Three-dimensional grid geometric refinement method and system based on texture boundary constraint, terminal and medium

Abstract

The invention discloses a three-dimensional grid geometric refinement method and system based on texture boundary constraint, a terminal and a medium, and the method comprises the steps: obtaining a three-dimensional grid model, and carrying out the enhancement processing of a texture image of the three-dimensional grid model; extracting linear boundary features from the enhanced texture image to obtain a texture boundary line segment; mapping the texture boundary line segments to the surface of the three-dimensional grid model and taking the texture boundary line segments as texture boundary constraints, and constructing a candidate point set containing the texture boundary constraints in combination with an original geometric structure of the three-dimensional grid model; and generating a refined three-dimensional mesh model through a constrained mesh construction mode, and realizing consistent alignment of geometric units and texture semantic boundaries. According to the method, a network construction mechanism of texture enhancement and boundary constraint is introduced, so that the photogrammetry three-dimensional grid model realizes consistent alignment of geometric units and texture semantic boundaries under the condition of given geometric resolution, and the three-dimensional grid model with more reasonable structural expression and more accurate boundary description is obtained.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional model representation technology, and in particular to a three-dimensional mesh geometry refinement method, system, terminal, and medium based on texture boundary constraints. Background Technology

[0002] With the development of UAV photogrammetry and multi-view image 3D reconstruction technology, image-based 3D mesh models have been widely used in fields such as urban fine modeling, digital twins, and smart cities. Compared with 2D images or point cloud data, mesh models can simultaneously express continuous geometric shapes and high-resolution texture information, and have significant advantages in terms of realism and structural representation.

[0003] In photogrammetric reconstruction, dense point clouds are typically converted into triangular meshes, which are then further simplified or subdivided. Existing mesh processing methods widely used in engineering and research mainly include mesh simplification methods based on Quadratic Error Metrics (QEM), variational shape approximation methods, and various classical surface subdivision algorithms. These methods primarily optimize for minimizing geometric errors or improving surface smoothness, reducing mesh complexity or enhancing local geometric regularity while preserving the overall geometric shape. However, these geometry-driven mesh processing methods generally do not incorporate significant boundary information from texture images into the meshing constraints. Real-world photogrammetric models often contain clear structural edges in their textures, such as building outlines, road boundaries, and feature boundaries. However, in existing uniform subdivision or simplification processes, triangular patches often span multiple texture and semantic regions, leading to misalignment between geometric units and real feature boundaries. Figure 1 As shown in the image.

[0004] This misalignment between geometric and texture boundaries reduces the accuracy of the model's structural representation and introduces uncertainty and error accumulation in subsequent geometric analysis, spatial measurement, and applications. To address these issues, existing research has attempted to improve model quality through planar constraints, edge-preserving reconstruction, or texture-based local optimization. However, these methods are mostly limited to specific objects or local regions, rely on empirical parameters, and are difficult to generalize to large-scale complex scenes. Furthermore, there is a lack of universal texture boundary constraint mechanisms that directly affect the triangulation process. In addition, photogrammetric texture images are affected by scale, compression, and noise, resulting in insufficient edge feature stability, which also limits the effective use of texture boundaries in geometric refinement.

[0005] Therefore, existing technologies struggle to achieve precise alignment between geometric units and texture structure boundaries while maintaining geometric regularity. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, this invention provides a method, system, terminal, and medium for three-dimensional mesh geometric refinement based on texture boundary constraints. The technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a method for refining the geometry of a three-dimensional mesh based on texture boundary constraints, the method comprising: A three-dimensional mesh model is obtained, and the texture image of the three-dimensional mesh model is enhanced. Linear boundary features are extracted from the enhanced texture image to obtain texture boundary line segments; The texture boundary line segments are mapped onto the surface of the 3D mesh model and used as texture boundary constraints. Combined with the original geometry of the 3D mesh model, a candidate point set containing the texture boundary constraints is constructed. A refined 3D mesh model is generated using a constrained meshing method, achieving consistent alignment between geometric units and texture semantic boundaries.

