Large-scale Triangular Mesh Continuous Detail Hierarchical Representation Method and Dynamic Reconstruction Method

Abstract

The present application relates to the technical field of data representation processing, and discloses a method for continuous detail hierarchical representation and dynamic reconstruction of large-scale triangular meshes. Among them, the representation method includes: performing top-level partitioning on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the nodes of the bottom-level partitioning structure, and the root nodes and intermediate nodes in the BVH structure are determined as the nodes of the top-level partitioning structure; performing bottom-level space partitioning on the triangular meshes included in the bottom-level partitioning structure nodes to generate a corresponding initial patch representation set with patches as the granularity; based on the initial patch representation set, performing mesh simplification operations from the bottom-level partitioning structure nodes to the top-level partitioning structure nodes to obtain a continuous multi-resolution mesh representation. The technical solution provided by the present application can effectively meet the storage, calculation, and real-time display requirements of large-scale mesh models.

Description

Technical Field

[0001] This application relates to the technical field of data representation processing, and particularly to a method for continuously detailed hierarchical representation and dynamic reconstruction of large-scale triangular meshes. Background Art

[0002] With the continuous increase in the scale of triangular mesh models, traditional computing and storage technologies have been difficult to meet the processing requirements of large-scale mesh models. To solve this problem, researchers have adopted technologies such as triangular mesh simplification, visibility culling, and level of detail (LOD), but these methods still face performance and storage bottlenecks when dealing with dynamic viewpoints, non-uniformly distributed scenes, and large-scale data.

[0003] Therefore, the current urgent problem to be solved is how to effectively meet the storage, computing, and real-time display requirements of large-scale mesh models. Summary of the Invention

[0004] This application provides a method for continuously detailed hierarchical representation and dynamic reconstruction of large-scale triangular meshes, achieving the technical effect of effectively meeting the storage, computing, and real-time display requirements of large-scale mesh models.

[0005] To achieve the above objective, the main technical solutions adopted in this application include:

[0006] In the first aspect, an embodiment of this application provides a method for continuously detailed hierarchical representation of large-scale triangular meshes. The representation method includes:

[0007] Performing top-level partitioning on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the nodes of the bottom-level partitioning structure, and the root node and intermediate nodes in the BVH structure are determined as the nodes of the top-level partitioning structure;

[0008] Performing bottom-level space partitioning on the triangular meshes included in the bottom-level partitioning structure nodes to generate a corresponding initial patch representation set with patches as the granularity;

[0009] Based on the initial patch representation set, performing mesh simplification operations from the bottom-level partitioning structure nodes to the top-level partitioning structure nodes to obtain a continuous multi-resolution mesh representation; wherein, the continuous multi-resolution mesh representation includes a simplified target patch representation set.

[0010] A method for continuous detailed hierarchical representation of large-scale triangular meshes provided by this embodiment improves the computational efficiency by performing top-level partitioning based on a large-scale complex scene graph and using a bounding volume hierarchy for efficient mesh simplification. By performing fine-grained patch space partitioning at the bottom layer, it can effectively retain scene details and accurately represent important regions, thus avoiding unnecessary global simplification. The bottom-up layer-by-layer simplification method, combined with multi-resolution mesh representation, can be flexibly adjusted at different detail levels to balance the rendering effect and computational complexity, especially suitable for real-time rendering or simulation of large-scale scenes. This solution not only reduces memory usage but also improves computational efficiency. The finally generated simplified patch representation set can effectively compress redundant information in the mesh while maintaining most shape features.

[0011] In one embodiment, during the process of performing bottom-layer space partitioning, the method further includes:

[0012] Identifying boundary vertices at the boundaries of each hierarchical partition;

[0013] Determining the dependency relationship between the boundary vertices and the initial patch representations in the initial patch representation set;

[0014] When the boundary vertices change, updating the geometric information of the corresponding initial patch representations in the initial patch representation set based on the dependency relationship.

[0015] By identifying the boundary vertices at the boundaries of each hierarchical partition in this embodiment, the key connection points in the mesh can be accurately determined, especially the vertices that cross different levels or region boundaries. This process provides accurate boundary data for mesh simplification, ensuring that the patch structure in the boundary region is reasonably processed during the simplification process. Next, by establishing the dependency relationship between the boundary vertices and the patch representations in the initial patch representation set, the influence of the boundary points on the patches can be accurately traced, avoiding over-simplification or loss of key information. Finally, when the boundary vertices change, the geometric information of the corresponding patches is updated in a timely manner according to these dependency relationships, ensuring the accuracy of the boundary details after mesh simplification, thus optimizing the mesh simplification quality and ensuring the accuracy and natural transition at the boundaries. This method not only improves the quality of mesh simplification but also optimizes the computational efficiency and rendering effect.

[0016] In one embodiment, based on the initial patch representation set, performing a mesh simplification operation from the bottom-layer partition structure nodes to the top-layer partition structure nodes to obtain a continuous multi-resolution mesh representation, including:

[0017] Performing a mesh simplification operation on the initial patch representation set corresponding to each bottom-layer partition structure node to obtain a corresponding simplified patch representation set;

[0018] Upward operation: Assign the simplified patch representation set to the parent nodes at the corresponding levels to obtain a high-level simplified representation;

[0019] Execute the upward operation layer by layer until reaching the root node in the top-level division structure node, obtaining the continuous multi-resolution grid representation from the bottom layer to the top layer.

[0020] In this embodiment, by simplifying the representation set of the underlying grid patches, the data volume is reduced, and the storage and calculation efficiency are improved. Then, through the upward operation, the simplified patch representation set is assigned to the parent nodes to further optimize the grid representation and ensure that the simplification relationship between different levels is retained. Finally, through layer-by-layer simplification, a continuous multi-resolution grid representation is ultimately obtained, enabling rendering or calculation to select an appropriate resolution according to requirements at different viewing distances, thereby optimizing performance and saving computing resources. Overall, by simplifying grid data, reducing computational complexity, and combining continuous multi-resolution grid representation, the rendering efficiency is improved, memory usage is optimized, and the calculation of large-scale scenes is accelerated.

[0021] In a second aspect, an embodiment of the present application provides a dynamic reconstruction method applied to image space error measurement. The dynamic reconstruction method applied to image space error measurement includes:

[0022] Characterize the obtained large-scale complex scene graph according to the above-mentioned characterization method to obtain a continuous multi-resolution grid representation;

[0023] Based on the continuous multi-resolution grid representation, traverse from the top-level division structure node to the bottom-level division structure node, perform error measurement on the frustum and the node bounding box corresponding to the node, and extract the corresponding continuous multi-resolution grid representation.

[0024] The dynamic reconstruction method applied to image space error measurement provided in this embodiment dynamically selects the most suitable grid resolution by traversing the relationship between the bounding box of the node and the frustum and combining error measurement. In this way, the grid detail level can be accurately selected under different viewing angles and node requirements, thereby ensuring the rendering quality and visual effect while optimizing the use of computing resources.

[0025] In one implementation, the performing error measurement on the frustum and the node bounding box corresponding to the node, and extracting the corresponding continuous multi-resolution grid representation includes:

[0026] Determine the nodes whose node bounding boxes are completely outside the frustum as completely invisible nodes, prune and remove the completely invisible nodes and the corresponding subtrees, and extract the corresponding continuous multi-resolution grid representation;

[0027] Determine the nodes whose node bounding boxes are completely within the frustum as fully visible nodes. For the fully visible nodes, traverse the corresponding underlying partition structure nodes downward, convert the simplification error corresponding to the fully visible nodes at the current level into a display error, and extract the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold.

