Grid model and scene optimization method and device, storage medium and processor

Abstract

The embodiment of the invention provides a grid model and scene optimization method and device, a storage medium, a processor and a computer program product, and belongs to the technical field of digital twinning. The grid model optimization method comprises the following steps: detecting and deleting repeated points, lines and surfaces in a three-dimensional grid model based on a redundancy detection and removal algorithm; the nodes, the edges and the weight values of the edges of the three-dimensional grid model are input into a pre-trained graph convolutional network model, a three-dimensional grid model conforming to a simplification target is obtained, training sets adopted for training the graph convolutional network model comprise three-dimensional grid model sets which are not simplified and are processed according to simplification rules and have different simplification levels, and the three-dimensional grid model sets are obtained; the step of training the graph convolutional network model comprises the substeps of comparing and learning a rule that edges in the corresponding three-dimensional grid model between different simplification levels are merged. According to the optimization method of the grid model, the data size of the three-dimensional model can be remarkably reduced, and the operation loading speed of a digital twin application scene is increased.

Description

Technical Field

[0001] The present invention relates to the field of digital twin technology, and specifically to a method, device, storage medium, processor and computer program product for optimizing a grid model and a scene. Background Art

[0002] Industrial enterprises are shifting from traditional physical equipment and manual monitoring to the use of digital models and virtual reality (VR) / augmented reality (AR) technology for management and process optimization. Digital twin technology can provide real-time status monitoring, fault prediction and performance optimization by virtually simulating equipment and systems in the physical world. However, 3D models exported from industrial design software (such as Bentley, AutoCAD, Inventor, Solidworks) usually contain a large amount of geometric data and complex texture information. When these 3D models with large amounts of data and a large number of faces are run in digital twin application scenarios, they often cause performance problems such as slow loading and rendering jams. Summary of the invention

[0003] The purpose of an embodiment of the present invention is to provide a mesh model optimization method that can significantly reduce the data volume of the three-dimensional model and improve the running and loading speed of the digital twin application scenario.

[0004] In order to achieve the above object, an embodiment of the present invention provides a mesh model optimization method for simplifying a three-dimensional mesh model exported by industrial design software. The mesh model optimization method includes:

[0005] Based on the redundancy detection and removal algorithm, it detects and deletes duplicate points, lines and surfaces in the 3D mesh model;

[0006] The nodes, edges, and edge weights of the 3D mesh model are input into the pre-trained graph convolutional network model to obtain a 3D mesh model that meets the simplification goal.

[0007] The training set used for training the graph convolutional network model includes a set of unsimplified 3D mesh models and a set of 3D mesh models processed according to simplification rules and meeting multiple simplification levels.

[0008] Training the graph convolutional network model includes: comparing and learning the rules of merging edges in corresponding three-dimensional mesh models of different simplification levels, or comparing and learning the rules of merging edges in an unsimplified three-dimensional mesh model and corresponding three-dimensional mesh models of different simplification levels;

[0009] The simplification target is one of a plurality of different simplification levels or a value between two simplification levels.

[0010] Optionally, the weight value is the length of the edge.

[0011] Optionally, the simplification level is determined based on a threshold of edges to be merged.

[0012] Optionally, the threshold is a ratio threshold, including 20%, 50%, and 80%.

[0013] Optionally, the mesh model optimization method further includes: reducing the file size of the three-dimensional mesh model processed by the graph convolutional network model by using a model compression algorithm,

[0014] Among them, the model compression algorithm includes a geometric compression algorithm and a texture compression algorithm;

[0015] The geometry compression algorithm uses linear transformation and quantization techniques to reduce the storage of vertex and face data;

[0016] The texture compression algorithm dynamically adjusts the texture compression ratio according to the display device parameters and resolution requirements.

[0017] Optionally, the type of graph convolutional network model used is consistent with the type of three-dimensional mesh model to be processed, where the types include: building type, person type, and equipment type.

[0018] On the other hand, an embodiment of the present invention provides a mesh model optimization device for simplifying a three-dimensional mesh model derived from industrial design software. The mesh model optimization device includes: a preprocessing module, a merging and simplification module, and a pre-trained graph convolutional network model.

[0019] Among them, the preprocessing module detects and deletes repeated points, lines, and surfaces in the three-dimensional mesh model based on the redundancy detection and removal algorithm;

[0020] The merging and simplification module inputs the nodes, edges, and edge weights of the 3D mesh model into the pre-trained graph convolutional network model to obtain a 3D mesh model that meets the simplification goal.