[0007] In one implementation, the texture image of the three-dimensional mesh model is enhanced, including: The texture image is enhanced using Real-ESRGAN to restore and enhance texture details and edges. Real-ESRGAN, based on a generative adversarial network framework, comprises a generator and a relativistic discriminator. The generator recovers high-frequency details from low-resolution textures, while the relativistic discriminator balances perceptual realism with structural fidelity, ensuring that the enhanced texture is both clear and conforms to the characteristics of the real scene.

[0008] In one implementation, linear boundary features are extracted from the enhanced texture image to obtain texture boundary line segments, including: A line segment detection algorithm is used to extract linear boundary features from the enhanced texture image to obtain texture boundary line segments; The detection scale of the line segment detection algorithm is set to 0.8.

[0009] In one implementation, mapping the texture boundary line segments to the surface of the 3D mesh model and using them as texture boundary constraints includes: For each triangular facet in the 3D mesh model, establish an affine transformation relationship from texture space to 3D space; Based on the affine transformation relationship, each sampling point in the texture boundary line segment is back-projected onto the surface of the three-dimensional mesh model to obtain a three-dimensional constraint edge that is consistent with the spatial position of the real ground object boundary. The three-dimensional constraint edge is used as the texture boundary constraint.

[0010] In one implementation, a candidate point set containing the texture boundary constraints is constructed by combining the original geometry of the 3D mesh model, including: Based on the original geometric structure of the three-dimensional mesh model, the texture boundary line segments mapped to the three-dimensional space are processed, and uniform sampling is performed on each triangular facet boundary to form a boundary sampling point set; The endpoints and boundary sampling point sets of the texture boundary line segments are collected to form a texture constraint point set; The boundary sampling point set and the texture constraint point set are merged, and repeated or nearly repeated points with a distance less than a preset threshold are filtered to obtain the merged boundary constraint point set. Based on the set of boundary constraint points, a candidate set of points containing the texture boundary constraints is constructed.

[0011] In one implementation, based on the boundary constraint point set, constructing a candidate point set containing the texture boundary constraints includes: Within each triangular facet of the 3D mesh model, an internal control point set is generated in an approximately equilateral triangle distribution according to the target spacing. The set of internal control points is merged with the set of boundary constraint points, and duplicate points are filtered again to obtain the set of candidate points.

[0012] In one implementation, a refined 3D mesh model is generated through a constrained meshing method to achieve consistent alignment between geometric units and texture semantic boundaries, including: Based on the candidate point set and texture boundary line segments, a constrained Delaunay triangulation is performed on each triangular facet of the 3D mesh model to generate a refined 3D mesh model. During the mesh construction process, texture-aligned constraint edges are forcibly retained as mesh edges, while new vertices are automatically introduced at the intersections of constraint lines, satisfying the Delaunay geometric optimality condition.

[0013] Secondly, embodiments of the present invention also provide a three-dimensional mesh geometry refinement system based on texture boundary constraints, wherein the system is used to implement the steps of the three-dimensional mesh geometry refinement method based on texture boundary constraints as described in any of the above solutions, and the system includes: The texture enhancement module is used to acquire a three-dimensional mesh model and enhance the texture image of the three-dimensional mesh model. The texture boundary extraction module is used to extract linear boundary features from the enhanced texture image to obtain texture boundary line segments. The constraint point set construction module is used to map the texture boundary line segments to the surface of the three-dimensional mesh model and use them as texture boundary constraints, and to construct a candidate point set containing the texture boundary constraints in combination with the original geometry of the three-dimensional mesh model. The constrained meshing module is used to generate a refined 3D mesh model through a constrained meshing method, achieving consistent alignment between geometric units and texture semantic boundaries.

[0014] Thirdly, embodiments of the present invention also provide a terminal, wherein the terminal includes a memory, a processor, and a three-dimensional mesh geometry refinement program based on texture boundary constraints stored in the memory and executable on the processor. When the processor executes the three-dimensional mesh geometry refinement program based on texture boundary constraints, it implements the steps of the three-dimensional mesh geometry refinement method based on texture boundary constraints of any of the above-mentioned schemes.

