A three-dimensional model discretization method and electronic device for furniture design

By constructing the graph structure and adaptive voxelization of furniture components connection relationships, the node accuracy is gradually optimized, and the problem of insufficient discretization accuracy of furniture design tetrahedrons in the existing technology is solved, and efficient human-causing engineering analysis is achieved.

CN118709456BActive Publication Date: 2024-11-15ENERGY RES INST OF JIANGXI ACAD OF SCI +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411198218.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-11-15
Estimated Expiration
2044-08-29

AI Technical Summary

Technical Problem

The existing furniture design software is difficult to meet the accuracy requirements during the process of tetrahedral discretization, resulting in large amounts of human-cause engineering analysis and calculation, and the accuracy requirements for different furniture components in the analysis are different, so the existing methods are difficult to meet.

Method used

By constructing a graph structure based on the connection relationship of furniture components, the initial grid is generated adaptively voxelized, and the accuracy difference is propagated through the iterative process, and the node accuracy is gradually optimized until the accuracy requirements are met.

Benefits of technology

The accuracy division of different furniture components is realized, the discretization efficiency and accuracy are improved, and the rapid human-causing engineering analysis is supported.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118709456B_ABST
    Figure CN118709456B_ABST
Patent Text Reader

Abstract

The present invention relates to a three-dimensional model discretization method and electronic device for furniture design. The method constructs a furniture model based on B-Rep representation as a graph and performs adaptive voxelization; then, all nodes in the graph are sorted from small to large according to node parameters to obtain a node sequence, and the mesh corresponding to each node is used as a boundary mesh to perform surface iteration and internal iteration respectively, and the average difference vector of the iteration is propagated to adjacent nodes through the nodes of the graph; finally, it is determined whether the accuracy requirement is met. The present invention can convert a furniture model based on B-Rep representation into a discretized representation based on tetrahedron, thereby supporting the ergonomic analysis of furniture, and has broad application prospects.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of computer graphics, and in particular, relates to a three-dimensional model discretization method and electronic equipment for furniture design. Technical Background

[0002] In the furniture design process, designers usually need to make trade-offs between multiple design schemes and design parameters. The results of the ergonomics analysis of the design model largely determine the choice of design schemes and design parameters. Although the existing furniture design software provides three-dimensional design functions, it has limitations in the tetrahedron discretization process for furniture. The shapes of furniture components vary. The general uniform sampling method is difficult to achieve the established accuracy requirements under the appropriate discrete tetrahedron scale representation. When the established accuracy is achieved, a large number of discrete tetrahedron expressions will be obtained, which makes the subsequent ergonomics analysis process very computationally intensive and difficult to apply in the field of furniture design. Different furniture components have different importance in the ergonomics analysis process. For example, for chair design, the shape expression of the back and seat is more important than the chair legs, and they also require higher accuracy under ergonomics analysis. Summary of the invention

[0003] In order to overcome these challenges, the present invention provides a three-dimensional model discretization method for furniture design. The method can construct a graph structure according to the composition relationship of furniture components, and then automatically / manually calibrate the initial discrete accuracy requirements of each graph node, and adaptively voxelize and generate an initial grid according to the initial accuracy requirements based on the B-Rep expression; through the initial grid, the nodes are sorted, and the initial grid corresponding to the sorted nodes is iterated. During the iteration process, the accuracy difference will propagate to the adjacent nodes and affect the iteration of the node. After all nodes are iterated, the current round of accuracy iteration is completed. If the established accuracy requirements are met, the iteration is exited, otherwise the next round of iteration is performed.

[0004] The present invention provides a three-dimensional model discretization method for furniture design, which specifically includes the following steps:

[0005] S1. Each entity of the furniture model represented by B-Rep is regarded as a furniture component, each entity is represented as a node of the graph, the edge of the graph represents the connection between entities, and the graph is constructed according to the connection relationship between entities;

[0006] S2. each entity is represented by a medial axis representation method, and adaptive voxelization is performed based on the medial axis representation;

[0007] S3. Sort all nodes in the graph according to the node parameters from small to large to obtain a node sequence, use the mesh corresponding to each node as the boundary mesh to perform surface iteration and internal iteration respectively, and propagate the average difference vector of the iteration to the adjacent nodes through the nodes of the graph;

[0008] S4. Calculate the accuracy of each node of the current discrete result obtained in each internal iteration to determine whether it meets the accuracy requirement. If not, perform the next round of surface iteration and internal iteration. If it meets the accuracy requirement, remove the current node from the graph and remove all edges connected to the node. Determine whether there are other nodes. If there are other nodes in the graph, start the next round of iteration from step S3 until there are no nodes in the graph and the discretization is completed.