[0021] The training set used for training the graph convolutional network model includes a set of unsimplified 3D mesh models and a set of 3D mesh models processed according to simplification rules and meeting multiple simplification levels.

[0022] Training the graph convolutional network model includes: comparing and learning the rules of merging edges in corresponding three-dimensional mesh models of different simplification levels, or comparing and learning the rules of merging edges in an unsimplified three-dimensional mesh model and corresponding three-dimensional mesh models of different simplification levels;

[0023] The simplification target is one of a plurality of different simplification levels or a value between two simplification levels.

[0024] Optionally, in the merge and reduce modules, the weight value is the length of the edge.

[0025] Optionally, in the merge and simplify module, the simplification level is determined based on a threshold of edges to be merged.

[0026] Optionally, the threshold is a ratio threshold, including 20%, 50%, and 80%.

[0027] Optionally, the mesh model optimization device further includes a model lightweight module, and further reduces the file size of the three-dimensional mesh model processed by the graph convolutional network model through a model compression algorithm.

[0028] Among them, the model compression algorithm includes a geometric compression algorithm and a texture compression algorithm;

[0029] The geometry compression algorithm uses linear transformation and quantization techniques to reduce the storage of vertex and face data;

[0030] The texture compression algorithm dynamically adjusts the texture compression ratio according to the display device parameters and resolution requirements.

[0031] Optionally, in the merging and simplification module, the type of graph convolutional network model used is consistent with the type of three-dimensional mesh model to be processed, where the types include: building type, person type, and equipment type.

[0032] On the other hand, an embodiment of the present invention provides a scene optimization method for optimizing an industrial digital twin scene, the scene optimization method comprising:

[0033] Optimize the three-dimensional mesh model initially obtained in the industrial digital twin scene according to the mesh model optimization method of the present application, and obtain three-dimensional mesh models of multiple preset simplified levels;

[0034] Based on the distance perception algorithm, three-dimensional mesh models with different simplification levels are loaded according to different viewing distance ranges;

[0035] Based on the switching transition algorithm, the switching effect of 3D mesh models with different simplification levels in different viewing distances is processed to avoid visual jumps;

[0036] Based on the texture optimization algorithm and the lighting optimization algorithm, the rendering effect of the three-dimensional mesh model included in the current scene is improved.

[0037] On the other hand, an embodiment of the present invention provides a scene optimization device for optimizing an industrial digital twin scene, the scene optimization device comprising: an optimization device for a grid model according to the present application, a dynamic loading module, a detail management module, a texture optimization module, and a global optimization module.

[0038] Among them, the mesh model optimization device is used to optimize the three-dimensional mesh model initially obtained in the industrial digital twin scene, and obtain a three-dimensional mesh model of multiple preset simplified levels;

[0039] Dynamic loading module, based on distance perception algorithm, loads 3D mesh models of different simplification levels according to different viewing distance ranges;

[0040] The detail management module handles the switching effect of 3D mesh models with different simplification levels at different viewing distances based on the switching transition algorithm to avoid visual jumps.

[0041] Texture optimization module, based on texture optimization algorithm, improves the rendering effect of the 3D mesh model included in the current scene;

[0042] The global optimization module improves the rendering effect of the three-dimensional mesh model included in the current scene based on the lighting optimization algorithm.

[0043] On the other hand, an embodiment of the present invention provides a processor configured to: execute the grid model optimization method of the present application, or execute the scene optimization method of the present application.

[0044] On the other hand, an embodiment of the present invention provides a machine-readable storage medium, which stores instructions, and when the instructions are executed by a processor, the processor is configured to: execute the grid model optimization method of the present application, or execute the scene optimization method of the present application.

[0045] On the other hand, an embodiment of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements the grid model optimization method of the present application, or executes the scene optimization method of the present application.

[0046] Through the above technical scheme, the three-dimensional mesh model is first preprocessed, and repeated points, lines, and surfaces in the three-dimensional mesh model are detected and deleted. Then, the nodes, edges, and edge weights of the preprocessed three-dimensional mesh model are input into a pre-trained graph convolutional network model to obtain a three-dimensional mesh model that meets the simplification goal. The training set used to train the graph convolutional network model includes a set of three-dimensional mesh models processed at different simplification levels. During the training process, the rules for merging edges in three-dimensional mesh models corresponding to different simplification levels are compared and learned, so that when the three-dimensional mesh model is processed by the pre-trained graph convolutional network model, a three-dimensional mesh model that meets the simplification goal can be obtained.