[0015] Fourthly, embodiments of the present invention also provide a computer-readable storage medium, wherein a three-dimensional mesh geometry refinement program based on texture boundary constraints is stored on the computer-readable storage medium, and the three-dimensional mesh geometry refinement program based on texture boundary constraints implements the steps of the three-dimensional mesh geometry refinement method based on texture boundary constraints as described in any of the above schemes on the computer-readable storage medium.

[0016] Beneficial Effects: Compared with existing technologies, this invention provides a 3D mesh geometry refinement method based on texture boundary constraints. First, a 3D mesh model is obtained, and its texture image is enhanced. Then, linear boundary features are extracted from the enhanced texture image to obtain texture boundary segments. Next, these texture boundary segments are mapped onto the surface of the 3D mesh model and used as texture boundary constraints. Combined with the original geometry of the 3D mesh model, a candidate point set containing the texture boundary constraints is constructed. Finally, a refined 3D mesh model is generated through a constrained mesh construction method, achieving consistent alignment between geometric units and texture semantic boundaries.

[0017] This invention introduces a mesh construction mechanism that combines texture enhancement and boundary constraints, enabling photogrammetric 3D mesh models to achieve consistent alignment between geometric units and texture semantic boundaries under given geometric resolution conditions. This results in a 3D mesh model with more reasonable structural expression and more accurate boundary characterization, improving the applicability and reliability of the model in quality assessment, structural analysis, and subsequent semantic applications. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating how triangular facets often span multiple texture and semantic regions during the process of uniform subdivision or simplification in existing technologies.

[0019] Figure 2 A flowchart of a preferred embodiment of the three-dimensional mesh geometry refinement method based on texture boundary constraints provided in this invention.

[0020] Figure 3A flowchart illustrating the technical route of the three-dimensional mesh geometry refinement method based on texture boundary constraints provided in this embodiment of the invention.

[0021] Figure 4 This is a schematic diagram of texture image enhancement processing in the three-dimensional mesh geometry refinement method based on texture boundary constraints provided in an embodiment of the present invention.

[0022] Figure 5 This is a schematic diagram of the three-dimensional mesh refinement process based on texture boundary constraints in the three-dimensional mesh geometry refinement method based on texture boundary constraints provided in the embodiments of the present invention.

[0023] Figure 6 This is a schematic diagram of the principle framework of a three-dimensional mesh geometry refinement system based on texture boundary constraints provided in an embodiment of the present invention.

[0024] Figure 7 A schematic diagram of a terminal provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0026] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content, operations, or steps, nor does it require execution in the described order. For example, some operations or steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0027] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0028] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. For example, "first control information" and "second control information" are only used to distinguish different control information and do not limit their order.

[0029] Those skilled in the art will understand that the words "first" and "second" do not limit the quantity or the order of execution, and that the words "first" and "second" do not necessarily imply that they are different.

[0030] It should also be understood that the term “and / or” as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0031] Existing 3D mesh geometry processing methods primarily optimize geometric errors or surface smoothness without incorporating texture semantic boundary constraints, leading to issues such as triangular patches spanning multiple texture semantic regions and inconsistencies between geometric units and real-world feature boundaries. This invention provides a 3D mesh geometry refinement method based on texture boundary constraints. The basic idea is as follows: First, the texture image of the photogrammetric 3D mesh model is enhanced to improve texture details and edge discernibility. Based on this, significant structural boundaries in the texture are extracted to obtain texture boundary segments, which are then mapped onto the 3D mesh surface as geometric constraints. Further, combining the original mesh geometry of the 3D mesh model, a point set containing boundary constraints is constructed under controllable refinement conditions. A new 3D mesh is generated through a constrained mesh construction method, thereby achieving consistent alignment between geometric units and texture semantic boundaries.

[0032] Specifically, the 3D mesh geometry refinement method based on texture boundary constraints in this embodiment can be applied to terminals, such as computers and other intelligent product terminals. In practical applications, such as... Figure 2 As shown in the figure, the 3D mesh geometry refinement method based on texture boundary constraints in this embodiment includes the following steps: Step S100: Obtain a three-dimensional mesh model and perform enhancement processing on the texture image of the three-dimensional mesh model.