[0028] Determine the nodes whose node bounding boxes partially fall on the frustum boundary as partially visible nodes. For the partially visible nodes, traverse the corresponding underlying partition structure nodes downward, convert the simplification error corresponding to the partially visible nodes at the current level into a display error, and extract the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold.

[0029] In this embodiment, by judging the node bounding boxes, the completely invisible nodes and their subtrees are removed, avoiding unnecessary calculations, reducing resource consumption and improving rendering efficiency. For the fully visible nodes, it is judged whether a higher-precision continuous multi-resolution mesh representation is needed by comparing the display error with the error threshold, so as to ensure the visual effect. For the partially visible nodes, by traversing and comparing the display error with the error threshold, it is ensured that even the partial nodes on the frustum boundary can be properly processed, avoiding rendering effect distortion and redundant calculations at the same time. Generally speaking, this method can efficiently manage computing resources, optimize memory usage, and at the same time improve the rendering performance and quality of large-scale complex scenes.

[0030] In one implementation, the extracting the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold includes:

[0031] When the display error is less than or equal to a preset display error threshold, extract the continuous multi-resolution mesh representation corresponding to the node.

[0032] In a third aspect, an embodiment of the present application provides a dynamic reconstruction method applied to object space error measurement. The dynamic reconstruction method applied to object space error measurement includes:

[0033] Characterize the obtained large-scale complex scene graph according to the above-mentioned characterization method to obtain a continuous multi-resolution mesh representation;

[0034] Based on the continuous multi-resolution mesh representation, traverse from the top-level partition structure nodes to the bottom-level partition structure nodes, and determine a dynamic adaptation threshold according to the traversal depth and a predefined initial distance threshold;

[0035] For the node bounding boxes corresponding to any two nodes, use the bounding box inflation processing technique to determine the query distance;

[0036] When the query distance is greater than the dynamic adaptation threshold, it is determined that there is no collision, and continuous multi-resolution grid representations in the child nodes corresponding to the any two nodes are extracted;

[0037] When the query distance is less than or equal to the dynamic adaptation threshold, it is determined that there is a collision, the simplification error corresponding to the any two nodes is evaluated, and the continuous multi-resolution grid representations are extracted according to the comparison result between the evaluation result and a preset simplification error threshold.

[0038] A dynamic reconstruction method applied to object space error metric provided in this embodiment realizes efficient collision detection by combining continuous multi-resolution grid representations and dynamic adaptation threshold setting. First, by performing multi-resolution grid representation on a complex scene graph, structured grid data is provided, which is convenient for selecting appropriate calculation precision at different levels of detail, thereby improving the collision detection efficiency. Second, by dynamically adjusting the threshold, according to the traversal depth and the initial distance threshold, it flexibly adapts to the detection requirements at different levels, avoiding overly fine calculations for nodes that are far away or irrelevant, and saving computing resources. At the same time, the bounding box inflation processing technology effectively filters out the nodes that need to be further processed, reducing the amount of calculation. Finally, through dynamic threshold judgment and extraction of multi-resolution grids, the balance between the accuracy and efficiency of collision detection in different distance cases is ensured.

[0039] In one embodiment, the extracting the continuous multi-resolution grid representations according to the comparison result between the evaluation result and a preset simplification error threshold includes:

[0040] When the sum of the simplification errors corresponding to the any two nodes is less than or equal to the simplification error threshold, the continuous multi-resolution grid representations in the child nodes corresponding to the any two nodes are extracted;

[0041] When the sum of the simplification errors corresponding to the any two nodes is greater than the simplification error threshold, the node with a larger simplification error is traversed downward until the sum of the simplification errors corresponding to the any two nodes is less than or equal to the simplification error threshold, and the continuous multi-resolution grid representations in the child nodes corresponding to the any two nodes are extracted.

[0042] In this embodiment, by setting a simplified error threshold and flexibly adjusting the calculation method according to the simplified error between nodes. When the sum of the simplified errors corresponding to two nodes is less than or equal to the threshold, the continuous multi-resolution grid representations in the child nodes corresponding to these nodes can be directly extracted, thus quickly completing the calculation, saving computing resources, and improving efficiency. When the sum of the simplified errors is greater than the threshold, the node with a larger simplified error will be traversed downward until a node that meets the error threshold is found, which can ensure the accuracy of the calculation. Through this strategy, the collision detection task can be efficiently and accurately processed in different scenarios, avoiding redundant calculations and ensuring the accuracy of the results, thereby achieving the best balance between calculation efficiency and accuracy.

[0043] In a fourth aspect, an embodiment of the present application provides a dynamic reconstruction method applied to unified error measurement. The dynamic reconstruction method applied to unified error measurement includes:

[0044] According to the dynamic reconstruction method applied to object space error measurement described above, the node whose node bounding box corresponding to a collision-free node completely falls outside the viewing frustum is determined as a completely invisible node, and pruning and removal are performed on the completely invisible node and the corresponding subtree;

[0045] According to the dynamic reconstruction method applied to object space error measurement described above, for the node with a collision, the dynamic reconstruction method applied to image space error measurement described above is used to extract the continuous multi-resolution grid representation corresponding to the node; or

[0046] According to the continuous multi-resolution grid representation extracted by the dynamic reconstruction method applied to object space error measurement described above, the corresponding continuous multi-resolution grid representation is further extracted by using the dynamic reconstruction method applied to image space error measurement described above.

[0047] The dynamic reconstruction method applied to unified error measurement provided in this embodiment can efficiently perform collision detection and visual display processing by combining the object space error measurement and the image space error measurement methods. In the object space error measurement stage, the collision risk and the position inside and outside the viewing frustum of the nodes are quickly screened to avoid unnecessary calculations; in the image space error measurement stage, for the nodes that may collide and the nodes inside the viewing frustum, a suitable grid representation is further selected to ensure the accuracy and quality of the display effect. Through the effective combination of these two error measurement methods, the processing efficiency can be greatly improved while ensuring the calculation accuracy, and the overall performance can be optimized.

[0048] In a fifth aspect, an embodiment of the present application provides a large-scale triangular mesh continuous detail hierarchical representation device, and the device includes:

[0049] The top-level partitioning unit is used to perform top-level partitioning based on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the bottom-level partitioning structure nodes, and the root node and intermediate nodes in the BVH structure are determined as the top-level partitioning structure nodes;

[0050] The bottom-level partitioning unit is used to perform bottom-level space partitioning based on the triangular meshes included in the bottom-level partitioning structure nodes, and generate a corresponding initial patch representation set with patches as the granularity;

[0051] The mesh representation generation unit is used to perform mesh simplification operations from the bottom-level partitioning structure nodes to the top-level partitioning structure nodes based on the initial patch representation set to obtain a continuous multi-resolution mesh representation; wherein, the continuous multi-resolution mesh representation includes a simplified target patch representation set.

[0052] In a sixth aspect, an embodiment of the present application provides a computer device, including:

[0053] A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the above-mentioned large-scale triangular mesh continuous detail hierarchical representation method, the above-mentioned dynamic reconstruction method applied to image space error metric, the above-mentioned dynamic reconstruction method applied to object space error metric, or the above-mentioned dynamic reconstruction method applied to unified error metric.