[0047] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0049] Figure 1 is a schematic diagram schematically showing an application environment of a grid model optimization method according to an embodiment of the present application;

[0050] Figure 2 is a schematic diagram schematically showing a flow chart of a method for optimizing a grid model according to an embodiment of the present application;

[0051] Figure 3 is a schematic diagram schematically showing a data flow of a scene optimization method according to an embodiment of the present application;

[0052] Figure 4 Schematically shows the internal structure of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0054] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0055] The grid model optimization method provided in this application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The terminal 102 is a client terminal used by the user responsible for network model creation and processing, which can be, for example, a browser web terminal, a native APP terminal, an H5 web terminal, or a mini program terminal. The server 104 can be a server that provides integrated services for creating and optimizing grid models. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices, and the server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0056] Figure 2 The flowchart of the method for optimizing the mesh model according to the embodiment of the present application is schematically shown. It should be noted that the model exported by the industrial design software usually contains a large amount of redundant geometric data, which causes the complexity and volume of the model to increase unnecessarily. The purpose of this embodiment is to minimize the redundant elements in the recognition model and merge them into a single element.

[0057] In this embodiment, this method is mainly applied to the above Figure 1 Taking the server 104 in FIG. 1 as an example, the following steps are included:

[0058] Step 201, based on a redundancy detection and removal algorithm, detecting and deleting duplicate points, lines, and surfaces in a three-dimensional mesh model;

[0059] Step 202: Input the nodes, edges and edge weights of the three-dimensional mesh model into a pre-trained graph convolutional network model to obtain a three-dimensional mesh model that meets the simplification goal.

[0060] It should be noted that in step 201, the initial three-dimensional mesh model is a three-dimensional mesh model exported by industrial design software, which may include repeated points, lines, and surfaces. Step 201 is based on a redundant detection and removal algorithm, and deleting repeated points, lines, and surfaces does not affect the accuracy of the model. In step 202, the training set used to train the graph convolutional network model includes a set of unsimplified three-dimensional mesh models and a set of three-dimensional mesh models processed according to simplification rules and conforming to multiple different simplification levels; training the graph convolutional network model includes: comparing and learning the rules of merging edges in corresponding three-dimensional mesh models of different simplification levels, or comparing and learning the rules of merging edges in unsimplified three-dimensional mesh models and corresponding three-dimensional mesh models of different simplification levels; the simplification target is a simplification level among multiple different simplification levels or an intermediate value between two simplification levels.

[0061] Specifically, in step 201, the following preprocessing steps may also be included: detection and optimization of duplicate points, duplicate lines and duplicate faces in the exported model to reduce redundant data of the model. In the model export stage, similar geometric elements are automatically identified and merged using geometric analysis and graphics processing techniques. The preprocessed three-dimensional mesh model is processed by a graph convolutional network model in step 202 to further reduce the number of nodes and edges included in the model. Among them, the weight value input into the pre-trained graph convolutional network model is the length value of the edge, and the simplification level is determined according to the threshold of the edge to be merged, wherein the threshold can be a proportional threshold: 20%, 50%, 80%. For example, the number of edges of the preprocessed three-dimensional mesh model is regarded as 100%, and the number of edges is reduced to 50% after merging by the graph convolutional network model.

[0062] It should be noted that the simplification level can be set as needed, for example, set to three levels of low precision, medium precision, and high precision, or set to ten different levels from 1 to 10. When training the graph convolutional network model, at least the corresponding training should be performed according to the various set simplification levels. When using the trained graph convolutional network model, it is preferred to use the trained simplification level as the simplification target. If the simplification target to be achieved is the value between two simplification levels, the relevant parameters in the graph convolutional network model can be modified according to the interpolation method.

[0063] In some implementations, step 202 may also include the following implementation steps: optimizing the model structure by merging and simplifying points, lines, and surfaces. Designing a merging algorithm based on spatial distribution characteristics and topological structure can significantly reduce the number of model surfaces and volume while maintaining geometric accuracy.

[0064] In some embodiments, an improved geometry detection and removal algorithm is also used, which is based on octree space partitioning and KD tree search technology, and can efficiently identify redundant elements in the model and merge them into a single element. In order to further reduce the complexity of the model, the geometry merging algorithm can also merge adjacent geometric faces to reduce the number of faces of the model while ensuring that the geometric accuracy is not reduced.

[0065] In some embodiments, the file size of the three-dimensional mesh model processed by the graph convolutional network model is also reduced by a model compression algorithm, wherein the model compression algorithm includes a geometry compression algorithm and a texture compression algorithm; the geometry compression algorithm uses linear transformation and quantization technology to reduce the storage of vertex and face data; the texture compression algorithm dynamically adjusts the texture compression ratio according to display device parameters and resolution requirements. The technologies used include texture compression formats (such as DXT, ASTC, etc.) and material simplification (such as reducing the resolution and bit depth of texture maps).