[0033] Combination Figure 3 As shown, this embodiment first acquires a 3D mesh model, obtaining a texture image and a geometric mesh. In photogrammetric 3D models, texture images are often limited by imaging scale, compression processing, and noise, resulting in the loss of high-frequency details, making it difficult to use boundaries extracted directly from the original texture as reliable geometric constraints. Therefore, this embodiment requires enhancing the texture image of the 3D mesh model.

[0034] In practical applications, Real-ESRGAN (a deep learning super-resolution reconstruction algorithm) is used to enhance the texture image, restoring and improving texture details and edges. This step restores and enhances linear features such as building outlines and road boundaries, improving edge contrast and providing clear and stable input for subsequent boundary detection. Specifically, as shown below... Figure 4As shown in the diagram. This embodiment of Real-ESRGAN is based on a generative adversarial network framework and consists of a generator and a relativistic discriminator. The generator is used to recover high-frequency details from low-resolution textures, while the relativistic discriminator balances perceptual realism and structural fidelity, ensuring that the enhanced texture is both clear and conforms to the characteristics of a real scene. The parameter settings include an upsampling factor of 2 (factor=2). This parameter improves edge contrast and detail clarity while avoiding high-frequency noise caused by over-sharpening. Higher upsampling factors may introduce unstable edge features, affecting the accuracy of subsequent line segment detection.

[0035] This invention enhances photogrammetric texture images before geometric meshing or thinning to improve texture details and edge discernibility, thereby ensuring the stability and reliability of subsequently extracted texture boundaries as geometric constraints. This technical feature distinguishes it from existing techniques that directly use the original texture or ignore texture quality.

[0036] Step S200: Extract linear boundary features from the enhanced texture image to obtain texture boundary line segments.

[0037] This embodiment automatically extracts significant linear boundary features from the enhanced texture image. These boundaries correspond to semantic boundaries such as road boundaries and feature boundaries in real-world scenes, serving as constraints for subsequent geometric refinement. This embodiment employs a Line Segment Detector (LSD) algorithm to extract linear boundary features from the enhanced texture image, obtaining texture boundary line segments. The detection scale of the LSD algorithm is set to 0.8, which controls the sensitivity of edge detection and the granularity of feature extraction. Based on step S200, significant structural boundaries in the texture can be automatically extracted, providing reliable texture boundary line segments for subsequent steps. .

[0038] Furthermore, the LSD algorithm is used in this invention to obtain texture or structural boundaries, and it is not limited to traditional line segment detection methods based on gradients or geometric features. In other implementations, deep learning-based structural wireframe or boundary extraction methods, such as HAWP (Holistically-Attracted Wireframe Parsing), can also be used to extract semantically consistent linear or curved structural boundaries from the overall image structure.

[0039] Step S300: Map the texture boundary line segments to the surface of the three-dimensional mesh model and use them as texture boundary constraints. Combine the original geometry of the three-dimensional mesh model to construct a candidate point set containing the texture boundary constraints.

[0040] Since the texture boundary segments detected in step S200 are located in a two-dimensional texture space, these texture boundary segments need to be mapped to the surface of the three-dimensional mesh model as constraints for subsequent geometric refinement. For each triangular facet in the three-dimensional mesh model, this embodiment establishes an affine transformation relationship from texture space to three-dimensional space. Specifically, let the three vertices of the triangular facet be: The corresponding texture coordinates are as follows: , in: This represents the homogeneous vector of texture coordinates for the k-th vertex (k=1, 2, 3). Let x and y be the x and y coordinates of the k-th vertex on the texture image, respectively. The three-dimensional spatial coordinates are as follows: ,in: Represents the three-dimensional geometric coordinates of the k-th vertex (k=1,2,3). It represents the three-dimensional real number space.