[0009] Further preferably, in step S1, the connection relationship between the entities is manually calibrated, or the connection relationship between the entities is automatically determined by a nearest neighbor patch threshold.

[0010] More preferably, in step S2, the shortest distance represented by the medial axis is divided by 2 as the side length of the voxel, and then voxelization is performed.

[0011] Further preferably, in the step S2, feature voxels are identified during the voxelization process, i.e., voxels on the surface of the solid are taken as feature voxels, and it is determined whether there are non-boundary feature voxels. If so, they are adaptively subdivided. If not, it is determined whether each feature voxel contains only a connected sub-mesh on the triangular mesh. If so, it is adaptively subdivided. After the adaptive subdivision, it returns to determine whether there are non-boundary feature voxels. If not, the voxelization process is terminated.

[0012] Further preferably, in the step S2, all voxels are divided into three categories, namely, external voxels, internal voxels and boundary voxels, by using a voxelization process and a scan conversion algorithm.

[0013] Further preferably, the surface iteration process is: according to the boundary grid of the extracted voxel or tetrahedral grid , for the boundary grid Perform several subdivisions to generate the approximate surface; The subdivision mesh generated after the subdivision is ; respectively record the boundary grid And a given triangular mesh The mesh vertices are control vertices and data points ; In each iteration, the control vertices in three-dimensional space are Map to feature space and find data points In feature space and subdivided grid The nearest vertex on the surface , and calculate the data points With Vertex The difference vector , Is a subdivision grid The vertices on the boundary mesh Control Vertices on Linear combination representation, assigning weighted difference vectors to control vertices Next, for each control vertex , take the weighted average of the assigned difference vectors to get the average difference vector; finally, add the average difference vector to the control vertex to generate a new boundary mesh of the control vertices.

[0014] Further preferably, the internal iteration process is: layering according to the connection relationship between internal vertices and boundaries, and moving vertices layer by layer starting from the boundary; the internal vertices in a ring neighborhood of the control vertex are divided into the first layer; among the remaining internal vertices, the internal vertices adjacent to a ring of first-layer vertices are the second layer, and so on; then, the position of the control vertex is fixed, and the first-layer internal vertices are moved to the new position according to the given average difference vector operation, while ensuring that the movement is performed within the feasible domain, next, the position of the first-layer internal vertices is fixed, and the position of the second-layer internal vertices is moved, and so on, until all levels of vertices are moved; this process is iterated until the ratio of the vertex movement distance to the diagonal of the bounding box is less than a preset threshold; to ensure the validity of the tetrahedral mesh, each iteration step needs to be restricted to the feasible domain; each internal vertex is adjacent to several tetrahedrons, and the movement of the vertex will affect the directed volume of the adjacent tetrahedral meshes; in order to keep the sign of the directed volume of the vertex adjacent to the tetrahedral mesh unchanged, the movement of the vertex should be limited to the same side of the plane opposite to the vertex, that is, within the polyhedron surrounded by all adjacent tetrahedral meshes.

[0015] The present invention also provides a non-volatile computer storage medium, which stores computer executable instructions. The computer executable instructions can execute the above-mentioned three-dimensional model discretization method for furniture design.

[0016] The present invention also provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned three-dimensional model discretization method for furniture design.

[0017] The present invention establishes a graph structure based on the connection relationship between each entity of the furniture design model: the graph structure of the present invention supports discretization processing, and can divide the various parts of the furniture according to the established requirements, so that the precision of different entities is different, so as to meet the different precision requirements of furniture ergonomics analysis.

[0018] The present invention is based on the calculation transfer of the graph structure: based on the graph structure, the present invention sorts the discrete precision according to the requirements, and transfers the average adjustment interpolation of the intermediate calculation according to the graph structure, which effectively improves the overall convergence speed of the discrete process.

[0019] The beneficial effects of the present invention include:

[0020] Improved discrete flexibility: Designers can adjust the discrete accuracy of each design entity according to the specific needs of human factors engineering analysis to achieve more suitable discrete data and support rapid human factors engineering analysis.

[0021] Improved discretization speed: By expressing the relationship between furniture entities as a graph structure and transmitting intermediate iteration results, high-precision discretization control can be achieved to ensure the discretization efficiency and quantity of furniture entities. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described in detail below with reference to the accompanying drawings.

