Interactive garden design simulation system and method based on virtual reality
By constructing an improved graph neural network knowledge graph and digital twin model, the problems of low visualization and poor interactivity in traditional garden design have been solved, realizing intelligent and efficient interaction throughout the entire garden design process, and improving design quality and efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HANGZHOU GEJING ARCHITECTURAL LANDSCAPE DESIGN CO LTD
- Filing Date
- 2026-01-15
- Publication Date
- 2026-04-28
AI Technical Summary
Traditional garden design relies on two-dimensional drawings and static models, resulting in low visualization, vague expression of element relationships, difficulty in achieving real-time interaction and scheme optimization during the design process, and inability to meet the diverse design adjustment needs of designers, thus hindering the improvement of efficiency and quality in garden design.
By performing structured analysis and three-dimensional feature extraction on basic garden design data, an improved graph neural network knowledge graph is constructed. Combined with digital twin technology, a digital twin model of the site is built, realizing semantic association of garden design elements and accurate matching of virtual scenes, and supporting real-time implantation and adjustment of design interaction logic.
It enables intelligent processing of the entire process of landscape design, from data analysis to interactive simulation, improving the visualization and interactive experience of the design, helping designers to accurately grasp the rationality and feasibility of the design scheme, and improving the efficiency and quality of landscape design.
Smart Images
Figure CN121936285A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of virtual reality design technology for gardens, specifically an interactive garden design simulation system and method based on virtual reality. Background Technology
[0002] Landscape design, as a comprehensive endeavor integrating ecology, aesthetics, and function, traditionally relies heavily on two-dimensional drawings and static models. This results in low visualization levels, unclear relationships between elements, and difficulty in intuitively presenting the spatial effects and interactive experiences of the design scheme. Existing interactive design simulation schemes have relatively simple interactive logic, failing to meet the diverse design adjustment needs of designers and hindering real-time interaction and scheme optimization during the design process. This restricts the efficiency and quality improvement of landscape design and fails to adapt to the modern trend of refined and intelligent landscape design.
[0003] Therefore, an interactive garden design simulation system and method based on virtual reality are provided. Summary of the Invention
[0004] To address the aforementioned technical problems, the present invention aims to provide an interactive garden design simulation system and method based on virtual reality.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an interactive garden design simulation method based on virtual reality, the method comprising: The acquired basic data of garden design is subjected to structured analysis and three-dimensional feature extraction to obtain a semantic set of garden design elements. A garden design knowledge graph corresponding to the semantic set of garden design elements is constructed based on an improved graph neural network algorithm. Based on digital twin technology, a digital model is created from the basic data of garden design to obtain a site digital twin model of the corresponding garden; the garden design knowledge graph is then mapped to the site digital twin model to obtain an initial interactive virtual garden scene. Obtain basic design interaction instructions, and implant interactive logic into the initial interactive garden virtual scene based on the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.
[0006] Furthermore, the process of performing structured analysis and three-dimensional feature extraction on the acquired basic data of garden design to obtain the semantic set of garden design elements includes: The acquired basic landscape design data is unified in multiple data formats to obtain a standardized basic dataset; the standardized basic dataset is parsed in a hierarchical structure to obtain a hierarchical basic data text set; the hierarchical basic data text set is used for element entity recognition and 3D feature extraction to obtain an element entity set and a 3D feature set. Relationships are extracted from the set of element entities to obtain an entity relationship set. Based on the three-dimensional feature set, the element entity set is semantically enhanced and vectorized, transforming the element entities with integrated three-dimensional features into corresponding element entity semantic vectors and three-dimensional feature vectors. The entity relationship set is then fused with the corresponding element entity semantic vectors and three-dimensional feature vectors to obtain a semantic set of garden design elements.
[0007] Furthermore, the process of constructing a garden design knowledge graph corresponding to the semantic set of garden design elements based on the improved graph neural network algorithm includes: The similarity of the semantic vectors of the element entities in the semantic set of the garden design elements is calculated to obtain the entity semantic similarity matrix; Based on the entity semantic similarity matrix, an improved GAT graph attention neural network is used to cluster the element entities to obtain the theme clusters of garden design elements; Cluster cores are calculated for each theme cluster in the set of theme clusters of garden design elements to obtain the core entity node set of garden design knowledge graph. Semantic association edges are matched for the core entity node set according to the entity relationship set. Based on the semantic association edges and the core entity node set, the garden design knowledge graph is constructed.
