embodiment, when performing the simplification process of step S3, a parallel computing method is used to process the contraction operation of each edge in parallel, so as to quickly complete the contraction of the edge while maintaining the mesh topology structure.

[0122] Specifically, the parallel processing step uses a parallel computing library (such as OpenMP) to set the number of threads and allocation strategy for parallel computing to ensure that the edge contraction operation can be executed simultaneously on multiple processor cores, thereby improving the processing speed. The specific steps include:

[0123] S301, assigning edges to threads: assigning all edges to be processed to different threads based on a set random extraction strategy, ensuring that each thread processes the same or similar number of edges;

[0124] S302, executing the shrinking operation in parallel: each thread independently and in parallel executes the edge shrinking operation, calculates the new vertex position, and updates the vertex and edge information;

[0125] S303, synchronization operation: after each round of parallel shrinking operation is completed, a synchronization operation is performed to ensure that the results of all threads are consistent and to prepare for the next round of edge shrinking operation.

[0126] In order to further demonstrate the effectiveness of the large-scale tetrahedron parallel deletion method based on quadratic error metric of the present invention in large-scale tetrahedron data processing, the method of the present invention is used to compare the results of processing the same large-scale tetrahedron with the VTK method currently commonly used for large-scale tetrahedron processing.

[0127] like Figure 2 and Figure 3 The figure shows a comparison diagram of the verification of the data processing effects of the method of the present invention and the VTK method using tetBox with a data set of 1 million grid data and 100,000 grid data respectively.

[0128] like Figure 2 As shown, the mesh obtained by the method of the present invention has a high integrity and is almost consistent with the original mesh, which verifies the effectiveness and practicality of the method of the present invention to a large extent. Although it can be found in detail that the method of the present invention has some flaws at the junction of the two surfaces, the method of the present invention has significant advantages over the simplification method of VTK. In addition, the number of meshes simplified by the method of the present invention is reduced by three times compared with the VTK method. These results are attributed to the k-function penalty term proposed in the present invention. On smooth surfaces where data changes are not obvious, the method of the present invention can greatly reduce the number of meshes, while in areas where planes meet and data suddenly change, the method of the present invention can reduce the amount of simplification and more accurately retain the meshes of these parts. Therefore, the method of the present invention can present a more standard rendering result than the VTK algorithm with a smaller number of meshes.

[0129] like Figure 3 As shown in the figure, the method proposed in the present invention uses a data set with a very small amount of data. It verifies its versatility for small-scale data. The data only has 120,000 grids. In such a small grid, it is obvious that the quality of grid compression is obviously declining. It is obvious that there are several gaps on the edges where the faces intersect. And there are obvious wrinkles and ripples in the grooves. The effect is not ideal. And the reduction of VTK can hardly reach the usable level. It can be seen that there are obvious defects. By comparing with Figure 2 The obvious differences can be found by comparing them. They used more than 1 million grid data, which is about 10 times the number of grids.

[0130] In summary, the method of the present invention can efficiently simplify large-scale tetrahedral meshes while ensuring that the simplified meshes have high geometric fidelity and computational efficiency.

Claims

1. A large-scale tetrahedron parallel deletion method based on quadratic error metric, characterized in that: The steps are: S1, read the tetrahedron mesh data including vertex coordinates, vertex information and vertex index of the tetrahedron, and complete the data structure initialization by parsing the tetrahedron mesh data; S2. Define the boundary surface and set the quadratic error matrix Q of the vertices on the non-boundary surface of the tetrahedron and the quadratic error matrix Q of the vertices on the boundary surface respectively. b , used to evaluate the shrinkage error of each edge; S3, randomly extract multiple edges of tetrahedrons, adopt the method of shrinking the edges in a way that minimizes the error, and delete the tetrahedrons related to the shrunk edges, and iteratively update the data structure until a preset simplification ratio or number of iterations is reached; wherein, the method of shrinking the edges in a way that minimizes the error, and deletes the tetrahedrons related to the shrunk edges is as follows: 1) randomly extract multiple tetrahedrons, and randomly extract an edge from each tetrahedron; 2) for each edge, respectively calculate the quadratic error matrix Q of the vertices at both ends of the edge, and obtain a new quadratic error matrix Q' by merging the quadratic error matrix Q of the vertices at both ends of each edge; 3) according to the calculation result of step S301, select the edge with the minimum quadratic error matrix Q' from the multiple randomly extracted edges as the shrunk edge; 4) based on the quadratic error matrix Q' of the shrunk edge, obtain the vertex position after the edge is shrunk; 5) update the coordinates in the vertex information of the tetrahedron that originally shared the same vertex with the shrunk edge; 6) mark the tetrahedron whose vertices are shrunk to the same position as the tetrahedron to be deleted and remove it in the subsequent data structure update.