[0024] Reference Figure 1 , a three-dimensional model discretization method for furniture design, specifically comprising the following steps:

[0025] S1. Each entity of the furniture model represented by B-Rep is regarded as a furniture component. Each entity is represented as a node of a graph. The edge of the graph represents the connection between entities. The graph is constructed according to the connection relationship between entities. The connection relationship between entities can be manually calibrated, or the connection relationship between entities can be automatically determined by the nearest neighbor patch threshold.

[0026] S2. Adaptively voxelize the entity: Each entity is represented by the medial axis through the medial axis representation, and adaptive voxelization is performed based on the medial axis representation. The adaptive method is to divide the shortest distance represented by the medial axis by 2 as the side length of the voxel, and then voxelize. In the voxelization process, the feature voxels are identified, that is, the voxels on the surface of the entity are taken as feature voxels, and it is determined whether there are non-boundary feature voxels. If so, adaptive subdivision is performed. If not, it is determined whether each feature voxel contains only a connected sub-mesh on the triangular mesh. If so, adaptive subdivision is performed. After adaptive subdivision, it is returned to determine whether there are non-boundary feature voxels. If not, the voxelization process is terminated. Through the voxelization process, all voxels are divided into three categories using the scan conversion algorithm: 1) external voxels, which are voxels located outside the surface of the entity; 2) internal voxels, which are voxels located inside the surface of the entity; 3) boundary voxels, which are feature voxels adjacent to external voxels. Boundary voxels must be feature voxels, but feature voxels are not necessarily boundary voxels.

[0027] S3. Sort all nodes in the graph according to the node parameters from small to large to obtain a node sequence, and use the mesh corresponding to each node as the boundary mesh to perform surface iteration and internal iteration respectively.

[0028] S3.1 Surface iteration: Extract boundary meshes from voxels or tetrahedral meshes , for the boundary grid Perform several subdivisions to generate the approximate surface. The subdivision mesh generated after the subdivision is . The boundary grid is And a given triangular mesh The mesh vertices are control vertices and data points ; In each iteration, the control vertices in three-dimensional space are Map to feature space and find data points In feature space and subdivided grid The nearest vertex on the surface , and calculate the data points With Vertex The difference vector , Is a subdivision grid The vertices on the boundary mesh Control Vertices on Linear combination representation, so the weighted difference vectors can be assigned to the control vertices Next, for each control vertex , take the weighted average of the assigned difference vectors to get the average difference vector; finally, add the average difference vector to the control vertex to generate a new boundary mesh of the control vertices.

[0029] S3.2 Internal iteration: Each iteration of the boundary mesh needs to be diffused to the interior of the tetrahedral mesh. Layers are formed according to the connection relationship between the internal vertices and the boundary, and the vertices are moved layer by layer starting from the boundary. The internal vertices in the one-ring neighborhood of the control vertex are divided into the first layer; among the remaining internal vertices, the internal vertices that are adjacent to the first-layer vertices in one ring are the second layer, and so on. Then, the position of the control vertex is fixed, and the first-layer internal vertices are moved to the new position according to the given average difference vector operation, while ensuring that the movement is performed within the feasible domain. Next, the position of the first-layer internal vertices is fixed, and the position of the second-layer internal vertices is moved, and so on, until all levels of vertices are moved; this process is iterated until the ratio of the vertex movement distance to the diagonal of the bounding box is less than the preset threshold. To ensure the validity of the tetrahedral mesh, each iteration needs to be restricted to the feasible domain. Each internal vertex is adjacent to several tetrahedral meshes, and the movement of the vertex will affect the directed volume of the adjacent tetrahedral meshes. In order to keep the sign of the directed volume of the tetrahedral meshes adjacent to the vertex unchanged, the movement of the vertex should be limited to the same side of the plane opposite to the vertex, that is, within the polyhedron enclosed by all adjacent tetrahedral meshes.

[0030] S3.3 Transfer update: Propagate the mean difference vector through the nodes of the graph to the neighboring nodes to control the node accuracy of the current iteration round.

[0031] S4. Accuracy judgment: Calculate the accuracy of each node of the current discrete result obtained in each internal iteration to determine whether it meets the accuracy requirements. If it does not meet the accuracy requirements, perform the next round of surface iteration and internal iteration. If it meets the accuracy requirements, remove the current node from the graph and all edges connected to the node. Determine whether there are other nodes. If there are other nodes in the graph, start the next round of iteration from step S3 until there are no nodes in the graph and the discretization is completed.