[0008] Furthermore, the process of calculating the cluster centers of each theme cluster in the garden design element theme cluster set to obtain the core entity node set of the garden design knowledge graph includes: Traverse each theme cluster in the collection of garden design element theme clusters; extract the element entity semantic vector and three-dimensional feature vector of all element entities within the corresponding theme cluster; The method of feature concatenation and weighted fusion is used to generate a fused feature vector for each entity; the entity weight within the cluster is calculated based on the semantic similarity of the entity; taking each element entity within the topic cluster as the benchmark, the mean semantic similarity between it and all other element entities within the cluster is calculated. The mean similarity of all entities within a cluster is normalized to obtain the weight coefficient of each entity within the topic cluster; based on the fused feature vector and the weight coefficient, the weighted average cluster center vector of the topic cluster is calculated. Calculate the cosine similarity between the fused feature vector of each element entity within a topic cluster and the weighted average cluster center vector. Select the element entity with the highest cosine similarity as the core entity node corresponding to that topic cluster. If multiple element entities have the same similarity to the weighted average cluster center vector and all of them are the maximum values, then all of them are taken as core entity nodes. Summarize the core entity nodes corresponding to all topic clusters and denote them as the core entity node set.
[0009] Furthermore, based on digital twin technology, the process of digitally modeling the basic data of landscape design to obtain a site-specific digital twin model of the corresponding landscape includes: Based on digital twin technology, three-dimensional modeling is carried out according to the topographic data, plant category attribute data, garden structure parameter data and regional cultural adaptation standard data in the basic data of garden design. In this way, corresponding site topographic digital twin sub-model, plant digital twin sub-model library, structure digital twin sub-model library and cultural symbol digital twin sub-model are obtained. Based on the functional zoning plan of the garden site, the site topography digital twin sub-model, the plant models in the plant digital twin sub-model library, the structure models in the structure digital twin sub-model library, and the cultural symbol digital twin sub-model are spatially combined and assembled to obtain the initial version of the site digital twin model. Collision detection and optimization were performed on the initial version of the site digital twin model to obtain the final site digital twin model.
[0010] Furthermore, the process of associating and mapping the landscape design knowledge graph with the site's digital twin model to obtain an initial interactive virtual landscape scene includes: Construct a mapping rule base between a knowledge graph of landscape design and a digital twin model of the site; Semantic annotation is performed on each sub-model in the site digital twin model to generate a model semantic tag set; Based on the model semantic tag set and mapping rule base, the core entity nodes in the landscape design knowledge graph are matched with the sub-models in the site digital twin model to obtain the entity-model mapping relationship set; Based on the entity-model mapping relationship set, the semantic association edges and core entity node set in the garden design knowledge graph are associated with the site digital twin model. The associated site digital twin model is then imported into the virtual reality engine to construct an initial interactive garden virtual scene with semantic interaction capabilities.
[0011] Furthermore, the process of embedding interactive logic into the initial interactive garden virtual scene based on the parsing results of the basic design interaction instructions to obtain an interactive garden virtual scene includes: The acquired basic design interaction commands are parsed to obtain the command type and command parameter set, and then a command parsing rule base is constructed. Based on the instruction type and parameter information set in the instruction parsing rule base, a garden design interaction logic rule base is constructed. The logical rules in the interactive logic rule base are implanted into the virtual reality engine of the initial interactive garden virtual scene; the initial interactive garden virtual scene after the interactive logic is implanted is functionally tested and optimized to obtain the final interactive garden virtual scene.
[0012] A second aspect of the present invention also provides an interactive garden design simulation system based on virtual reality, including a knowledge graph construction module, a digital model construction module, a digital model mapping module, and a basic scene construction module; The knowledge graph construction module is used to perform structured parsing and three-dimensional feature extraction on the acquired basic data of garden design to obtain a semantic set of garden design elements, and to construct a garden design knowledge graph corresponding to the semantic set of garden design elements based on an improved graph neural network algorithm. The digital model building module, based on digital twin technology, performs digital modeling on the basic data of garden design to obtain the corresponding digital twin model of the garden site; The digital model mapping module is used to associate and map the garden design knowledge graph with the site digital twin model to obtain an initial interactive garden virtual scene; The scene construction module is used to obtain basic design interaction instructions, and to implant interactive logic into the initial interactive garden virtual scene according to the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.