[0054] In a seventh aspect, an embodiment of the present application provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to cause a computer to execute the above-mentioned large-scale triangular mesh continuous detail hierarchical representation method, the above-mentioned dynamic reconstruction method applied to image space error metric, the dynamic reconstruction method applied to the object space error metric, or the above-mentioned dynamic reconstruction method applied to unified error metric. Description of the Drawings

[0055] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0056] Figure 1 It is a flowchart of a large-scale triangular mesh continuous detail hierarchical representation method provided by an embodiment of the present application;

[0057] Figure 2Flowchart in the process of performing underlying space division provided by an embodiment of this application;

[0058] Figure 3 Flowchart of step S5 provided by an embodiment of this application;

[0059] Figure 4 Flowchart of a dynamic reconstruction method applied to image space error metric provided by an embodiment of this application;

[0060] Figure 5 Flowchart of step S103 provided by an embodiment of this application;

[0061] Figure 6 Flowchart of a dynamic reconstruction method applied to object space error metric provided by an embodiment of this application;

[0062] Figure 7 Block diagram of a large-scale triangular mesh continuous detail hierarchical representation device provided by an embodiment of this application;

[0063] Figure 8 Structural schematic diagram of a computer device provided by an embodiment of this application. Detailed implementation manners

[0064] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Apparently, the described embodiments are some but not all of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of this application without creative efforts shall fall within the scope of protection of this application.

[0065] Triangle Mesh is a widely used representation form in 3D geometric models. With the gradual development of applications related to triangular mesh models, their data scale has been continuously growing, and the computational complexity has also been rising rapidly. However, the processing capabilities of existing hardware are already difficult to meet the processing requirements of large-scale triangular mesh models in complex real-world application scenarios. These models cannot be directly transmitted to display hardware, and in some cases, it is even difficult to load them into the main memory at one time. This bottleneck has become a key problem restricting the application of large-scale triangular mesh models. Therefore, it is necessary to consider multiple factors such as computing power, storage, transmission, and business characteristics, and develop dedicated optimization algorithms to meet the actual needs of large-scale mesh applications.

[0066] To reduce the scale of triangular mesh datasets in digital scenarios, common techniques include triangular mesh simplification and visibility culling. The basic idea of visibility culling is to quickly eliminate mesh elements that have no impact on the final display by estimating which geometric primitives contribute to the image of the current viewport. Common implementation techniques include back-face culling, frustum culling, and occlusion culling. However, relying solely on visibility culling techniques may lead to performance fluctuations because their effectiveness is affected by scene characteristics and the spatial position of the viewpoint. In contrast, mesh simplification techniques reduce the scale of the scene by decreasing the complexity of the mesh, mainly divided into global algorithms and local algorithms. Global algorithms usually process the entire mesh. Although they are simple to implement and have a relatively fast processing speed, the quality of the simplified mesh generated is poor and topological consistency cannot be ensured. Local algorithms operate within local regions of the mesh, gradually reducing complexity, and have higher execution efficiency and robustness. The localized simplification algorithm based on quadratic error metric (QEM) is a representative of this type of technology.

[0067] However, models with a single resolution have limitations when dealing with dynamic viewpoints or object pose changes. To better meet the application requirements, the Levels of Detail (LOD) technology emerged. The LOD technology can optimize the display performance by providing multi-resolution model representations and adjusting the level of detail of the model according to the change of the viewpoint at runtime. LOD technology is divided into discrete LOD, progressive LOD, and continuous LOD. Among them, continuous LOD has advantages that cannot be compared with the other two types of LOD schemes because it can record the relationship between each local operation operator and the modified set it depends on and support non-uniform spatial distribution.

[0068] As the scale of triangular mesh models continues to grow, traditional in-core algorithms that rely on main memory are difficult to meet the processing requirements of large mesh models, and out-of-core mesh simplification technology has emerged. Out-of-core technology uses methods such as underlying caching and on-demand loading to break through the limitations of storage and computing resources. Currently, a solution that combines out-of-core caching, hierarchical volume-height field hybrid approximation, and real-time ray tracing and other technologies has been proposed to address the problem of large mesh models in single-machine interaction and visualization. However, such methods are still limited by storage access latency, and with the increase in mesh scale, the problem of data structure expansion becomes increasingly prominent. In addition to these mainstream simplification methods, new technologies such as image-based "proxy" technology, discrete image sampling interpolation technology, and point-based rendering technology (Splatting) are also being gradually explored, providing new ideas for reducing the complexity of large-scale mesh models.

[0069] Although some achievements have been made in reducing the complexity of large-scale grid models, current research still mainly focuses on a single display problem. In the application background of high complexity, multi-factor coupling, and simultaneous display and calculation requirements, exploring high-performance technical solutions based on the characteristics of modern hardware platforms to meet the representation, display, and calculation requirements of large-scale grid models still has important scientific research value and practical urgency.

[0070] To solve the above technical problems, according to the embodiments of the present application, an embodiment of a method for continuous detail hierarchical representation of large-scale triangular grids is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0071] In this embodiment, a method for continuous detail hierarchical representation of large-scale triangular grids is provided. Figure 1 The flowchart of a method for continuous detail hierarchical representation of large-scale triangular grids provided by the embodiments of the present application is as Figure 1 shown, and the process includes the following steps:

[0072] Step S1, perform top-level partitioning on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the nodes of the bottom-level partitioning structure, and the root nodes and intermediate nodes in the BVH structure are determined as the nodes of the top-level partitioning structure.

[0073] Specifically, the large-scale complex scene graph is essentially a hierarchical data structure used to describe the geometric and semantic relationships of objects in the scene. The scene contains various types of objects, and the geometric forms (such as shape, size) and semantic information (such as object category, function) of each object are interrelated. According to the geometric features (such as the shape, size of the object) and scene semantics (such as object category, function) information, a compact hierarchical bounding volume structure is adaptively applied to efficiently decompose the single large-scale complex scene graph into a multi-level and multi-node representation form. Each node is represented by a bounding volume (such as AABB, OBB, K-Dop, etc.) to form a hierarchical BVH structure. In the BVH structure, each node contains a node bounding box, and the nodes are connected through a tree structure. The hierarchical structure of the nodes ranges from the root node (representing the scope of the entire scene) to the leaf nodes (representing the geometric data of specific objects).

[0074] Top-level partitioning structure nodes refer to the root nodes and intermediate nodes in the BVH structure, which do not directly store specific geometric data. Instead, they store node bounding boxes (such as AABB, OBB, etc.) to quickly determine whether the objects contained in the node are likely to interact with other objects. In addition to the node bounding box, the top-level partitioning structure nodes also store the node geometric error and the spatial transformation matrix. The node geometric error represents the error between the geometric body contained in the current node and the original model, which can help optimize subsequent calculations, such as simplifying the mesh. The spatial transformation matrix is used to describe the spatial transformation of the node (such as translation, rotation, scaling), which is particularly suitable for the rigid body motion in dynamic scenes. By storing these transformation matrices, the top-level nodes can adapt to dynamically changing scenes.

[0075] The bottom-level partitioning structure nodes correspond to the leaf nodes in the BVH structure, and these nodes directly store specific geometric data (such as triangular meshes, point cloud data, etc.). The storage and calculation overheads of these nodes are relatively large, but they are the parts that need to be accessed during actual collision detection, ray tracing, etc.