[0066] In some implementations, a training data set is also designed according to the type of the three-dimensional mesh model, and a graph convolutional network model of a corresponding type is obtained. For example, for different classifications such as buildings, people, and equipment, corresponding training data sets are collected respectively, and then a graph convolutional network model of a building, a graph convolutional network model of a person, and a graph convolutional network model of an equipment are obtained by training respectively. When processing a three-dimensional mesh model, the type of the graph convolutional network model used is consistent with the type of the three-dimensional mesh model to be processed.

[0067] Compared with the prior art, the technical advantages of this embodiment are:

[0068] 1. Established a multi-layer simplification mechanism for the model

[0069] First, the redundant detection and removal algorithm is used for preprocessing to remove duplicate elements in the initial 3D mesh model. Then, the number of nodes and edges of the model is further reduced based on the pre-trained graph convolutional network model. The number of faces of the model is further reduced through the improved geometry detection and removal algorithm.

[0070] 2. Established a model lightweight mechanism

[0071] Use geometric compression algorithms to reduce model size, and make the model lighter without significantly reducing visual effects through vertex compression, edge compression, and face compression. Reduce the model's storage space and amount of data transmitted by compressing and optimizing the model's material and texture data.

[0072] An embodiment of the present invention provides a mesh model optimization device, which is used to simplify a three-dimensional mesh model exported by industrial design software. The mesh model optimization device includes: a preprocessing module, a merging and simplification module, and a pre-trained graph convolutional network model, wherein the preprocessing module detects and deletes repeated points, lines, and surfaces in the three-dimensional mesh model based on a redundancy detection and removal algorithm; the merging and simplification module inputs the nodes, edges, and edge weight values ​​of the three-dimensional mesh model into the pre-trained graph convolutional network model to obtain a three-dimensional mesh model that meets the simplification target, wherein the training set used for training the graph convolutional network model includes an unsimplified three-dimensional mesh model set and a three-dimensional mesh model set processed according to simplification rules and meeting multiple different simplification levels; training the graph convolutional network model includes: comparing and learning the rules of merging edges in corresponding three-dimensional mesh models of different simplification levels, or comparing and learning the rules of merging edges in the unsimplified three-dimensional mesh model and the corresponding three-dimensional mesh models of different simplification levels; the simplification target is one simplification level among multiple different simplification levels or an intermediate value between two simplification levels.

[0073] In some embodiments, in the merging and simplifying module, the weight value is the length of the edge; the simplification level is determined according to the threshold of the edges to be merged. Optionally, the threshold is a ratio threshold, including 20%, 50%, and 80%.

[0074] In some embodiments, the mesh model optimization device also includes a model lightweight module, and also reduces the file size of the three-dimensional mesh model processed by the graph convolutional network model through a model compression algorithm, wherein the model compression algorithm includes a geometric compression algorithm and a texture compression algorithm; the geometric compression algorithm uses linear transformation and quantization technology to reduce the storage amount of vertex and face data; the texture compression algorithm dynamically adjusts the texture compression ratio according to the display device parameters and resolution requirements.

[0075] In some embodiments, in the merging and simplification module, the type of graph convolutional network model used is consistent with the type of three-dimensional mesh model to be processed, where the types include: building type, character type, and equipment type.

[0076] The embodiment of the present invention provides a scene optimization method for optimizing industrial digital twin scenes. Figure 1 Based on the optimization method embodiment of the mesh model shown in FIG. 1 , other technical means are also used to further optimize the loading and rendering methods of the three-dimensional mesh model in the scene, such as Figure 3 As shown, the data conversion process of the scene optimization method is shown. The scene optimization method includes the following steps:

[0077] Step 301: Optimizing the three-dimensional mesh model initially obtained in the industrial digital twin scene according to the mesh model optimization method of the present application, and obtaining three-dimensional mesh models of multiple preset simplified levels;

[0078] Step 302: Based on the distance perception algorithm, load three-dimensional mesh models of different simplification levels according to different viewing distance ranges;

[0079] Step 303: Based on the switching transition algorithm, the switching effect of the three-dimensional mesh models with different simplification levels in different viewing distance ranges is processed to avoid visual jumps;

[0080] Step 304: Based on the texture optimization algorithm and the lighting optimization algorithm, the rendering effect of the three-dimensional mesh model included in the current scene is improved.