[0013] Compared with existing technologies, the beneficial effects of this invention are as follows: The interactive garden design simulation method and system based on virtual reality provided by this invention, through structured analysis and three-dimensional feature extraction of basic garden design data, combined with the construction of a knowledge graph using an improved graph neural network, fully explores the semantic relationships of design elements, providing solid intelligent semantic support for the virtual scene; by using digital twin technology to construct a digital twin model of the site and achieving association mapping with the knowledge graph, it ensures accurate matching between the virtual scene and the actual site characteristics, intuitively presenting the spatial effect of the garden design; by parsing basic design interaction instructions and embedding interaction logic, it enriches the interactive functions of the virtual scene, realizing real-time interaction and scheme adjustment during the design process. The entire method realizes intelligent processing of the entire process of garden design from data analysis and virtual modeling to interactive simulation, improving the visualization and interactive experience of the design, helping designers accurately grasp the rationality and feasibility of the design scheme, improving the efficiency and quality of garden design, and at the same time, the various modules of the system work together efficiently, adapting to different types of garden design scenarios, and have broad application value. Attached Figure Description
[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.
[0015] Figure 1 This is a schematic diagram illustrating the steps of an interactive garden design simulation method based on virtual reality.
[0016] Figure 2 This is a schematic diagram of a module of an interactive garden design simulation system based on virtual reality. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0019] Example 1: like Figure 1 As shown, an interactive garden design simulation method based on virtual reality includes the following steps: Step S1: Perform structured analysis and three-dimensional feature extraction on the acquired basic data of garden design to obtain a semantic set of garden design elements, and construct a garden design knowledge graph corresponding to the semantic set of garden design elements based on an improved graph neural network algorithm. Step S2: Based on digital twin technology, digital model the basic data of garden design to obtain the site digital twin model of the corresponding garden; associate and map the garden design knowledge graph with the site digital twin model to obtain the initial interactive garden virtual scene; Step S3: Obtain basic design interaction instructions, and implant interactive logic into the initial interactive garden virtual scene according to the parsing result of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.
[0020] It should be further explained that, in the specific implementation process, the process of performing structured analysis and three-dimensional feature extraction on the acquired basic data of landscape design to obtain the semantic set of landscape design elements includes: It should be noted that the basic data for garden design includes topographic data, plant category attribute data, garden structure parameter data, and regional cultural adaptation standard data. The topographic data includes, but is not limited to, elevation data, slope and aspect data, hydrological distribution data, and soil type data of the garden site. These are fundamental data used to characterize the physical form of the site, such as elevation model data obtained through UAV lidar scanning and soil pH and fertility data obtained through soil sampling analysis. The plant category attribute data refers to trees, shrubs, herbs, vines, etc., suitable for the climatic conditions of the garden site. Biological attribute data of plants, including but not limited to plant height, crown width, growth cycle, suitable light intensity, water requirements, and seasonal changes in landscape; parameter data of garden structures, including but not limited to the size parameters, material properties, structural forms, and construction process requirements of garden facilities such as pavilions, pergolas, landscape walls, water features, and garden paths; regional cultural adaptation standard data, including but not limited to design standards and guidelines that match the historical culture, folk customs, and architectural style of the area where the garden site is located, such as the white wall and black tile style standard of Jiangnan water town gardens and the symmetrical layout guidelines of northern royal gardens.
[0021] Optionally, in this embodiment of the application, the acquired basic data of landscape design is unified in multiple source data formats to obtain a standardized basic dataset; It should be noted that the basic data for garden design exists in multiple heterogeneous formats, such as LAS point cloud format for topographic data, Excel spreadsheet format for plant category attribute data, CAD drawing format for garden structure parameter data, and PDF document format for regional cultural adaptation standard data. The data conversion tool will uniformly convert the data in different formats into Extensible Markup Language format to achieve standardized storage of the data structure.