[0076] Optimizing the BVH structure through top-level partitioning can effectively handle large-scale complex scene graphs, especially in dynamic scenes. By hierarchically organizing geometric data, the calculation process can be accelerated and the storage overhead can be reduced.

[0077] Step S3: Based on the triangular meshes contained in the bottom-level partitioning structure nodes, perform bottom-level space partitioning to generate a corresponding initial patch representation set with patches as the granularity.

[0078] Specifically, a patch is a set of triangles with spatial locality. Each patch contains a group of adjacent triangles, which are locally related in space and can thus be processed as a whole. After top-level partitioning, the large-scale scene is decomposed into multiple BVH sub-nodes. Next, starting from the BVH leaf nodes, perform space partitioning on the triangular meshes in each leaf node to generate a representation set of the large-scale mesh model with patches as the granularity. Multiple methods can be used for bottom-level space partitioning, such as KD-Tree, Octree, or Uniform Grid, and KD-Tree is preferably used. The selection of these methods depends on the specific application scenario and data characteristics. The main reason for choosing patches as the processing unit is their good spatial locality, which helps optimize the memory access consistency of the GPU and reduce the memory access and transmission overheads of data.

[0079] Step S5: Based on the initial patch representation set, perform mesh simplification operations from the bottom-level partitioning structure nodes to the top-level partitioning structure nodes to obtain a continuous multi-resolution mesh representation; among them, the continuous multi-resolution mesh representation includes the simplified target patch representation set.

[0080] Specifically, starting from the leaf nodes of the underlying spatial partitioning structure, a simplification operation is performed on the initial patch representations in each leaf node. The algorithm for the simplification operation can adopt QEM (Quadric Error Metrics) or a feature-preserving technique based on the Laplacian operator to reduce the number of triangular patches while trying to preserve the features of the geometric shape. The simplified rough triangular patch set is assigned to the upper-level parent nodes of each child node, and a compact bounding volume BV (Bounding Volume) and a simplification error ε (the Hausdorff distance between the triangular meshes before and after simplification) are calculated for each node. The simplification-assignment process is iteratively executed level by level, and finally a continuous multi-resolution mesh representation with an increasing error layer by layer from the leaf nodes to the root node of the underlying partitioning structure is formed. Through this simplification-assignment strategy, the finally formed mesh has the characteristics of multi-resolution. That is, different levels represent different mesh details, and different resolutions can be selected according to needs to adapt to different applications (such as real-time rendering, large-scale simulation, etc.).

[0081] A method for continuous detail hierarchical representation of a large-scale triangular mesh provided in this embodiment improves the computational efficiency by performing a top-level partitioning based on a large-scale complex scene graph and using a hierarchical bounding volume structure for efficient mesh simplification. By performing a fine-grained patch space partitioning at the bottom layer, it can effectively retain scene details and accurately represent important regions, thereby avoiding unnecessary global simplification. The bottom-up layer-by-layer simplification method, combined with the multi-resolution mesh representation, can be flexibly adjusted at different detail levels to balance the rendering effect and the computational complexity, and is particularly suitable for real-time rendering or simulation of large-scale scenes. This solution not only reduces the memory usage but also improves the computational efficiency. The finally generated set of simplified patch representations can effectively compress the redundant information in the mesh while maintaining most of the shape features.

[0082] Figure 2 The following is a flowchart of the process for performing the underlying spatial partitioning provided in the embodiment of the present application, and this process may include the following steps:

[0083] Step S31: Identify the boundary vertices at the partitioning boundaries of each level.

[0084] Step S33: Determine the dependency relationship between the boundary vertices and the initial patch representations in the initial patch representation set.

[0085] Step S35: When the boundary vertices change, update the geometric information of the initial patch representations in the corresponding initial patch representation set based on the dependency relationship.

[0086] In the process of grid representation and simplification in large-scale complex scenes, ensuring the geometric consistency of the multi-resolution model is a key issue. Especially after the connection relationship changes due to grid simplification, it is necessary to specifically consider the grids across node boundaries to avoid cracks between patches. Specifically, for the division boundaries at each level, mark the boundary vertices located on the boundary, and record the relationship between each boundary vertex and the initial patch representation that depends on it, so that when the boundary vertex changes, the affected patches can be quickly located. This can be achieved by recording the list of boundary vertices on which each initial patch representation depends. At the same time, record the list of associated patches sharing the boundary vertex. When the position or attribute of the boundary vertex changes, ensure that all initial patch representations depending on this vertex can synchronously update their geometric information to maintain geometric consistency. Specifically, in grid editing operations (such as grid simplification, vertex movement, etc.), in order to avoid cracks, after the boundary vertex changes, all relevant initial patch representations must be updated quickly and accurately. This requires that when updating the initial patch representation, the updates across different levels of nodes can be synchronized. If the attribute of a boundary vertex changes, this change should not only affect the initial patch representation at the current level, but also affect the initial patch representations of other levels across this boundary, so as to ensure that each initial patch representation is consistent with other relevant initial patch representations. This dynamic update mechanism ensures that even after the connection relationship changes due to grid simplification, the initial patch representations between nodes at the same level and between nodes at different levels still maintain geometric consistency.

[0087] This embodiment can accurately determine the key connection points in the grid by identifying the boundary vertices of the division boundaries at each level, especially the vertices across different levels or regional boundaries. This process provides accurate boundary data for grid simplification, ensuring that the patch structure in the boundary area is reasonably processed during the simplification process. Next, by establishing the dependency relationship between the boundary vertices and the patch representations in the set of initial patch representations, the influence of the boundary points on the patches can be accurately traced, avoiding over-simplification or loss of key information. Finally, when the boundary vertex changes, the geometric information of the corresponding patches is updated in a timely manner according to these dependency relationships, ensuring the accuracy of the boundary details after grid simplification, thus optimizing the quality of grid simplification and ensuring the accuracy and natural transition at the boundary. This method not only improves the quality of grid simplification, but also optimizes the computational efficiency and rendering effect.

[0088] Figure 3 The flowchart of step S5 provided by the embodiment of the present application is shown, and this process may include the following steps:

[0089] Step S51, perform a grid simplification operation on the set of initial patch representations corresponding to each underlying division structure node to obtain a corresponding set of simplified patch representations.

[0090] Specifically, mesh simplification is performed on the initial face representation set in the underlying partition structure node. This process involves using mature mesh simplification algorithms, such as Quadric Error Metrics (QEM) or feature-preserving techniques based on the Laplacian operator, to simplify the face set of the triangular mesh. The goal of simplification is to reduce the number of faces in the mesh and maintain the shape and structure of the mesh as much as possible, thereby improving the efficiency of rendering and storage.

[0091] Step S53, upward operation: assigning the simplified face representation set to the parent node of the corresponding level to obtain a simplified representation at a higher level.

[0092] Specifically, the simplified results of the bottom-level nodes will be passed to the previous layer through merging or simplification operations until the root node of the top layer. Each parent node not only receives the simplified results of the child nodes, but also needs to perform further simplification and merging to eventually form a more concise mesh representation. The parent node merges the simplified face sets of all child nodes to form a simplified mesh at this level. This merge is not just data splicing, but also a geometric reconstruction, which usually involves further simplification of the vertices and faces of the parent node. The simplified results of the parent node will affect the simplification of higher levels until the top layer of the entire mesh.