2. The large-scale tetrahedron parallel deletion method based on quadratic error metric according to claim 1, characterized in that: In step S1, parsing and completing the initialization of data structures includes: parsing vertex coordinates into floating-point arrays, parsing vertex indices of tetrahedrons into integers, creating a vertex list for storing coordinate information of all vertices, creating a tetrahedron list for storing vertex indices of tetrahedrons, creating an edge list for storing vertex index information of each edge in the tetrahedron, and initializing data structures related to tetrahedral mesh processing, including quadratic error matrices of edges and error metrics of tetrahedrons.

3. The large-scale tetrahedron parallel deletion method based on quadratic error metric according to claim 2, characterized in that: In the created vertex list, all vertex coordinates and vertex-related information obtained by parsing are stored one by one in the vertex list, and each vertex has a unique index in the list. Vertex-related information includes physical quantities and colors; in the created tetrahedron list, the vertex indexes of all tetrahedrons are stored one by one in the table; in the created edge list, the vertex index information of each edge in the tetrahedron is stored one by one in the table.

4. The large-scale tetrahedron parallel deletion method based on quadratic error metric according to claim 1, characterized in that: In step S2, the boundary surface is defined as: a surface formed by coordinates of three points that do not have two identical points forming the plane; wherein, (1) For the vertices on the non-boundary surfaces of the tetrahedron, the quadratic error matrix Q is defined as: Suppose there is a point p and a neighboring point x in a high-dimensional space, and we need to calculate the error from x to several planes; Set the basis vectors [e1,e2,…,e d ] forms the tangent plane where p is located, [e d+1 ,…,e n ] is perpendicular to the plane where p lies; Then the error function is defined, that is, the square of the distance from point x to the plane is expressed as: Choose a set of basis vectors [e i ] so that the tangent plane [e i ] remains unchanged, and we get: definition: but, The expression of the quadratic error matrix Q for each vertex of each tetrahedron is: Q p (x)=x T Ax-2x T Ap+p T Ap, In the above two formulas, b=-Ap,c=p T Ap; (2) For the vertices on the boundary surface of the tetrahedron, the quadratic error matrix Q b Defined as: Assume (v1, v2, v3) are the adjacent boundary vertices of boundary vertex v1, and assume that the plane normal vectors are n1 and n2, then define the k function by the cross product of the normal vectors of adjacent boundary surfaces: Furthermore, the quadratic error matrix is ​​given by Q b The expression is: Q b (v1)=Q(v1)+W b *k(v1), <h2 style=";text-align:left;direction:ltr">Q<h2 style=";text-align:left;direction:ltr"> b <h2 style=";text-align:left;direction:ltr"> (v2) = Q(v2) + W<h2 style=";text-align:left;direction:ltr"> b <h2 style=";text-align:left;direction:ltr"> *k(v2), In the formula, Q b (v1) and Q b (v2) is the quadratic error function at the boundary surface vertices v1 and v2; W b is a user-defined weight factor; k(v1) and k(v2) are the cross products of the boundary normal vectors at vertices v1 and v2.

5. The large-scale tetrahedron parallel deletion method based on quadratic error metric according to claim 1, characterized in that: The specific implementation steps of step S3 are: S301, before each edge contraction operation starts, initialize related data structures, such as a vertex list, a tetrahedron list, and an edge list; set a counter for recording the number of deleted tetrahedrons, and initialize the simplification ratio and the number of iterations; S302, performing edge shrinkage and tetrahedron deletion operations by adopting a method of shrinking the edge in a way that minimizes the error and deleting the tetrahedrons associated with the shrunk edge: S303, data structure update, including updating the vertex list, updating the edge list and updating the tetrahedron list; S304, calculate the simplification ratio of the current grid after this iteration, and determine whether step S3 is completed: 1) If the preset simplification ratio or number of iterations is reached, step S404 is completed and the final updated data structure is output; 2) If the preset simplification ratio or number of iterations is not reached, steps S302 to S304 are repeated.

6. The large-scale tetrahedron parallel deletion method based on quadratic error metric according to claim 1, characterized in that: Step S3 is executed by using a parallel computing method through a parallel computing library, and the steps are as follows: S301, assigning edges to threads: assigning all edges to be processed to different threads according to a set strategy, ensuring that each thread processes the same or similar number of edges; S302, executing the shrinking operation in parallel: each thread independently and in parallel executes the edge shrinking operation, calculates the new vertex position, and updates the vertex and edge information; S303, synchronization operation: after each round of parallel shrinking operation is completed, a synchronization operation is performed to ensure that the results of all threads are consistent and to prepare for the next round of edge shrinking operation.