[0022] The standardized basic dataset is subjected to hierarchical structured parsing to obtain a hierarchical basic data text set. The hierarchical structured parsing is carried out according to a three-level structure of site layer, element layer, and attribute layer. The site layer parsing extracts macro information such as the boundary range and functional zoning of the garden site. The element layer parsing extracts core design elements such as topography, plants, structures, and cultural symbols. The attribute layer parsing extracts the parameter indicators and constraints corresponding to each element. The hierarchical structure division and information extraction can be completed based on regular expressions and path query language.
[0023] The hierarchical basic data text set is subjected to element entity recognition and 3D feature extraction to obtain an element entity set and a 3D feature set. The element entity recognition includes extracting core entities from each level of data, including but not limited to terrain entities, plant entities, structure entities, and cultural entities, such as slopes in terrain entities, camphor trees in plant entities, granite garden paths in structure entities, and latticed windows in cultural entities. The 3D feature extraction includes, but is not limited to, extracting elevation undulation features and slope change features for terrain entities, extracting crown 3D morphology features and seasonal change 3D visualization features for plant entities, and extracting 3D size features and spatial layout features for structure entities. The parametric extraction of 3D features can be completed using point cloud processing algorithms and the API interface of 3D modeling software.
[0024] Relationships are extracted from the element entity set to obtain an entity relationship set, and semantic enhancement and vectorization encoding are performed on the element entity set based on the three-dimensional feature set. The relationship extraction refers to identifying semantic and spatial associations between element entities, such as the semantic relationship between camphor trees and acidic soil, and the positional relationship between a hexagonal pavilion and the central square. The semantic enhancement refers to embedding the parameter information of the three-dimensional feature set into the semantic description of the element entities, such as fusing the three-dimensional parameters of camphor trees with the semantics of plant entities. The vectorization encoding can utilize an improved Word2Vec model to transform the element entities with fused three-dimensional features into corresponding element entity semantic vectors and three-dimensional feature vectors. By fusing the entity relationship set with the corresponding element entity semantic vectors and three-dimensional feature vectors, a semantic set of garden design elements is obtained.
[0025] It should be further explained that, in the specific implementation process, the specific process of constructing the garden design knowledge graph corresponding to the semantic set of garden design elements based on the improved graph neural network algorithm includes: Optionally, in this embodiment of the application, the similarity of the semantic vectors of the element entities in the semantic set of the garden design elements is calculated to obtain an entity semantic similarity matrix; The cosine similarity algorithm is used to calculate the similarity between the semantic vectors of any two element entities. The similarity calculation formula is as follows: ;in, This represents the similarity score between two entities, with a range of values. A larger value indicates a stronger semantic association between entities; , The first The and the first The semantic vector of each element entity.
[0026] Based on the entity semantic similarity matrix, an improved GAT (Graph Attention Network) is used to cluster the element entities to obtain thematic clusters of garden design elements, specifically: Neighboring nodes are selected based on the entity semantic similarity matrix, and a semantic similarity threshold is set. Traverse the entity semantic similarity matrix and match it with the target entity. similarity All entities Include target entities The initial set of neighborhood nodes is obtained; combined with the entity relationship set of the semantic set of garden design elements, entities with clear semantic or spatial relationships are added to the initial set of neighborhood nodes, resulting in the final set of neighborhood nodes of the improved GAT graph attention neural network. This addresses the problem of insufficient semantic relevance caused by random sampling of neighborhood nodes in traditional GAT graph attention neural networks.
[0027] The improved GAT graph attention neural network introduces a three-dimensional feature attention weight factor. When calculating the attention coefficient of entity nodes, the similarity of three-dimensional features is included in the weight calculation. The modified formula for calculating the attention coefficient is as follows: ;in, , This is the semantic vector of the corresponding element entity; , This is the three-dimensional feature vector of the corresponding element entity; For element entities The set of neighboring nodes; This is a weight vector used to calculate the attention coefficient between entity node pairs, measuring the strength of the association between nodes.
[0028] During the iterative training of the GAT graph attention neural network, the average semantic similarity within each entity cluster is calculated after each iteration. When the growth rate of the average semantic similarity is less than 1% for three consecutive iterations, the clustering process is deemed to have converged, the model training is terminated, and over-clustering is avoided. The converged entity cluster set is then output and denoted as the garden design element theme cluster set. The garden design element theme cluster set includes, but is not limited to, tree plant clusters, water feature structure clusters, and regional cultural symbol clusters, thereby achieving accurate aggregation of semantically closely related element entities.