[0093] Step S55, performing upward operations layer by layer until reaching the root node in the top-level partition structure node, and obtaining a continuous multi-resolution grid representation from the bottom layer to the top layer.

[0094] Specifically, at each level, the compact bounding volume BV (Bounding Volume) and the simplification error ε (Hausdorff distance between the triangle mesh before and after simplification) are calculated. This information is used to support fast node selection and error metric calculation at runtime. Finally, a continuous multi-resolution mesh representation with increasing error from leaf nodes to root nodes is formed.

[0095] During the node simplification process, the nodes represented by the continuous multi-resolution mesh are saved to the external memory in a timely manner to optimize the storage and access efficiency. Before each write operation to the external memory, the face representation set in the node is packaged and a triangle strip is constructed to optimize the memory access implementation of spatial locality. The irregular face storage set is encoded into storage blocks of equal size to facilitate the design of efficient memory access strategies for large-capacity data, such as memory pool technology. The memory pool technology reduces the overhead of dynamic memory allocation by pre-allocating memory blocks, further improving the overall performance of the algorithm when processing large-scale triangular meshes.

[0096] In this embodiment, by simplifying the representation set of the underlying mesh patches, the data volume is reduced, and the storage and calculation efficiency are improved. Then, through the upward operation, the simplified patch representation set is assigned to the parent node to further optimize the mesh representation and ensure that the simplification relationship between different levels is retained. Finally, through layer-by-layer simplification, a continuous multi-resolution mesh representation is finally obtained, enabling rendering or calculation to select an appropriate resolution according to requirements at different viewing distances, thereby optimizing performance and saving computing resources. Overall, by simplifying the mesh data, reducing the computational complexity, and combining the continuous multi-resolution mesh representation, the rendering efficiency is improved, the memory usage is optimized, and the calculation of large-scale scenes is accelerated.

[0097] An embodiment of the present application provides a dynamic reconstruction method applied to image space error measurement, as Figure 4 shown, the flowchart of a dynamic reconstruction method applied to image space error measurement provided by an embodiment of the present application includes the following steps:

[0098] Step S101, perform characterization on the obtained large-scale complex scene graph according to the above characterization method to obtain a continuous multi-resolution mesh representation.

[0099] Step S103, based on the continuous multi-resolution mesh representation, traverse from the top-level division structure node to the bottom-level division structure node, perform error measurement on the node bounding box corresponding to the frustum, and extract the corresponding continuous multi-resolution mesh representation.

[0100] Specifically, a continuous multi-resolution mesh representation is obtained according to Steps S1 - S5. The continuous multi-resolution mesh representation means that each node in the scene contains a corresponding mesh, and the resolution of the mesh is dynamically adjusted according to different viewing distances, ensuring both graphic details and improved calculation efficiency during the rendering process.

[0101] Starting from the root node at the top layer, traverse each node downward according to the tree structure. This is a hierarchical traversal process. The root node represents the top-level structure of the scene, and each child node represents a more detailed area or sub-structure. During the traversal, it is necessary to extract the key information contained in each node, mainly the node bounding box, simplification error, and spatial transformation matrix. The node is successively transformed from the model coordinate system to the world coordinate system and the view coordinate system, and the error metric between the frustum and the node bounding box is performed. The frustum is the visible area under the camera's perspective, which represents the area of the scene that the camera can see. The error metric is performed by comparing the difference between the frustum and the node bounding box. Based on the result of the error metric, a continuous multi-resolution mesh representation that meets the display error requirements is selected for subsequent rendering processing. In this process, areas with a closer viewing distance will use a high-resolution continuous multi-resolution mesh representation, while areas far from the view point will use a low-resolution continuous multi-resolution mesh representation. This dynamic selection based on the viewing distance helps to reduce the computational load and ensure the details and realism of the rendering result.

[0102] A dynamic reconstruction method applied to image space error metric provided in this embodiment dynamically selects the most suitable mesh resolution by traversing the relationship between the bounding box of the node and the frustum and combining the error metric. In this way, the mesh detail level can be accurately selected under different perspectives and node requirements, thereby ensuring the rendering quality and visual effect while optimizing the use of computing resources.

[0103] Figure 5 It is a flowchart of step S103 provided by an embodiment of the present application, and this process includes the following steps:

[0104] In step S1031, the nodes whose node bounding boxes completely fall outside the frustum are determined as completely invisible nodes, and the completely invisible nodes and the corresponding subtrees are pruned and removed, and the corresponding continuous multi-resolution mesh representation is extracted.

[0105] Specifically, the algorithm starts from the top-level root node and traverses each level downward in turn. For each node, the node is first successively transformed from the model coordinate system to the world coordinate system and the view coordinate system through a series of transformations. And it is judged whether the node bounding box completely falls outside the frustum. When the bounding box of a certain node is completely outside the frustum, this node and all its corresponding subtrees are considered to be completely invisible under the current view point. Therefore, all these invisible nodes and their child nodes will be removed from the rendering queue. To avoid further processing, this operation significantly reduces the rendering overhead. After removing the invisible nodes, the continuous multi-resolution mesh representation of the remaining visible nodes is extracted for subsequent rendering or collision detection. For completely visible or partially visible nodes, continue to traverse downward to extract the continuous multi-resolution mesh representation that meets the error requirements.

[0106] Step S1033: Determine the nodes whose node bounding boxes are completely within the frustum as fully visible nodes. For fully visible nodes, traverse the corresponding underlying partition structure nodes downward, convert the simplification error corresponding to the fully visible nodes at the current level into a display error, and extract the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold.

[0107] Specifically, when a node bounding box is completely within the frustum, the node and its subtree are considered fully visible. For a fully visible node, first generate a list of all its child nodes. Then, recursively traverse each child node. By projecting the simplification error of each child node (usually defined in the model space or world space) into the display image space, the display error at the current viewing point is obtained. Compare this display error with the predefined error threshold: If the display error is less than or equal to the error threshold, the continuous multi-resolution mesh representation of the current node meets the display requirements, and the traversal terminates. If the display error is greater than the error threshold, continue traversing the underlying space partition level to find a continuous multi-resolution mesh representation with a higher resolution. The display error is calculated by projecting the simplification error into the image space, usually involving multiple coordinate system conversions, including from the model space, world space to the viewing point coordinate system, and then to the screen space. This ensures that the error metric is consistent with the actual display effect on the screen.

[0108] Step S1035: Determine the nodes whose node bounding boxes partially fall on the frustum boundary as partially visible nodes. For partially visible nodes, traverse the corresponding underlying partition structure nodes downward, convert the simplification error corresponding to the partially visible nodes at the current level into a display error, and extract the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold.

[0109] Specifically, if the node bounding box intersects with the frustum but is not completely within it, the node is a partially visible node. For partially visible nodes, it is also necessary to continue recursively traversing their child nodes and calculate the projection of the corresponding simplification error into the display image space. Similarly, by comparing the display error with the error threshold, it is decided whether to continue the in-depth traversal.

[0110] In one embodiment, when the display error is less than or equal to a preset display error threshold, extract the continuous multi-resolution mesh representation corresponding to the node.