[0032] The following is a detailed explanation of the design of a chair as an example:

[0033] First, the 3D model of the armchair is imported and converted into the boundary representation (B-Rep) format. In this format, the various components of the armchair (such as the seat, backrest, armrests, legs, etc.) are clearly represented as different entities, and the geometric boundaries and topological relationships between each entity are also clearly defined. Then, each entity is regarded as a node in the graph, and the graph is constructed based on the actual connection relationships between entities. These connection relationships can be manually calibrated between the connection relationships between each entity, or automatically determined by the nearest neighbor patch threshold.

[0034] Next, each entity of the chair is adaptively voxelized. The medial axis representation of each entity is extracted through the medial axis transform (MAT) technique as the basis for voxelization. Based on the medial axis representation and the local geometric features of the entity, the initial size of the voxel is dynamically determined, and adaptive subdivision is performed during the voxelization process. In the curved areas of the armrests and backrests, the voxels are set more densely to capture more geometric details; while in straight areas such as the chair legs, the voxels are relatively sparse to save computing resources. At the same time, the voxels located on the surface of the entity are identified as feature voxels. Through the scan conversion algorithm, all voxels are classified into external voxels, internal voxels, and boundary voxels. Note that boundary voxels must be feature voxels, but feature voxels are not necessarily boundary voxels.

[0035] To make the initial boundary grid As close to a triangular mesh as possible , each feature voxel needs to meet the following two conditions: each feature voxel is a boundary voxel; each feature voxel contains only triangular meshes A connected sub-mesh on a given triangular mesh P. If a feature voxel does not meet any of the above conditions, the following consequences will occur: the shape of the boundary mesh V will deviate from the shape of the given triangular mesh P, resulting in distortion of the geometric information; or the boundary mesh V is inconsistent with the triangular mesh P in topological structure, that is, they are not the same. Therefore, when a feature voxel that does not meet the conditions is detected, an adaptive subdivision operation will be performed on it. Specifically, the voxels in the row where this feature voxel is located are divided into two, and the size of the voxels is refined. To avoid generating voxels that are too small, further adaptive subdivision will be stopped when the side length of the subdivided voxel is less than 1 / 8 of the initial side length.

[0036] For each entity component of the armchair, boundary mesh iterative optimization is performed to improve the discretization quality. First, the boundary mesh is initially subdivided to generate a subdivided mesh. Then, the boundary mesh is gradually optimized using surface iteration and internal iteration strategies. In the surface iteration, the subdivided mesh vertices are mapped to the original triangular mesh of the armchair, and the distance errors between the vertices and the original triangular mesh vertices are calculated. These errors are reduced by moving the vertex positions, so that the boundary mesh gradually approaches the design surface. In the internal iteration, the internal vertices are layered according to the connection relationship with the boundary, and the internal structure of the boundary mesh is optimized by moving the vertex positions. The movement process is ensured to be carried out within the feasible domain to maintain the validity of the tetrahedral mesh. For slender structures such as chair legs, special optimization strategies are used to avoid mesh distortion.

[0037] After each iteration, the optimization information obtained in the surface iteration is transferred to the adjacent nodes through the connection relationship of the graph to ensure the coordination and consistency of the mesh optimization between the different components of the armchair.

[0038] The current discretization result is evaluated according to the preset accuracy standards (such as vertex deviation threshold, mesh shape error, etc.). If the evaluation result meets the accuracy requirements, the iteration process is terminated; otherwise, the next round of iterative optimization is continued. In this way, it can be ensured that the discretization result of the armchair 3D model meets the design requirements.

[0039] In another embodiment, a non-volatile computer storage medium is provided, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute a three-dimensional model discretization method for furniture design in the above embodiment.

[0040] This embodiment provides an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a three-dimensional model discretization method for furniture design according to the above embodiment.

[0041] The specific implementation scheme described above further illustrates the purpose, technical scheme and technical effect of the present invention in detail. It should be understood that the above description is only a specific implementation scheme of the present invention and is not intended to limit the scope of the present invention. Any equivalent changes and modifications made by any technician in the field without departing from the ideas and principles of the present invention should fall within the scope of protection of the present invention.