[0029] Cluster cores are calculated for each theme cluster in the garden design element theme cluster set to obtain the core entity node set of the garden design knowledge graph. Then, each theme cluster in the garden design element theme cluster set is traversed. , ; The total number of theme clusters for garden design elements; extract all element entities within the corresponding theme cluster. element entity semantic vector With three-dimensional feature vectors ( , For thematic clusters (Number of element entities within) The fused feature vector for each entity is generated by feature concatenation and weighted fusion. The calculation formula is as follows: ;in, These are the feature fusion weight coefficients, with values ranging from [value range missing]. .
[0030] Calculate intra-cluster entity weights based on entity semantic similarity, using topic clusters. Each element entity within Based on this, calculate its relationship with all other element entities within the cluster. ( The mean semantic similarity of () is calculated using the following formula: ;in, This represents the similarity value between two entities stored in the entity semantic similarity matrix.
[0031] The mean similarity of all entities within a cluster is normalized to obtain the weight coefficient of each entity within the topic cluster. The calculation formula is as follows: ; It should be noted that the weighting coefficients It reflects the semantic representativeness of an entity within a topic cluster. The larger the value, the closer the relationship between the element entity and other element entities within the cluster, and the more suitable it is as a cluster center candidate.
[0032] According to the fusion feature vector and weighting coefficients Calculate topic clusters The weighted average cluster center vector is calculated using the following formula: ;in, This is a weighted average cluster center vector; it integrates the topic clusters. It uses the semantic and three-dimensional features of all element entities, and assigns higher weights to entities with strong semantic representativeness, which can accurately represent the core features of the entire topic cluster.
[0033] Computational topic clusters The fused feature vector of each element entity within the entity with weighted average cluster center vector The cosine similarity is used to select the element entity with the highest cosine similarity. As this thematic cluster The corresponding core entity nodes; if multiple element entities have the same similarity to the weighted average cluster center vector and all of them are the maximum values, then they are all regarded as core entity nodes.
[0034] The core entity nodes corresponding to all topic clusters are summarized and denoted as the core entity node set; Based on the entity relationship set, corresponding semantic association edges are matched for the core entity node set. Based on the semantic association edges and the core entity node set, a garden design knowledge graph is constructed.
[0035] It should be noted that the semantic association edges are obtained based on the semantic association definition between element entities.
[0036] It should be further explained that, in the specific implementation process, the process of digitally modeling the basic data of the garden design based on digital twin technology to obtain the corresponding digital twin model of the garden site includes: Optionally, in this embodiment of the application, based on digital twin technology, three-dimensional modeling is performed according to the topographic data, plant category attribute data, garden structure parameter data and regional cultural adaptation standard data in the basic data of garden design. Through the three-dimensional data features in the topographic data, plant category attribute data, garden structure parameter data and regional cultural adaptation standard data, the corresponding site topographic digital twin sub-model, plant digital twin sub-model library, structure digital twin sub-model library and cultural symbol digital twin sub-model are obtained. Based on the functional zoning plan of the garden site, the site topography digital twin sub-model, the plant models in the plant digital twin sub-model library, the structure models in the structure digital twin sub-model library, and the cultural symbol digital twin sub-model are spatially combined and assembled to obtain the initial version of the site digital twin model. Collision detection and optimization are performed on the initial version of the site digital twin model to obtain the final site digital twin model; The collision detection refers to detecting spatial positional conflicts between elements in the model, such as overlap between plant models and structure models, or elevation mismatch between structure models and terrain models. The bounding box collision detection algorithm is used to identify the conflicts. For the detected conflicts, the positions are adjusted and the parameters are optimized in accordance with the landscape design specifications to ensure the rationality and feasibility of the model.
[0037] It should be further explained that, in the specific implementation process, the process of associating and mapping the aforementioned garden design knowledge graph with the site's digital twin model to obtain the initial interactive virtual garden scene includes: Optionally, in this embodiment of the application, a mapping rule base for the garden design knowledge graph and the site digital twin model is constructed; Among them, the mapping rules include the one-to-one correspondence rules between core entity nodes and digital models. For example, the mapping rules between the camphor tree element entity in the garden design knowledge graph and the camphor tree 3D model in the plant digital twin sub-model library, and the mapping rules between semantic association edges and digital models. For example, the mapping rules between spatial adjacency relationships in the garden design knowledge graph and the positional adjacency constraints of two element nodes in the digital model, and the mapping rules between attribute information and corresponding digital model parameters.