[0111] In this embodiment, by judging the node bounding box, nodes that are completely invisible and their subtrees are removed, avoiding unnecessary calculations, reducing resource consumption, and improving rendering efficiency. For nodes that are completely visible, it is judged whether a higher-precision continuous multi-resolution mesh representation is needed by comparing the display error with the error threshold, so as to ensure the visual effect. For partially visible nodes, by traversing and comparing the display error with the error threshold, it is ensured that even partial nodes on the boundary of the viewing frustum can be properly processed, avoiding rendering effect distortion and redundant calculations at the same time. Generally speaking, this method can efficiently manage computing resources, optimize memory usage, and improve the rendering performance and quality of large-scale complex scenes.

[0112] An embodiment of the present application provides a dynamic reconstruction method applied to object space error measurement, as Figure 6 shown, an embodiment of the present application provides a flowchart of a dynamic reconstruction method applied to object space error measurement, and this process includes the following steps:

[0113] Step S201, represent the obtained large-scale complex scene graph according to the above-mentioned representation method to obtain a continuous multi-resolution mesh representation.

[0114] Step S203, based on the continuous multi-resolution mesh representation, traverse from the top-level division structure node to the bottom-level division structure node, and determine the dynamic adaptation threshold according to the traversal depth and a predefined initial distance threshold.

[0115] Specifically, in collision detection, the dynamic adaptation threshold is used to reduce the number of false positive collision pairs and improve the detection accuracy. Starting from the root node at the top level, traverse down to the bottom-level division structure node. Calculate the dynamic adaptation threshold according to the traversal depth and the initial distance threshold. As the traversal depth increases, a smaller distance threshold is adopted to perform more accurate and strict queries. Among them, the dynamic adaptation threshold δ = δ0 / f(l), where δ0 is the initial distance threshold for traversing the root node defined by the user, and f(l) is a monotonically increasing function of the hierarchical depth l. The design of the dynamic adaptation threshold can effectively reduce the amount of calculation and improve the accuracy of collision detection. Especially between nodes at deeper levels, the detection distance will be smaller, and potential collisions can be captured in detail.

[0116] Step S205, for the node bounding boxes corresponding to any two nodes, use the bounding box dilation processing technology to determine the query distance.

[0117] Specifically, the two node bounding boxes are dilated using the bounding box dilation (Dilate) processing technology to make them occupy a larger range in space. Perform an intersection test between the dilated node bounding boxes to determine the query distance Overlap(B a , B b), where B a , B b are respectively the node bounding boxes of two nodes to be queried after dilation processing. The degree of intersection (Overlap) will be the basis for judging whether there are potential collision pairs.

[0118] Step S207, when the query distance is greater than the dynamic adaptation threshold, it is determined that there is no collision, and the continuous multi-resolution grid representations in the child nodes corresponding to any two nodes are extracted.

[0119] Specifically, applying the BVTT algorithm, a list of potential collision pairs (N a , N b ) in the scene is formed, where N a , N b are respectively nodes from different scene objects. If the query distance is greater than the dynamic adaptation threshold, it is determined that there is no collision. The continuous multi-resolution grid representations in the child nodes corresponding to the collision-free nodes are extracted for subsequent processing.

[0120] Step S209, when the query distance is less than or equal to the dynamic adaptation threshold, it is determined that there is a collision. The simplified error corresponding to any two nodes is evaluated, and the continuous multi-resolution grid representation is extracted according to the comparison result between the evaluation result and the preset simplified error threshold.

[0121] Specifically, if the query distance is less than or equal to the dynamic adaptation threshold, it is determined that there is a collision. The simplified error corresponding to the colliding nodes is evaluated. The evaluation result is compared with the preset simplified error threshold. If the sum of the simplified errors is less than or equal to the simplified error threshold, the continuous multi-resolution grid representation of the current node is extracted. If the sum of the simplified errors is greater than the simplified error threshold, the node with the larger simplified error is traversed downward to perform higher-resolution detection calculations. The setting of the simplified error threshold is a process of balancing performance and accuracy. If the error is small, the amount of calculation can be reduced; if the error is large, more refined calculations are required to ensure accuracy.

[0122] A dynamic reconstruction method applied to object space error measurement provided in this embodiment realizes efficient collision detection by combining continuous multi-resolution grid representation and dynamic adaptive threshold setting. First, by performing multi-resolution grid representation on a complex scene graph, structured grid data is provided, which facilitates selecting appropriate calculation precision at different levels of detail, thereby improving the efficiency of collision detection. Second, by dynamically adjusting the threshold, according to the traversal depth and the initial distance threshold, it flexibly adapts to the detection requirements at different levels, avoiding overly fine calculations for nodes that are far away or irrelevant, and saving computing resources. At the same time, the bounding box inflation processing technology effectively filters out the nodes that need further processing, reducing the amount of calculation. Finally, through dynamic threshold judgment and extraction of multi-resolution grids, the balance between the accuracy and efficiency of collision detection in different distance situations is ensured.

[0123] In one embodiment, when the sum of the simplification errors corresponding to any two nodes is less than or equal to the simplification error threshold, the continuous multi-resolution grid representation in the child nodes corresponding to the two nodes is extracted.

[0124] When the sum of the simplification errors corresponding to any two nodes is greater than the simplification error threshold, the node with the larger simplification error is traversed downward until the sum of the simplification errors corresponding to any two nodes is less than or equal to the simplification error threshold, and the continuous multi-resolution grid representation in the child nodes corresponding to the two nodes is extracted.

[0125] Specifically, if Overlap(B a , B b ) and ε a +ε b ≤h, where ε a , ε b are the simplification errors corresponding to the two nodes respectively, and h is the simplification error threshold predefined by the user, then it is determined that there is a collision in this area of the scene. This indicates that the continuous multi-resolution grid representations corresponding to these two nodes are accurate enough and there is no need to continue in-depth calculation. Extract the continuous multi-resolution grid representation in the child nodes, that is, directly use the lower-level grid representations of these nodes for collision detection. Since the error is small enough, such an extraction process can avoid the complexity of further calculation and ensure that the accuracy of collision detection has reached the application requirements.

[0126] If Overlap(B a , B b ) and ε a +ε b>h, that is, there is spatial overlap between the bounding boxes of two nodes, but the resolution of the continuous multi-resolution mesh representation in the node is low, and the error from the original mesh is too large. Then it is determined that there is a potential collision of the object in the scene in this area. Higher-resolution detection calculations need to be performed. Each time, select the node with a large simplification error ε among them, recursively traverse downward the hierarchical structure of the underlying space division, and perform collision calculations according to the above rules until the sum of the simplification errors meets the condition of being less than or equal to the threshold.

[0127] In this embodiment, by setting a simplification error threshold, the calculation method is flexibly adjusted according to the simplification error between nodes. When the sum of the simplification errors corresponding to two nodes is less than or equal to the threshold, the continuous multi-resolution mesh representations in the child nodes corresponding to these nodes can be directly extracted, thus quickly completing the calculation, saving computing resources, and improving efficiency. When the sum of the simplification errors is greater than the threshold, the nodes with larger simplification errors will be traversed downward until nodes that meet the error threshold are found, which can ensure the accuracy of the calculation. Through this strategy, the collision detection task can be efficiently and accurately processed in different scenarios, avoiding redundant calculations, while ensuring the accuracy of the results, so as to achieve the best balance between computing efficiency and accuracy.