[0041] The affine transformation relation is expressed as: (1) Where B is the three-dimensional affine transformation matrix. For any texture sampling point located within the texture region of this triangular facet. Its corresponding position in three-dimensional space is given by the following formula: (2) in: This represents the coordinates of the i-th sampling point after mapping to three-dimensional space, where i is the index number of the sampling point. Let be the texture coordinate vector of the i-th sampling point. Let x and y be the horizontal and vertical coordinates of the i-th sampling point on the texture image, respectively. B is the affine transformation matrix obtained in formula (1).

[0042] Through the above affine transformation relationship, this embodiment can transform texture boundary lines. Each sampling point in the model is back-projected onto the surface of the three-dimensional mesh model to obtain a three-dimensional constraint edge that is consistent with the spatial position of the real ground object boundary. The three-dimensional constraint edge is used as the texture boundary constraint.

[0043] Unlike existing methods that rely solely on boundary information in two-dimensional texture space or on three-dimensional geometric features for mesh construction, this embodiment explicitly maps the structural boundaries obtained from texture images or semantic results to three-dimensional constraint edges on the three-dimensional mesh surface through the correspondence between texture space and three-dimensional geometric space. In this way, the generation position of the triangular mesh is directly constrained by the structural boundaries, preventing individual triangular faces from spanning multiple texture or semantic regions, thereby achieving consistent alignment between geometric units and the boundaries of real-world features.

[0044] Step S300: Map the texture boundary line segments to the surface of the three-dimensional mesh model and use them as texture boundary constraints. Combine the original geometry of the three-dimensional mesh model to construct a candidate point set containing the texture boundary constraints.

[0045] Before constrained mesh construction, a set of candidate points needs to be constructed that simultaneously reflects the original geometric structure, texture boundary constraints, and the regularity within the patches. Specifically, refer to... Figure 5 As shown, based on the original geometry of the 3D mesh model, this embodiment processes the texture boundary segments mapped to 3D space and uniformly samples them on the boundary of each triangular facet to form a boundary sampling point set. Simultaneously, the endpoints and boundary sampling point sets of texture boundary line segments are aggregated to form a texture constraint point set. Next, the boundary sampling point set and the texture constraint point set are merged, and duplicate or nearly duplicate points with a distance less than a preset threshold are filtered out to obtain the merged boundary constraint point set. ,as follows: (3) in, Indicates based on distance threshold Point removal and filtering operations are performed. A candidate point set containing the texture boundary constraints is constructed based on the boundary constraint point set.

[0046] Specifically, in this implementation, within each triangular facet of the 3D mesh model, according to the target spacing... The internal control point set is generated using an approximately equilateral triangle distribution. Then, the set of internal control points is merged with the set of boundary constraint points, and duplicate points are filtered again to obtain the candidate point set, as shown below: (4) For the point merging and deduplication operation (defined as in formula (3)), this embodiment can adjust the target spacing. This enables the layering and control of the degree of geometric refinement.

[0047] Step S400: Generate a refined 3D mesh model through a constrained mesh construction method to achieve consistent alignment between geometric units and texture semantic boundaries.

[0048] In practical applications, this embodiment can be based on a candidate point set. and texture boundary lines Perform constrained Delaunay triangulation on each triangular facet of the 3D mesh model to generate a refined 3D mesh model, as shown below: (5) Among them, This represents the Delaunay triangulation operator. During mesh construction, texture-aligned constraint edges are forcibly preserved as mesh edges, while new vertices are automatically introduced at the intersections of constraint lines, satisfying the Delaunay geometric optimality condition. The specific process is as follows: Figure 5 As shown in the image.

[0049] The sampling density and meshing strategy used in this embodiment are not limited to a sampling method based on an approximately equilateral triangle distribution. In other implementations, Poisson disk sampling, random sampling, or other adaptive sampling strategies can also be used to adjust the sampling density according to the location of structural boundaries, texture complexity, or semantic changes.

[0050] As can be seen, based on the introduction of structural boundary constraints, this embodiment further constructs a candidate point set that integrates boundary sampling points, texture constraint points, and control points inside the patch, and increases the point density near the structural boundary, while maintaining a relatively uniform geometric distribution in non-critical areas. Then, a new three-dimensional mesh model is generated through a constrained mesh construction method.