[0038] Semantic annotation is performed on each sub-model in the site digital twin model to generate a model semantic tag set; Semantic annotation refers to adding semantic tags to each sub-model that are consistent with the core entity nodes in the garden design knowledge graph. For example, adding multi-layer semantic tags of plant-tree-camphor to the camphor tree 3D model, and adding semantic tags of structure-hexagonal pavilion to the hexagonal pavilion 3D model.
[0039] Based on the model semantic tag set and mapping rule base, the core entity nodes in the landscape design knowledge graph are matched with the sub-models in the site digital twin model to obtain the entity-model mapping relationship set; Among them, semantic tag matching adopts a combination of exact matching and fuzzy matching. For entities and models with completely identical semantic tags, exact matching is used; for entities and models with similar semantic tags, fuzzy matching is performed based on the similarity matrix of the corresponding core entity nodes in the landscape design knowledge graph.
[0040] Based on the entity-model mapping relationship set, the semantic association edges and core entity node set in the landscape design knowledge graph are associated with the site digital twin model, thereby realizing the semantic enhancement and constraint of the landscape design knowledge graph on the three-dimensional model.
[0041] The associated site digital twin model is imported into the virtual reality engine to construct an initial interactive garden virtual scene with semantic interaction capabilities; The virtual reality engine includes a lighting system, a weather system, and an interactive control system. The lighting system simulates realistic lighting effects based on the location and seasonal changes of the site. The weather system provides a visual simulation of different weather conditions, such as sunny, rainy, and snowy days. The interactive control system allows users to browse and operate the scene using virtual reality devices.
[0042] It should be noted that the site digital twin model constructed through digital twin technology achieves a realistic restoration of the garden site; by mapping the garden design knowledge graph with the digital twin model, the virtual scene is endowed with semantic-level interactive capabilities, solving the problem that traditional virtual garden scenes can only achieve visual browsing and lack design knowledge support.
[0043] It should be further explained that, in the specific implementation process, the process of embedding interactive logic into the initial interactive garden virtual scene based on the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene, includes: It should be noted that the basic design interaction commands refer to commonly used operation commands in the garden design process, including but not limited to commands for adding elements, deleting elements, moving elements, modifying element parameters, switching scene perspectives, and saving design schemes.
[0044] Optionally, in this embodiment of the application, the obtained basic design interaction instructions are parsed to obtain the instruction type and instruction parameter set, and then an instruction parsing rule base is constructed; Specifically, natural language processing technology is used to parse basic design interaction commands, identify the type of corresponding basic design interaction commands, extract the parameter information of corresponding basic design interaction commands, and build a command parsing rule base to ensure that all basic design interaction commands can be uniformly parsed into standardized command types and parameter information sets; for example, the parameter information for adding plant commands includes, but is not limited to, plant type, planting location, and planting quantity.
[0045] Based on the instruction type and parameter information set in the instruction parsing rule base, a garden design interaction logic rule base is constructed. The interaction logic rule base includes, but is not limited to, element operation logic rules, design constraint logic rules, and knowledge association logic rules. The element operation logic rules define the specific execution flow for adding, deleting, moving, and modifying elements. For example, the operation logic for adding a plant element is: select plant category → determine planting location → detect soil and light conditions at the planting location → generate plant model → add to virtual scene. The design constraint logic rules are constructed based on constraint relationships in the landscape design knowledge graph. For example, the plant planting location must match the soil pH, and the spacing between structures must meet fire safety regulations. The knowledge association logic rules define the association method between operation commands and the landscape design knowledge graph. For example, when a user modifies plant parameters, the system automatically associates the attribute information of the corresponding plant element in the landscape design knowledge graph and provides suggestions on a reasonable range for parameter modification.
[0046] The logical rules in the interactive logic rule base are implanted into the virtual reality engine of the initial interactive garden virtual scene; Specifically, corresponding execution scripts are written for different types of basic design interaction commands. For example, a script for adding plant elements is written to realize the function of calling a specified plant model from the plant sub-model library and generating it at a specified location in the initial interactive garden virtual scene; a script for design constraint detection is written to realize the function of real-time constraint detection and prompting for user operations; and all scripts are integrated with the interaction system of the virtual reality engine to realize the linkage between commands and scenes.
