Geographic scene parameterization three-dimensional modeling method and system driven by large language model, storage medium and program product

Abstract

The invention discloses a geographic scene parameterization three-dimensional modeling method and system driven by a large language model, a storage medium and a program product, belongs to the field of geographic information science, and solves the problems that a geographic 3D scene reconstruction method is time-consuming in the data acquisition and processing process, generated models and scenes lack semantic information and the like. The method comprises the following steps: constructing a progressive three-dimensional scene modeling knowledge graph of a geographic scene; a reasoning context screening mechanism based on spatial autocorrelation and information entropy is constructed based on the geographic entity space and the progressive three-dimensional scene modeling knowledge graph, and a prompt text used for LLMs reasoning is obtained; constructing an implicit association reasoning rule between entities guided by a thinking chain based on a prompt text of LLMs reasoning, and updating a progressive three-dimensional scene modeling knowledge graph; and performing parameterized geographic scene three-dimensional modeling based on a three-dimensional model generation method of a primitive model and modeling object semantics and the updated progressive three-dimensional scene modeling knowledge graph. The method is used for parameterized geographic scene three-dimensional modeling.

Description

Technical Field

[0001] A large language model-driven geographic scene parameterized three-dimensional modeling method, system, storage medium and program product, used for parameterized geographic scene three-dimensional modeling, belonging to the field of geographic information science. Background Art

[0002] Three-dimensional (3D) geographic scene modeling is an important research direction in the intersection of geographic information science and computer graphics. Different from ordinary 3D scenes, 3D geographic scenes are digital abstractions and expressions of real geographic environments, so they need to maintain spatial, semantic and temporal consistency with the real environment. 3D geographic scenes play a key role in applications such as digital twin cities, smart cities and virtual geographic environments, providing intuitive and interactive digital spatial decision support for urban planning, emergency management, environmental monitoring and navigation services (Biljecki et al., 2015; Dang et al., 2022). However, due to the complexity, diversity and dynamics of the geographic environment, there are still many technical challenges in building and maintaining large-scale realistic 3D geographic scene models (Ghamisi et al., 2019).

[0003] At present, there are two main methods for 3D geographic scene modeling: 3D reconstruction method and parametric modeling. 3D reconstruction method is a technology that directly constructs a 3D model from data collected from the real world (such as images, point clouds). Common technologies include image-based Structure from Motion (SfM) and Multi-View Stereo (MVS), and laser scanning-based LiDAR reconstruction. The main advantage of this method is that it can generate highly realistic and accurate models, especially suitable for capturing complex geographic environment details, with high geometric accuracy and visual realism. However, 3D reconstruction also has some limitations. First, 3D reconstruction is time-consuming in the process of data acquisition and processing, especially for the reconstruction of large-scale scenes. More importantly, the generated models and scenes often lack semantic information. Semantic information is the core element of 3D geographic scenes and GIS. It not only gives geometric models rich attributes and meanings, but also is the key foundation for spatial query, spatial analysis, and intelligent decision-making. Parametric modeling is a method based on the abstraction and generalization of the real world, and generates 3D models by setting parameters and rules. The core of this method is to transform complex geographic entities and phenomena into a series of parameters and rules with semantics. First, by analyzing the data features of the real world, key semantic attributes and relationships are extracted; then, modeling parameters and rules are formulated based on this semantic information. Common techniques include grammar-based modeling and procedural modeling, which essentially encode semantic knowledge into the modeling process. The advantage of this method is that it can quickly generate large-scale complex scenes, can generate multiple variants by adjusting parameters, and the model itself has rich semantic information. However, this method has limitations in accurately restoring scene details.

[0004] The fundamental reason for the above limitations is that parametric modeling attempts to describe complex and diverse real environments through limited, universal parameters and rules, which requires parametric modeling to strike a balance between universality and specificity. For example, extracting modeling information from more data, adding parameters and rules, and manual adjustments can make the three-dimensional scene more detailed, but this may cause parametric modeling to lose its original efficiency advantage. In general, the key to this limitation lies in the balance between efficiency and accuracy. Without changing the basic paradigm of parametric modeling, improving the degree of intelligence is the most effective way to solve this limitation. Intelligent parametric modeling can not only more effectively integrate and analyze the data used for modeling and extract more valuable information. It can also adaptively change the modeling strategy according to the particularity of the modeling object.

[0005] In summary, the 3D geographic scene modeling method in the prior art has the following technical problems:

[0006] 1. The 3D reconstruction method is time-consuming in the process of data acquisition and processing, and the generated models and scenes lack semantic information, which is not conducive to spatial query, spatial analysis, intelligent decision-making, etc.;

[0007] 2. The use of parametric modeling has problems with low intelligence, efficiency or accuracy. Summary of the invention

[0008] The purpose of the present invention is to provide a large language model driven parameterized three-dimensional modeling method, system, storage medium and program product for geographic scenes, so as to solve the problems that the 3D reconstruction method is time-consuming in the process of data acquisition and processing, and the generated models and scenes lack semantic information, which is not conducive to spatial query, spatial analysis, intelligent decision-making, etc.; and the use of parameterized modeling has the problem of low intelligence, efficiency or accuracy.