[0081] An embodiment of the present invention provides a scene optimization device for optimizing an industrial digital twin scene, the scene optimization device comprising: an optimization device for a mesh model according to the present application, a dynamic loading module, a detail management module, a texture optimization module and a global optimization module, wherein the mesh model optimization device is used to optimize the three-dimensional mesh model initially obtained in the industrial digital twin scene, and obtain a three-dimensional mesh model of a preset plurality of simplified levels; the dynamic loading module, based on a distance perception algorithm, loads three-dimensional mesh models of different simplified levels according to different viewing ranges; the detail management module, based on a switching transition algorithm, processes the switching effect of three-dimensional mesh models of different simplified levels in different viewing ranges to avoid visual jumps; the texture optimization module, based on a texture optimization algorithm, improves the rendering effect of the three-dimensional mesh model included in the current scene; the global optimization module, based on an illumination optimization algorithm, improves the rendering effect of the three-dimensional mesh model included in the current scene.

[0082] Based on the above-mentioned scene optimization device embodiment, the present invention further provides a preferred embodiment of a scene optimization device, including the following modules:

[0083] 1) Fully automatic geometry optimization module

[0084] An improved geometry detection and removal algorithm is used. This algorithm is based on octree space partitioning and KD tree search technology, which can efficiently identify redundant elements in the model and merge them into a single element. In order to further reduce the complexity of the model, the geometry merging algorithm can also merge adjacent geometric faces to reduce the number of faces in the model while ensuring that the geometric accuracy is not reduced.

[0085] Redundant data detection and removal algorithm: This algorithm mainly detects and optimizes duplicate points, duplicate lines and duplicate surfaces in the exported model to reduce redundant data in the model. In the model export stage, geometric analysis and graphics processing technology are used to automatically identify and merge similar geometric elements.

[0086] Geometric merging and simplification strategy: Optimize the model structure by merging and simplifying points, lines, and surfaces. Design a merging algorithm based on spatial distribution characteristics and topological structure to significantly reduce the number of model faces and volume while maintaining geometric accuracy.

[0087] 2) Model lightweight module

[0088] The model lightweight module is mainly used to reduce the file size and complexity of the model to improve the loading and rendering speed. By compressing the model data, including the compression of geometry data and texture data, the storage occupancy and transmission time of the model can be significantly reduced. Geometry compression uses linear transformation and quantization technology to reduce the storage volume of vertex and face data; material and texture compression is optimized through image compression algorithms (such as JPEG, PNG compression) and 3D graphics-specific compression formats (such as DDS, PVR). In addition, an adaptive texture compression strategy is used to dynamically adjust the texture compression ratio according to different display devices and resolution requirements.

[0089] Model compression technology: Use geometric compression algorithms to reduce the model size, and make the model lighter without significantly reducing the visual effect through vertex compression, edge compression, and face compression.

[0090] Material and texture compression algorithm: By compressing and optimizing the material and texture data of the model, the storage space and transmission data volume of the model are reduced. The technologies used include texture compression formats (such as DXT, ASTC, etc.) and material simplification (such as reducing the resolution and bit depth of texture maps).

[0091] 3) Dynamic loading and unloading of modules

[0092] The dynamic loading and unloading module realizes dynamic management of models through perspective perception and distance perception technology. The perspective perception unloading algorithm manages model data by constructing a spatial partitioning structure of the scene (such as quadtree and octree). When the user's perspective changes, the system can quickly determine which models are within the field of view and which models need to be unloaded, thereby dynamically releasing video memory and memory resources. Distance perception loading technology dynamically selects to load different LOD models by calculating the distance between the user's perspective and the model. With this strategy, the system can reduce unnecessary high-precision model loading and significantly optimize resource utilization.

[0093] View-aware offloading algorithm: This algorithm can detect the user's viewpoint in the digital twin scene in real time and automatically offload models that are not visible from the current viewpoint to free up video memory and memory resources. Based on changes in user viewpoints, spatial partitioning algorithms (such as quadtrees and octrees) are used to optimize the model management and offloading process.

[0094] Distance-aware loading technology: Dynamically load models of different precisions based on the distance between the user's viewing angle and the model. When the viewing distance is far, low-precision models are loaded first; when the viewing distance is close, high-precision models are switched to ensure a balance between visual effects and resource consumption.