[0047] Functional testing and optimization were performed on the initial interactive virtual garden scene after the interactive logic was implanted to obtain the final interactive virtual garden scene; Functional testing includes instruction response testing, constraint detection testing, and knowledge association testing; script optimization and rule adjustment are carried out for problems found in the testing to ensure the stability and accuracy of virtual scene interaction.
[0048] Example 2: like Figure 2 As shown, an interactive garden design simulation system based on virtual reality is disclosed. The system includes, but is not limited to, a knowledge graph construction module, a digital model construction module, a digital model mapping module, and a basic scene construction module. The knowledge graph construction module is used to perform structured parsing and three-dimensional feature extraction on the acquired basic data of garden design to obtain a semantic set of garden design elements, and to construct a garden design knowledge graph corresponding to the semantic set of garden design elements based on an improved graph neural network algorithm. The digital model building module, based on digital twin technology, performs digital modeling on the basic data of garden design to obtain the corresponding digital twin model of the garden site; The digital model mapping module is used to associate and map the garden design knowledge graph with the site digital twin model to obtain an initial interactive garden virtual scene; The scene construction module is used to obtain basic design interaction instructions, and to implant interactive logic into the initial interactive garden virtual scene according to the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.
[0049] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.
[0050] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0051] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.
[0052] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0053] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.
[0054] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0055] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0056] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. An interactive garden design simulation method based on virtual reality, characterized in that, The method includes: The acquired basic data of garden design is subjected to structured analysis and three-dimensional feature extraction to obtain a semantic set of garden design elements. A garden design knowledge graph corresponding to the semantic set of garden design elements is constructed based on an improved graph neural network algorithm. Based on digital twin technology, a digital model is created from the basic data of garden design to obtain a site digital twin model of the corresponding garden; the garden design knowledge graph is then mapped to the site digital twin model to obtain an initial interactive virtual garden scene. Obtain basic design interaction instructions, and implant interactive logic into the initial interactive garden virtual scene based on the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.
2. The interactive garden design simulation method based on virtual reality according to claim 1, characterized in that, The process of performing structured analysis and three-dimensional feature extraction on the acquired basic data of landscape design to obtain the semantic set of landscape design elements includes: The acquired basic landscape design data is unified in multiple data formats to obtain a standardized basic dataset; the standardized basic dataset is parsed in a hierarchical structure to obtain a hierarchical basic data text set; the hierarchical basic data text set is used for element entity recognition and 3D feature extraction to obtain an element entity set and a 3D feature set. Relationships are extracted from the set of element entities to obtain an entity relationship set. Based on the three-dimensional feature set, the element entity set is semantically enhanced and vectorized, transforming the element entities with integrated three-dimensional features into corresponding element entity semantic vectors and three-dimensional feature vectors. The entity relationship set is then fused with the corresponding element entity semantic vectors and three-dimensional feature vectors to obtain a semantic set of garden design elements.
3. The interactive garden design simulation method based on virtual reality according to claim 2, characterized in that, The process of constructing a garden design knowledge graph corresponding to the semantic set of garden design elements based on an improved graph neural network algorithm includes: The similarity of the semantic vectors of the element entities in the semantic set of the garden design elements is calculated to obtain the entity semantic similarity matrix; Based on the entity semantic similarity matrix, an improved GAT graph attention neural network is used to cluster the element entities to obtain the theme clusters of garden design elements; Cluster cores are calculated for each theme cluster in the set of theme clusters of garden design elements to obtain the core entity node set of garden design knowledge graph. Semantic association edges are matched for the core entity node set according to the entity relationship set. Based on the semantic association edges and the core entity node set, the garden design knowledge graph is constructed.