[0051] In summary, this invention first obtains a 3D mesh model and enhances the texture image of the 3D mesh model. Then, linear boundary features are extracted from the enhanced texture image to obtain texture boundary segments. Next, the texture boundary segments are mapped onto the surface of the 3D mesh model and used as texture boundary constraints. Combined with the original geometry of the 3D mesh model, a candidate point set containing the texture boundary constraints is constructed. Finally, a refined 3D mesh model is generated through a constrained mesh construction method, achieving consistent alignment between geometric units and texture semantic boundaries.

[0052] This invention introduces a mesh construction mechanism that combines texture enhancement and boundary constraints, enabling photogrammetric 3D mesh models to achieve consistent alignment between geometric units and texture semantic boundaries under given geometric resolution conditions. This results in a 3D mesh model with more reasonable structural expression and more accurate boundary characterization, improving the applicability and reliability of the model in quality assessment, structural analysis, and subsequent semantic applications.

[0053] Based on the above embodiments, the present invention also provides a three-dimensional mesh geometry refinement system based on texture boundary constraints. This system is used to implement the steps in the above method embodiments, such as... Figure 6As shown, the system includes: a texture enhancement module 10, a texture boundary extraction module 20, a constraint point set construction module 30, and a constraint mesh construction module 40. The texture enhancement module 10 is used to acquire a 3D mesh model and enhance the texture image of the 3D mesh model. The texture boundary extraction module 20 is used to extract linear boundary features from the enhanced texture image to obtain texture boundary line segments. The constraint point set construction module 30 is used to map the texture boundary line segments onto the surface of the 3D mesh model as texture boundary constraints, and, combined with the original geometric structure of the 3D mesh model, construct a candidate point set containing the texture boundary constraints. The constraint mesh construction module 40 is used to generate a refined 3D mesh model through a constrained mesh construction method, achieving consistent alignment between geometric units and texture semantic boundaries.

[0054] The principle of each module in the texture boundary constraint-based 3D mesh geometry refinement system of this embodiment is the same as that of each step in the above method embodiment, and will not be elaborated further here.

[0055] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 7 As shown. The terminal may include one or more processors 100 ( Figure 7 (Only one is shown in the diagram), memory 101, and computer program 102 stored in memory 101 and executable on one or more processors 100. For example, a 3D mesh geometry refinement program based on texture boundary constraints. When one or more processors 100 execute computer program 102, they can implement various steps in the embodiment of the 3D mesh geometry refinement method based on texture boundary constraints. Alternatively, when one or more processors 100 execute computer program 102, they can implement the functions of various modules / units in the embodiment of the 3D mesh geometry refinement system based on texture boundary constraints, without limitation herein.

[0056] In one embodiment, the processor 100 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0057] In one embodiment, memory 101 may be an internal storage unit of an electronic device, such as a hard drive or RAM. Memory 101 may also be an external storage device of the electronic device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital Card (SD), or Flash Card. Furthermore, memory 101 may include both internal and external storage units. Memory 101 is used to store computer programs and other programs and data required by the terminal. Memory 101 can also be used to temporarily store data that has been output or will be output.

[0058] Those skilled in the art will understand that Figure 7 The block diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0059] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), direct memory bus RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for geometric refinement of three-dimensional meshes based on texture boundary constraints, characterized in that, The method includes: A three-dimensional mesh model is obtained, and the texture image of the three-dimensional mesh model is enhanced. Linear boundary features are extracted from the enhanced texture image to obtain texture boundary line segments; The texture boundary line segments are mapped onto the surface of the 3D mesh model and used as texture boundary constraints. Combined with the original geometry of the 3D mesh model, a candidate point set containing the texture boundary constraints is constructed. A refined 3D mesh model is generated using a constrained meshing method, achieving consistent alignment between geometric units and texture semantic boundaries.

2. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 1, characterized in that, Enhancement processing of the texture image of the three-dimensional mesh model includes: The texture image is enhanced using Real-ESRGAN to restore and enhance texture details and edges. Real-ESRGAN, based on a generative adversarial network framework, comprises a generator and a relativistic discriminator. The generator recovers high-frequency details from low-resolution textures, while the relativistic discriminator balances perceptual realism with structural fidelity, ensuring that the enhanced texture is both clear and conforms to the characteristics of the real scene.

3. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 1, characterized in that, Linear boundary features are extracted from the enhanced texture image to obtain texture boundary line segments, including: A line segment detection algorithm is used to extract linear boundary features from the enhanced texture image to obtain texture boundary line segments; The detection scale of the line segment detection algorithm is set to 0.

8.

4. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 1, characterized in that, Mapping the texture boundary line segments to the surface of the 3D mesh model and using them as texture boundary constraints includes: For each triangular facet in the 3D mesh model, establish an affine transformation relationship from texture space to 3D space; Based on the affine transformation relationship, each sampling point in the texture boundary line segment is back-projected onto the surface of the three-dimensional mesh model to obtain a three-dimensional constraint edge that is consistent with the spatial position of the real ground object boundary. The three-dimensional constraint edge is used as the texture boundary constraint.

5. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 1, characterized in that, Based on the original geometry of the 3D mesh model, a candidate point set containing the texture boundary constraints is constructed, including: Based on the original geometric structure of the three-dimensional mesh model, the texture boundary line segments mapped to the three-dimensional space are processed, and uniform sampling is performed on each triangular facet boundary to form a boundary sampling point set; The endpoints and boundary sampling point sets of the texture boundary line segments are collected to form a texture constraint point set; The boundary sampling point set and the texture constraint point set are merged, and repeated or nearly repeated points with a distance less than a preset threshold are filtered to obtain the merged boundary constraint point set. Based on the set of boundary constraint points, a candidate set of points containing the texture boundary constraints is constructed.

6. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 5, characterized in that, Based on the boundary constraint point set, a candidate point set containing the texture boundary constraints is constructed, including: Within each triangular facet of the 3D mesh model, an internal control point set is generated in an approximately equilateral triangle distribution according to the target spacing. The set of internal control points is merged with the set of boundary constraint points, and duplicate points are filtered again to obtain the set of candidate points.

7. The three-dimensional mesh geometric refinement method based on texture boundary constraints according to claim 1, characterized in that, A refined 3D mesh model is generated using a constrained meshing method to achieve consistent alignment between geometric units and texture semantic boundaries, including: Based on the candidate point set and texture boundary line segments, a constrained Delaunay triangulation is performed on each triangular facet of the 3D mesh model to generate a refined 3D mesh model. During the mesh construction process, texture-aligned constraint edges are forcibly retained as mesh edges, while new vertices are automatically introduced at the intersections of constraint lines, satisfying the Delaunay geometric optimality condition.

8. A three-dimensional mesh geometry refinement system based on texture boundary constraints, characterized in that, The system is used to implement the steps of the three-dimensional mesh geometry refinement method based on texture boundary constraints as described in any one of claims 1-7, and the system comprises: The texture enhancement module is used to acquire a three-dimensional mesh model and enhance the texture image of the three-dimensional mesh model. The texture boundary extraction module is used to extract linear boundary features from the enhanced texture image to obtain texture boundary line segments. The constraint point set construction module is used to map the texture boundary line segments to the surface of the three-dimensional mesh model and use them as texture boundary constraints, and to construct a candidate point set containing the texture boundary constraints in combination with the original geometry of the three-dimensional mesh model. The constrained meshing module is used to generate a refined 3D mesh model through a constrained meshing method, achieving consistent alignment between geometric units and texture semantic boundaries.

9. A terminal, characterized in that, The terminal includes a memory, a processor, and a texture boundary constraint-based three-dimensional mesh geometry refinement program stored in the memory and executable on the processor. When the processor executes the texture boundary constraint-based three-dimensional mesh geometry refinement program, it implements the steps of the texture boundary constraint-based three-dimensional mesh geometry refinement method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a three-dimensional mesh geometry refinement program based on texture boundary constraints, the three-dimensional mesh geometry refinement program based on texture boundary constraints implementing the steps of the three-dimensional mesh geometry refinement method based on texture boundary constraints as described in any one of claims 1-7 on the computer-readable storage medium.

Images