[0095] 4) Level of Detail (LOD) module

[0096] The Level of Detail (LOD) module optimizes the efficiency of model rendering through automatic LOD generation and switching smooth transition technology. The automatic LOD generation algorithm automatically generates high, medium, and low precision versions according to the geometric complexity of the model, ensuring that the appropriate LOD model can be loaded at different viewing distances. When the LOD of the model is switched, the LOD switching smooth transition technology uses progressive detail blending technology to gradually replace the low-precision model with a high-precision model, avoiding visual jumps and flickering. This strategy significantly improves the user's visual experience at different viewing distances.

[0097] Automatic LOD generation algorithm: Based on the geometric complexity of the model and the application scenario requirements, high, medium and low precision model versions are automatically generated. Through geometric simplification algorithms (such as progressive mesh simplification, QEM simplification, etc.), LOD models suitable for different viewing distances are automatically generated.

[0098] LOD switching smooth transition technology: During the LOD switching process of the model, transition technology (such as progressive detail blending and fade-in and fade-out technology) is used to avoid visual jumps when LOD switching and improve user experience.

[0099] 5) Texture optimization module

[0100] The texture optimization module optimizes the texture resource management in the scene by dynamically adjusting the texture resolution and optimizing the texture loading strategy. The dynamic texture resolution adjustment algorithm automatically selects the appropriate texture resolution according to the complexity of the scene and the usage of system resources, thereby reducing video memory and memory consumption while ensuring visual effects. The texture loading optimization strategy adopts delayed loading and on-demand loading methods, loading the corresponding texture resources only when the user's perspective requires it, further reducing memory usage and loading time.

[0101] Dynamic texture resolution adjustment algorithm: Dynamically adjust the texture resolution and details according to scene requirements and resource usage. Automatically select the appropriate texture resolution by real-time monitoring of GPU memory usage and performance load.

[0102] Texture loading optimization strategy: Use delayed loading and on-demand loading strategies to load texture resources only when needed to reduce memory usage and loading time.

[0103] 6) Overall scene optimization module

[0104] The overall scene optimization module improves the rendering efficiency and visual effects of the scene through lighting optimization and object merging and optimization technology. The lighting optimization algorithm pre-calculates the lighting effects in the scene through global lighting pre-calculation, reducing the burden of real-time calculation. At the same time, real-time lighting dynamic adjustment dynamically adjusts the lighting effects according to the user's perspective and scene changes to ensure the continuity and authenticity of the visual effects. The object merging and optimization technology adopts a batch merging strategy for static objects in the scene, reducing rendering batches and improving the rendering efficiency of the scene.

[0105] Lighting optimization algorithm: Through global lighting pre-calculation and real-time lighting dynamic adjustment, the lighting effect in the scene is optimized to reduce the burden of real-time calculation. Combined with physically based rendering technology (PBR), the authenticity and efficiency of the lighting effect are ensured.

[0106] Object merging and optimization technology: For static objects in the scene, batch merging and optimization technology is used to reduce the number of objects and rendering batches, and improve scene rendering efficiency.

[0107] Compared with the prior art, the technical advantages of this embodiment are:

[0108] 1) Significantly improve system operation efficiency

[0109] Optimize loading and rendering speed: Through geometric element optimization, model lightweighting, dynamic loading and unloading strategies, the size and complexity of model files are reduced, and the loading and rendering speed of models is accelerated. In particular, dynamic loading and unloading technology can load the required model in real time according to the viewing angle, avoiding unnecessary resource consumption and significantly improving the overall operating efficiency of the digital twin scene.

[0110] Reduce rendering latency: LOD management technology and texture optimization strategies are used to ensure that appropriate model accuracy and textures are loaded at different viewing distances and perspectives, reducing the latency of rendering highly complex models and improving the system's response speed and rendering performance.

[0111] 2) Reduce resource consumption

[0112] Reduce video memory and memory usage: Through perspective-aware unloading and distance-aware loading technology, it can dynamically judge and manage the models and textures that need to be loaded, unload models that are not visible in the current perspective, reduce the usage of video memory and memory, and thus reduce the burden on hardware resources.

[0113] Optimize storage and transmission requirements: Geometry compression and material compression technologies reduce the file size of 3D models and textures, bringing significant optimization effects in data storage and network transmission, especially in industrial Internet and cloud application scenarios that require frequent data exchange, greatly improving operational efficiency and stability.

[0114] 3) Improve user experience

[0115] Smooth interactive experience: By reducing loading time and improving rendering speed, users can get a smoother operating experience, especially in complex digital twin scenes. Users can quickly switch perspectives, zoom and rotate models, and experience more natural and real-time interactive effects.

[0116] High-quality visual effects: LOD switching smooth transition technology and dynamic texture optimization are used to ensure that appropriate model accuracy and texture quality are displayed at different viewing distances and viewing angles, avoiding visual jumps and texture distortion during LOD switching and improving the overall visual experience.