4. The interactive garden design simulation method based on virtual reality according to claim 3, characterized in that, The process of calculating the cluster centers of each theme cluster in the garden design element theme cluster set to obtain the core entity node set of the garden design knowledge graph includes: Traverse each theme cluster in the collection of garden design element theme clusters; extract the element entity semantic vector and three-dimensional feature vector of all element entities within the corresponding theme cluster; The method of feature concatenation and weighted fusion is used to generate a fused feature vector for each entity; the entity weight within the cluster is calculated based on the semantic similarity of the entity; taking each element entity within the topic cluster as the benchmark, the mean semantic similarity between it and all other element entities within the cluster is calculated. The mean similarity of all entities within a cluster is normalized to obtain the weight coefficient of each entity within the topic cluster; based on the fused feature vector and the weight coefficient, the weighted average cluster center vector of the topic cluster is calculated. Calculate the cosine similarity between the fused feature vector of each element entity within a topic cluster and the weighted average cluster center vector. Select the element entity with the highest cosine similarity as the core entity node corresponding to that topic cluster. If multiple element entities have the same similarity to the weighted average cluster center vector and all of them are the maximum values, then all of them are taken as core entity nodes. Summarize the core entity nodes corresponding to all topic clusters and denote them as the core entity node set.
5. The interactive garden design simulation method based on virtual reality according to claim 4, characterized in that, The process of obtaining a site-specific digital twin model of a garden by digitally modeling basic garden design data based on digital twin technology includes: Based on digital twin technology, three-dimensional modeling is carried out according to the topographic data, plant category attribute data, garden structure parameter data and regional cultural adaptation standard data in the basic data of garden design. In this way, corresponding site topographic digital twin sub-model, plant digital twin sub-model library, structure digital twin sub-model library and cultural symbol digital twin sub-model are obtained. Based on the functional zoning plan of the garden site, the site topography digital twin sub-model, the plant models in the plant digital twin sub-model library, the structure models in the structure digital twin sub-model library, and the cultural symbol digital twin sub-model are spatially combined and assembled to obtain the initial version of the site digital twin model. Collision detection and optimization were performed on the initial version of the site digital twin model to obtain the final site digital twin model.
6. The interactive garden design simulation method based on virtual reality according to claim 5, characterized in that, The process of mapping the landscape design knowledge graph to the site digital twin model to obtain the initial interactive virtual landscape scene includes: Construct a mapping rule base between a knowledge graph of landscape design and a digital twin model of the site; Semantic annotation is performed on each sub-model in the site digital twin model to generate a model semantic tag set; Based on the model semantic tag set and mapping rule base, the core entity nodes in the landscape design knowledge graph are matched with the sub-models in the site digital twin model to obtain the entity-model mapping relationship set; Based on the entity-model mapping relationship set, the semantic association edges and core entity node set in the garden design knowledge graph are associated with the site digital twin model. The associated site digital twin model is then imported into the virtual reality engine to construct an initial interactive garden virtual scene with semantic interaction capabilities.
7. The interactive garden design simulation method based on virtual reality according to claim 6, characterized in that, The process of embedding interactive logic into the initial interactive garden virtual scene based on the parsing results of the basic design interaction instructions to obtain an interactive garden virtual scene includes: The acquired basic design interaction commands are parsed to obtain the command type and command parameter set, and then a command parsing rule base is constructed. Based on the instruction type and parameter information set in the instruction parsing rule base, a garden design interaction logic rule base is constructed. The logical rules in the interactive logic rule base are implanted into the virtual reality engine of the initial interactive garden virtual scene; the initial interactive garden virtual scene after the interactive logic is implanted is functionally tested and optimized to obtain the final interactive garden virtual scene.
8. An interactive garden design simulation system based on virtual reality, implementing the interactive garden design simulation method based on virtual reality as described in any one of claims 1 to 7, characterized in that, include: Knowledge graph construction module, digital model construction module, digital model mapping module, and basic scenario construction module; The knowledge graph construction module is used to perform structured parsing and three-dimensional feature extraction on the acquired basic data of garden design to obtain a semantic set of garden design elements, and to construct a garden design knowledge graph corresponding to the semantic set of garden design elements based on an improved graph neural network algorithm. The digital model building module, based on digital twin technology, performs digital modeling on the basic data of garden design to obtain the corresponding digital twin model of the garden site; The digital model mapping module is used to associate and map the garden design knowledge graph with the site digital twin model to obtain an initial interactive garden virtual scene; The scene construction module is used to obtain basic design interaction instructions, and to implant interactive logic into the initial interactive garden virtual scene according to the parsing results of the basic design interaction instructions, thereby obtaining an interactive garden virtual scene.