[0128] A dynamic reconstruction method applied to unified error metric provided by an embodiment of the present application, the dynamic reconstruction method applied to unified error metric includes:

[0129] According to the above dynamic reconstruction method applied to object space error metric, determine the nodes whose bounding boxes corresponding to the collision-free nodes are completely outside the frustum as completely invisible nodes, and prune and remove the completely invisible nodes and their corresponding subtrees;

[0130] According to the above dynamic reconstruction method applied to object space error metric, for the nodes with collisions, use the above dynamic reconstruction method applied to image space error metric to extract the continuous multi-resolution mesh representations corresponding to the nodes.

[0131] Specifically, first, judge whether a node may collide through the dynamic reconstruction method applied to object space error metric. For collision-free nodes, further judge whether they are completely outside the frustum. If a node is completely outside the frustum, prune and remove the node and its subtree. If a node is partially or completely inside the frustum, use the dynamic reconstruction method applied to image space error metric for this node, and select the continuous multi-resolution mesh representation that meets the display error.

[0132] For nodes with collisions, use the dynamic reconstruction method applied to image space error metric, and select the continuous multi-resolution mesh representation corresponding to the appropriate nodes according to the error threshold.

[0133] Or, according to the continuous multi-resolution grid representation extracted by the above dynamic reconstruction method applied to the object space error metric, the corresponding continuous multi-resolution grid representation is further extracted by using the above dynamic reconstruction method applied to the image space error metric.

[0134] Specifically, based on the object space error metric, a suitable continuous multi-resolution grid representation is further selected by using the dynamic reconstruction method applied to the image space error metric to ensure that the visual quality requirements are met during display.

[0135] The combination of these two error metric methods forms a hierarchical, error threshold-based decision framework. First, a large-scale error metric is performed in the object space to quickly screen out nodes that may collide or are completely outside the frustum. Then, for the nodes within the frustum, the dynamic reconstruction method of the image space error metric is further applied to select a suitable continuous multi-resolution grid representation to ensure that the display effect meets the requirements. For the nodes with collisions, they are further refined through the dynamic reconstruction method of the image space error metric to ensure the accuracy of collision detection and the visual effect. Thus, a single framework supports a wide range of application scenarios with both display and computing.

[0136] A dynamic reconstruction method applied to unified error metric provided in this embodiment can efficiently perform collision detection and visual display processing by combining the object space error metric and the image space error metric. In the object space error metric stage, the collision risk and the position inside and outside the frustum of nodes are quickly screened to avoid unnecessary calculations; in the image space error metric stage, for the nodes that may collide and the nodes within the frustum, a more suitable grid representation is further selected to ensure the accuracy and quality of the display effect. Through the effective combination of these two error metric methods, while ensuring the calculation accuracy, the processing efficiency can be greatly improved and the overall performance can be optimized.

[0137] Correspondingly, please refer to Figure 7 which is a block diagram of a large-scale triangular mesh continuous detail hierarchical representation device provided in an embodiment of the present application. The device includes:

[0138] A top-level division unit 101, configured to perform top-level division on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the nodes of the bottom-level division structure, and the root nodes and intermediate nodes in the BVH structure are determined as the nodes of the top-level division structure;

[0139] A bottom-level division unit 103, configured to perform bottom-level space division on the triangular meshes included in the bottom-level division structure nodes to generate a corresponding initial patch representation set with patches as the granularity;

[0140] A mesh representation generation unit 105 is configured to perform mesh simplification operations from bottom-level division structure nodes to top-level division structure nodes based on an initial patch representation set, so as to obtain a continuous multi-resolution mesh representation; wherein, the continuous multi-resolution mesh representation includes a set of target patch representations after simplification.

[0141] In some alternative embodiments, during the process of performing bottom-level space division, the method further includes:

[0142] Identifying boundary vertices at the division boundaries of each level;

[0143] Determining the dependency relationships between the boundary vertices and the initial patch representations in the initial patch representation set;

[0144] When the boundary vertices change, updating the geometric information of the initial patch representations in the corresponding initial patch representation set based on the dependency relationships.

[0145] In some alternative embodiments, performing mesh simplification operations from bottom-level division structure nodes to top-level division structure nodes based on an initial patch representation set to obtain a continuous multi-resolution mesh representation includes:

[0146] Performing mesh simplification operations on the initial patch representation set corresponding to each bottom-level division structure node to obtain a corresponding set of simplified patch representations;

[0147] Upward operation: Assigning the set of simplified patch representations to the parent nodes at the corresponding levels to obtain a high-level simplified representation;

[0148] Performing the upward operation layer by layer until reaching the root node in the top-level division structure node, so as to obtain a continuous multi-resolution mesh representation from bottom to top.

[0149] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding embodiments above, and will not be elaborated here.

[0150] A continuous multi-resolution mesh representation device in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0151] Please refer to Figure 8 , Figure 8 which is a schematic structural diagram of a computer device provided by an embodiment of the present application, as Figure 8As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including a high-speed interface and a low-speed interface. Each component communicates with each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some alternative embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a set of blade servers, or a multi-processor system). Figure 8 In [the figure], a processor 10 is taken as an example.

[0152] The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above-mentioned hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above-mentioned programmable logic device can be a complex programmable logic device, a field-programmable gate array, a generic array logic, or any combination thereof.

[0153] Among them, the memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0154] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely set relative to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The memory 20 can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memories.

[0156] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

[0157] The embodiments of the present application also provide a computer-readable storage medium. The methods according to the embodiments of the present application can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the methods described herein can be stored in such software processes on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the methods shown in the above embodiments are implemented.

[0158] The devices and units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0159] For the convenience of description, the above devices are described by dividing them into various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0160] Those skilled in the art should understand that the embodiments of the present application can be provided as methods and devices. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] The present application is described with reference to the flowcharts and / or block diagrams of methods and devices according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a dedicated computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing the processFigure 1 one or more processes and / or blocks Figure 1 a device for the functions specified in one or more blocks

[0162] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device that implements the functions in the process Figure 1 one or more processes and / or blocks Figure 1 specified in one or more blocks

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the process Figure 1 one or more processes and / or blocks Figure 1 specified in one or more blocks

[0164] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the presence of additional identical elements in the process, method, commodity or device including the said element

[0165] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiment

[0166] The above description is only for the embodiments of this application and is not used to limit this application. For those skilled in the art, this application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included within the scope of the claims of this application

[0167] Although the embodiments of this application are described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations fall within the scope defined by the appended claims

Claims

1. A method for continuously representing hierarchical details of a large-scale triangular mesh, characterized in that The described characterization method includes: Performing top-level partitioning on the obtained large-scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the nodes of the bottom-level partitioning structure, and the root node and intermediate nodes in the BVH structure are determined as the nodes of the top-level partitioning structure; Performing bottom-level space partitioning based on the triangular meshes included in the nodes of the bottom-level partitioning structure to generate a corresponding initial patch representation set with patches as the granularity; wherein, during the process of performing bottom-level space partitioning, the method further includes: identifying the boundary vertices at the partitioning boundaries of each level; determining the dependency relationships between the boundary vertices and the initial patch representations in the initial patch representation set; in the case where the boundary vertices change, updating the geometric information of the corresponding initial patch representations in the initial patch representation set based on the dependency relationships; Based on the initial patch representation set, performing mesh simplification operations from the nodes of the bottom-level partitioning structure to the nodes of the top-level partitioning structure to obtain a continuous multi-resolution mesh representation; wherein, the continuous multi-resolution mesh representation includes a set of simplified target patch representations.