[0117] 4) Enhance the scalability and maintainability of the system

[0118] Modular design: The present invention adopts modular design, and the optimization algorithm and management strategy are independent of their respective modules. This design improves the scalability and maintainability of the system. The introduction of new technologies or the upgrade of existing technologies can be carried out without interfering with other modules, which is convenient for subsequent expansion and maintenance.

[0119] Adaptable to various application scenarios: The optimization method of the present invention is not only applicable to digital twin applications in the industrial field, but can also be extended to other fields that require efficient 3D model management and rendering, such as architectural visualization, virtual reality, augmented reality, game development, etc. The versatility and adaptability of the system give it a wider application prospect and market value.

[0120] An embodiment of the present invention provides a scene optimization device, which includes a processor and a memory. The above-mentioned preprocessing module, merging and simplification module, model lightweight module, dynamic loading module, detail management module, texture optimization module, global optimization module and pre-trained graph convolutional network model are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0121] The processor contains a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set to optimize the industrial digital twin scene or the 3D mesh model used in the scene by adjusting the kernel parameters.

[0122] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0123] An embodiment of the present invention provides a machine-readable storage medium, on which instructions are stored. When the instructions are executed by a processor, the processor is configured to: execute the grid model optimization method of the present application, or execute the scene optimization method of the present application.

[0124] An embodiment of the present invention provides a processor configured to: execute the grid model optimization method of the present application, or execute the scene optimization method of the present application.

[0125] The embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored in the memory and can be run on the processor. When the processor executes the program, the steps of the method for optimizing the grid model of the present application are implemented, or the steps of the method for optimizing the scene of the present application are executed. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0126] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing the optimization method steps of initializing the grid model of the present application, or executing the program of the scene optimization method steps of the present application.

[0127] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, a data quality inspection method of the present application is implemented. The display screen A04 of the computer device can be a liquid crystal display or an electronic ink display, and the input device A05 of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0128] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0129] In one embodiment, the data quality inspection, evaluation, and optimization device provided by the present application can be implemented in the form of a computer program. The computer program can be used in Figure 4 The computer device shown in the figure is run. The memory of the computer device can store various program modules constituting the online payment device of the present application, such as the preprocessing module, merging and simplification module, model lightweight module, dynamic loading module, detail management module, map optimization module, and global optimization module data in the above device embodiment. The computer program composed of various program modules enables the processor to execute the steps of the data quality inspection method of each embodiment of the present application described in this specification.

[0130] Figure 4 The computer device shown can perform step 301 through the grid model optimization device in the scene optimization device of the present application. The computer device can perform step 302 through the dynamic loading module. And so on.

[0131] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0132] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0133] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0135] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0136] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0137] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0138] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0139] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A mesh model optimization method for simplifying a three-dimensional mesh model derived from industrial design software, the mesh model optimization method comprising: Based on a redundancy detection and removal algorithm, detecting and deleting duplicate points, lines, and surfaces in the three-dimensional mesh model; Input the nodes, edges and edge weights of the three-dimensional mesh model into a pre-trained graph convolutional network model to obtain a three-dimensional mesh model that meets the simplification goal, The training set used to train the graph convolutional network model includes a set of unsimplified three-dimensional mesh models and a set of three-dimensional mesh models processed according to simplification rules and conforming to multiple different simplification levels; Training the graph convolutional network model includes: comparing and learning the rules of merging edges in the three-dimensional mesh models corresponding to the different simplification levels, or comparing and learning the rules of merging edges in the unsimplified three-dimensional mesh model and the corresponding three-dimensional mesh models of different simplification levels; The simplification target is a simplification level among the multiple different simplification levels or a value between two simplification levels.

2. The method for optimizing a mesh model according to claim 1, characterized in that: The weight value is the length of the edge.

3. The method for optimizing a mesh model according to claim 1, characterized in that: The simplification level is determined according to a threshold of edges to be merged.

4. The method for optimizing a mesh model according to claim 3, characterized in that: The threshold is a ratio threshold, including 20%, 50%, and 80%.

5. The method for optimizing a grid model according to any one of claims 1 to 4, characterized in that: Also includes: The file size of the three-dimensional mesh model processed by the graph convolutional network model is reduced by a model compression algorithm. Wherein, the model compression algorithm includes a geometric compression algorithm and a texture compression algorithm; The geometric compression algorithm uses linear transformation and quantization technology to reduce the storage volume of vertex and surface data; The texture compression algorithm dynamically adjusts the texture compression ratio according to display device parameters and resolution requirements.