Claims

1. A three-dimensional model discretization method for furniture design, characterized in that: The following steps are involved: S1. Each entity of the furniture model represented by B-Rep is regarded as a furniture component, each entity is represented as a node of the graph, the edge of the graph represents the connection between entities, and the graph is constructed according to the connection relationship between entities; S2. each entity is represented by a medial axis representation method, and adaptive voxelization is performed based on the medial axis representation; S3. Sort all nodes in the graph according to the node parameters from small to large to obtain a node sequence, use the mesh corresponding to each node as the boundary mesh to perform surface iteration and internal iteration respectively, and propagate the average difference vector of the iteration to the adjacent nodes through the nodes of the graph; S4. Calculate the accuracy of each node of the current discrete result obtained in each internal iteration, and determine whether it meets the accuracy requirement. If it does not meet the accuracy requirement, perform the next round of surface iteration and internal iteration. If it meets the accuracy requirement, remove the current node from the graph, and remove all edges connected to the node, and determine whether there are other nodes. If there are other nodes in the graph, start the next round of iteration from step S3 until there are no nodes in the graph and the discretization is completed; The surface iteration process is as follows: Extract the boundary mesh of voxels or tetrahedral meshes , for the boundary grid Perform several subdivisions to generate the approximate surface; The subdivision mesh generated after the subdivision is ; respectively record the boundary grid And a given triangular mesh The mesh vertices are control vertices and data points ; In each iteration, the control vertices in three-dimensional space are Map to feature space and find data points In feature space and subdivided grid The nearest vertex on the surface , and calculate the data points With Vertex The difference vector , Is a subdivision grid The vertices on the boundary mesh Control Vertices on Linear combination representation, assigning weighted difference vectors to control vertices Next, for each control vertex , take the weighted average of the assigned difference vectors to get the average difference vector; finally, add the average difference vector to the control vertex to generate a new boundary mesh The control vertices of The internal iteration process is as follows: the layers are divided according to the connection relationship between the internal vertices and the boundaries, and the vertices are moved layer by layer starting from the boundaries; the internal vertices in a ring neighborhood of the control vertex are divided into the first layer; among the remaining internal vertices, the internal vertices adjacent to the first layer vertices in a ring are the second layer, and so on; then, the position of the control vertex is fixed, and the first layer of internal vertices are moved to the new position according to the given average difference vector operation, while ensuring that the movement is carried out within the feasible domain, next, the position of the first layer of internal vertices is fixed, and the position of the second layer of internal vertices is moved, and so on, until all levels of vertices are moved; this process is iterated until the ratio of the vertex movement distance to the diagonal of the bounding box is less than the preset threshold; to ensure the validity of the tetrahedral mesh, each iteration step needs to be restricted to the feasible domain; each internal vertex is adjacent to several tetrahedrons, and the movement of the vertex will affect the directed volume of the adjacent tetrahedral meshes; in order to keep the sign of the directed volume of the vertex adjacent to the tetrahedral mesh unchanged, the movement of the vertex should be limited to the same side of the plane opposite to the vertex, that is, the polyhedron surrounded by all adjacent tetrahedral meshes.

2. The three-dimensional model discretization method for furniture design according to claim 1, characterized in that: In step S1, the connection relationship between each entity is manually calibrated, or the connection relationship between each entity is automatically determined by the nearest neighbor patch threshold.

3. The three-dimensional model discretization method for furniture design according to claim 1, characterized in that: In step S2, the shortest distance represented by the medial axis is divided by 2 as the side length of the voxel, and then voxelization is performed.

4. The three-dimensional model discretization method for furniture design according to claim 1, characterized in that: In step S2, feature voxels are identified during the voxelization process, that is, voxels on the surface of the solid are taken as feature voxels, and it is determined whether there are non-boundary feature voxels. If so, adaptive subdivision is performed. If not, it is determined whether each feature voxel contains only a connected sub-mesh on the triangular mesh. If so, adaptive subdivision is performed. After adaptive subdivision, it returns to determine whether there are non-boundary feature voxels. If not, the voxelization process is terminated.

5. The three-dimensional model discretization method for furniture design according to claim 1, characterized in that: In step S2, through the voxelization process, all voxels are divided into three categories: external voxels, internal voxels and boundary voxels using a scan conversion algorithm.

6. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions execute the three-dimensional model discretization method for furniture design as described in any one of claims 1-5.

7. An electronic device comprising: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and wherein the instructions are executed by the at least one processor so that the at least one processor executes the three-dimensional model discretization method for furniture design as described in any one of claims 1-5.

Citation Information

Patent Citations

  • Iteration generation method of hexahedral mesh model

    CN103824331A

  • Three-dimensional CAD (computer-aided design) reconstruction method for pixel type topological optimization result

    CN114707388A

  • Automobile modeling design method and device

    CN117010093A

  • Three-dimensional model modeling method and device, electronic equipment and storage medium

    CN118334265A

  • Method and apparatus for fast voxelization of volumetric models

    US6556199B1