2. The characterization method according to claim 1, wherein The performing mesh simplification operations from the nodes of the bottom-level partitioning structure to the nodes of the top-level partitioning structure based on the initial patch representation set to obtain a continuous multi-resolution mesh representation includes: Performing mesh simplification operations on the initial patch representation set corresponding to each node of the bottom-level partitioning structure to obtain a corresponding set of simplified patch representations; Upward operation: Assigning the set of simplified patch representations to the parent nodes at the corresponding levels to obtain a high-level simplified representation; Performing the upward operation layer by layer until reaching the root node in the nodes of the top-level partitioning structure to obtain the continuous multi-resolution mesh representation from the bottom level to the top level.

3. A dynamic reconstruction method applied to image spatial error measurement, characterized in that, The dynamic reconstruction method applied to image space error metric includes: Characterizing the obtained large-scale complex scene graph according to the characterization method described in any one of claims 1-2 to obtain a continuous multi-resolution mesh representation; Based on the continuous multi-resolution mesh representation, traversing from the nodes of the top-level partitioning structure to the nodes of the bottom-level partitioning structure, performing error metric on the frustum and the node bounding box corresponding to the node, and extracting the corresponding continuous multi-resolution mesh representation.

4. The dynamic reconstruction method applied to image spatial error measurement according to claim 3, wherein, The performing error metric on the frustum and the node bounding box corresponding to the node and extracting the corresponding continuous multi-resolution mesh representation includes: Determining the nodes whose node bounding boxes are completely outside the frustum as completely invisible nodes, pruning and removing the completely invisible nodes and the corresponding subtrees, and extracting the corresponding continuous multi-resolution mesh representation; Determining the nodes whose node bounding boxes are completely inside the frustum as completely visible nodes, for the completely visible nodes, traversing downward to the corresponding nodes of the bottom-level partitioning structure, and converting the simplified error corresponding to the completely visible nodes at the current level into a display error, and extracting the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold. Nodes whose node bounding boxes partially fall within the frustum boundary are determined as partially visible nodes. For these partially visible nodes, traverse the corresponding underlying hierarchical structure nodes downward, and convert the simplification error corresponding to the partially visible nodes at the current level into a display error. According to the comparison result between the display error and a predefined error threshold, extract the corresponding continuous multi-resolution mesh representation.

5. The dynamic reconstruction method applied to image spatial error metric according to claim 4, characterized in that The extracting the corresponding continuous multi-resolution mesh representation according to the comparison result between the display error and a predefined error threshold includes: When the display error is less than or equal to a preset display error threshold, extract the continuous multi-resolution mesh representation corresponding to the node.

6. A dynamic reconstruction method applied to object space error measurement, characterized in that, The dynamic reconstruction method applied to object space error metric includes: Characterize the obtained large-scale complex scene graph according to the characterization method of any one of claims 1-2 to obtain a continuous multi-resolution mesh representation; Based on the continuous multi-resolution mesh representation, traverse from the top-level hierarchical structure nodes to the bottom-level hierarchical structure nodes, and determine a dynamic adaptation threshold according to the traversal depth and a predefined initial distance threshold; For the node bounding boxes corresponding to any two nodes, use the bounding box inflation processing technique to determine the query distance; When the query distance is greater than the dynamic adaptation threshold, determine it as collision-free, and extract the continuous multi-resolution mesh representation in the child nodes corresponding to the any two nodes; When the query distance is less than or equal to the dynamic adaptation threshold, determine it as having a collision, evaluate the simplification error corresponding to the any two nodes, and extract the continuous multi-resolution mesh representation according to the comparison result between the evaluation result and a preset simplification error threshold.

7. The dynamic reconstruction method applied to object space error metric according to claim 6, characterized in that The extracting the continuous multi-resolution mesh representation according to the comparison result between the evaluation result and a preset simplification error threshold includes: When the sum of the simplification errors corresponding to the any two nodes is less than or equal to the simplification error threshold, extract the continuous multi-resolution mesh representation in the child nodes corresponding to the any two nodes; When the sum of the simplification errors corresponding to the any two nodes is greater than the simplification error threshold, traverse the node with the larger simplification error downward until the sum of the simplification errors corresponding to the any two nodes is less than or equal to the simplification error threshold, and extract the continuous multi-resolution mesh representation in the child nodes corresponding to the any two nodes.

8. A dynamic reconstruction method applied to unified error metrics, characterized in that, The dynamic reconstruction method applied to unified error metric includes: According to the dynamic reconstruction method applied to object space error metric in claim 6, determine nodes whose node bounding boxes corresponding to collision-free nodes completely fall outside the frustum as completely invisible nodes, and prune and remove the completely invisible nodes and their corresponding subtrees; According to the dynamic reconstruction method applied to object space error metric in claim 6, for nodes having a collision, use the dynamic reconstruction method applied to image space error metric in claim 3 to extract the continuous multi-resolution mesh representation corresponding to the node; or The continuous multi - resolution grid representation extracted by the dynamic reconstruction method applied to object space error metric according to claim 6 is further used to extract the corresponding continuous multi - resolution grid representation by the dynamic reconstruction method applied to image space error metric according to claim 3.

9. A large-scale triangular mesh continuous detail hierarchical representation device, characterized in that, The device includes: A top - level partitioning unit, configured to perform top - level partitioning on the obtained large - scale complex scene graph to obtain a BVH structure; wherein, the leaf nodes in the BVH structure are determined as the bottom - level partitioning structure nodes, and the root node and intermediate nodes in the BVH structure are determined as the top - level partitioning structure nodes; A bottom - level partitioning unit, configured to perform bottom - level space partitioning based on the triangular meshes included in the bottom - level partitioning structure nodes to generate a corresponding initial patch representation set with patches as the granularity; wherein, during the process of performing bottom - level space partitioning, the bottom - level partitioning unit is further configured to identify boundary vertices at the boundaries of each level of partitioning; determine the dependency relationship between the boundary vertices and the initial patch representations in the initial patch representation set; and update the geometric information of the initial patch representations in the corresponding initial patch representation set based on the dependency relationship when the boundary vertices change. A grid representation generation unit, configured to perform grid simplification operations from the bottom - level partitioning structure nodes to the top - level partitioning structure nodes based on the initial patch representation set to obtain a continuous multi - resolution grid representation; wherein, the continuous multi - resolution grid representation includes a simplified set of target patch representations.

10. A computer device, characterized in that, Including: A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the large - scale triangular mesh continuous detail hierarchical representation method according to any one of claims 1 to 2.

11. A computer device, characterized in that, Including: A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the dynamic reconstruction method applied to image space error metric according to claim 4.

12. A computer device, characterized in that, Including: A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the dynamic reconstruction method applied to object space error metric according to claim 7.

13. A computer device, characterized in that, Including: A memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the dynamic reconstruction method applied to unified error metric according to claim 8.

14. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer - readable storage medium, and the computer instructions are used to cause a computer to execute the large - scale triangular mesh continuous detail hierarchical representation method according to any one of claims 1 to 2.

15. A computer-readable storage medium, characterized in that, Computer instructions are stored on the computer - readable storage medium, and the computer instructions are used to cause a computer to execute the dynamic reconstruction method applied to image space error metric according to claim 4.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to execute the dynamic reconstruction method for object space error metric according to claim 7.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to execute the dynamic reconstruction method for unified error metric according to claim 8.

Images