6. The method for optimizing a mesh model according to claim 1, characterized in that: The type of graph convolutional network model used is consistent with the type of three-dimensional mesh model to be processed, wherein the types include: building type, person type, and equipment type.

7. A mesh model optimization device for simplifying a three-dimensional mesh model derived from industrial design software, characterized in that: The grid model optimization device includes: a preprocessing module, a merging and simplification module and a pre-trained graph convolutional network model, Wherein, the preprocessing module detects and deletes repeated points, lines and surfaces in the three-dimensional grid model based on a redundancy detection and removal algorithm; The merging and simplification module inputs the nodes, edges and edge weights of the three-dimensional mesh model into a pre-trained graph convolutional network model to obtain a three-dimensional mesh model that meets the simplification goal. The training set used to train the graph convolutional network model includes a set of unsimplified three-dimensional mesh models and a set of three-dimensional mesh models processed according to simplification rules and conforming to multiple different simplification levels. Training the graph convolutional network model includes: comparing and learning the rules of merging edges in the three-dimensional mesh models corresponding to the different simplification levels, or comparing and learning the rules of merging edges in the unsimplified three-dimensional mesh model and the corresponding three-dimensional mesh models of different simplification levels; The simplification target is a simplification level among the multiple different simplification levels or a value between two simplification levels.

8. The grid model optimization device according to claim 7, characterized in that: In the merging and simplifying module, the weight value is the length of the edge.

9. The grid model optimization device according to claim 7, characterized in that: In the merge and simplify module, the simplification level is determined according to a threshold of edges to be merged.

10. The grid model optimization device according to claim 9, characterized in that: The threshold is a ratio threshold, including 20%, 50%, and 80%.

11. The grid model optimization device according to any one of claims 7 to 10, characterized in that: It also includes a model lightweight module, and reduces the file size of the three-dimensional mesh model processed by the graph convolutional network model through a model compression algorithm. Wherein, the model compression algorithm includes a geometric compression algorithm and a texture compression algorithm; The geometric compression algorithm uses linear transformation and quantization technology to reduce the storage volume of vertex and surface data; The texture compression algorithm dynamically adjusts the texture compression ratio according to display device parameters and resolution requirements.

12. The grid model optimization device according to claim 7, characterized in that: In the merging and simplification module, the type of graph convolutional network model used is consistent with the type of three-dimensional mesh model to be processed, wherein the types include: building type, person type, and equipment type.

13. A scene optimization method for optimizing industrial digital twin scenes, characterized in that: The scene optimization method comprises: The mesh model optimization method according to any one of claims 1 to 6 optimizes the three-dimensional mesh model initially obtained in the industrial digital twin scene, and obtains three-dimensional mesh models of multiple preset simplified levels; Based on the distance perception algorithm, three-dimensional mesh models with different simplification levels are loaded according to different viewing distance ranges; Based on the switching transition algorithm, the switching effect of 3D mesh models with different simplification levels in different viewing distances is processed to avoid visual jumps; Based on the texture optimization algorithm and the lighting optimization algorithm, the rendering effect of the three-dimensional mesh model included in the current scene is improved.

14. A scene optimization device for optimizing industrial digital twin scenes, characterized in that: The scene optimization device comprises: a dynamic loading module, a detail management module, a texture optimization module, a global optimization module and an optimization device for a grid model according to any one of claims 7 to 13, The mesh model optimization device is used to optimize the three-dimensional mesh model initially obtained in the industrial digital twin scene and obtain a three-dimensional mesh model of a preset plurality of simplified levels; The dynamic loading module loads three-dimensional mesh models of different simplification levels according to different viewing distance ranges based on a distance perception algorithm; The detail management module processes the switching effect of three-dimensional mesh models with different simplification levels in different viewing distances based on the switching transition algorithm to avoid visual jumps; The texture optimization module improves the rendering effect of the three-dimensional mesh model included in the current scene based on the texture optimization algorithm; The global optimization module improves the rendering effect of the three-dimensional mesh model included in the current scene based on the lighting optimization algorithm.

15. A processor, characterized in that: The method is configured to: execute the mesh model optimization method according to any one of claims 1 to 6, or execute the scene optimization method according to claim 13.

16. A machine-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to: execute the grid model optimization method according to any one of claims 1 to 6, or execute the scene optimization method according to claim 13.

17. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements the method for optimizing a mesh model according to any one of claims 1 to 6, or performs the method for optimizing a scene according to claim 13.

Images