[0009] In order to achieve the above object, the technical solution adopted by the present invention is:

[0010] A large language model-driven parameterized three-dimensional modeling method for geographic scenes comprises the following steps:

[0011] Step 1: Construct a knowledge graph of progressive 3D scene modeling of geographic scenes;

[0012] Step 2: Based on the geographic entity space and progressive 3D scene modeling knowledge graph, a reasoning context screening mechanism based on spatial autocorrelation and information entropy is constructed to obtain prompt text for LLMs reasoning;

[0013] Step 3: Based on the prompt text of LLMs reasoning, the implicit association reasoning rules between entities guided by the thinking chain are constructed to update the progressive 3D scene modeling knowledge graph;

[0014] Step 4: Perform parametric geographic scene 3D modeling based on the 3D model generation method based on the primitive model and modeling object semantics and the updated progressive 3D scene modeling knowledge graph.

[0015] Further, the specific steps of step 1 are:

[0016] The scene is expressed and abstracted from the three dimensions of geography, geometry and rendering to realize the construction of a progressive knowledge graph from geographic semantics to rendering data; that is, through the semantic mapping mechanism, different granularity semantics are associated and converted from the three dimensions of geography, geometry and rendering of the geographic scene, and finally an associated multi-dimensional semantic graph is obtained, that is, a progressive three-dimensional scene modeling knowledge graph is obtained, wherein the associated multi-dimensional semantic graph includes an explicitly associated geographic semantic knowledge graph, a geometric semantic knowledge graph and a rendering semantic knowledge graph. The geographic knowledge graph is used to abstract and conceptualize geographic information in the real world, the geometric semantic knowledge graph is used to further refine the geographic semantics to support the specific geometric modeling process, and the rendering semantic knowledge graph is used to further abstract on the basis of geometric semantics to support high-realistic visual rendering. Explicit association refers to the direct establishment of deterministic mappings between semantics of different dimensions through the entity connection of the knowledge graph.

[0017] Further, the specific steps of step 2 are:

[0018] Taking the spatial location of the geographic entity to which the inferred object belongs as the center, the Moran's I index is used to calculate the spatial autocorrelation under the spatial distance threshold of the geographic entity, and the distance with the largest absolute value of the spatial autocorrelation is selected as the final spatial reasoning range. The formula of the Moran's I index is:

[0019]

[0020]

[0021] in, Represents the Moran's I index value, n means that the spatial distance from the geographic entity of the inferred object is less than D k The total number of geographic entity spatial objects, D k represents the k-th geographic entity spatial distance threshold, x i and x j are the attribute values ​​of the inferred objects i and j respectively, is the average value of the attribute, w ij is the spatial weight of the geographic entity between the inferred objects i and j, d is the interval distance between the geographic entity spaces, is a non-zero positive integer;

[0022] Based on the obtained spatial reasoning scope, the entity in the knowledge graph to which the inferred object belongs in the progressive 3D scene modeling knowledge graph is taken as the center, and sub-networks are formed with different hop numbers. The number of relationships and attributes in the network is used as the information richness. Then, the information entropy is used to calculate the information richness of each sub-network. When the information entropy reaches the threshold H sWhen the text description of the parameters of all entities within the hop number and the relationship between entities is used as the prompt text of LLMs reasoning, it is represented as a local subgraph G in the form of a graph. q , the formula for calculating the information entropy in different hops with a certain entity as the center is:

[0023]

[0024] Among them, H k (v) represents the information entropy of sub-networks with different hop numbers when a certain entity is the center, N h (v) represents the set of all nodes with a hop count of h when the entity in the knowledge graph to which the object to be inferred belongs is the central node v. A(N h (v)) represents the attribute set in the sub-network with a central node of v and a hop count of h. h (v) represents the set of relations in a subnetwork with a central node of v and a hop count of h. i ) and P(r i ) are attributes a i and the relationship i In the set A(N h (v)) and R(N h The probability in (v).

[0025] Further, the specific steps of step 3 are:

[0026] Step 3.1: Sub-graph G q Context description text C converted into natural language form q , using Embedding technology and external knowledge base K to describe the context text C q Convert it into a vector, and then calculate the distance between the vectors by cosine similarity, and use the context description text corresponding to the vector with the closest distance as the external knowledge base K q ,Finally, according to the given query entity e q and query relation r q Retrieve external knowledge base K q Get the target context description text to construct a thought chain prompt P, where the thought chain prompt P includes query description, knowledge description and reasoning step prompts, where the external knowledge base K q is a knowledge base of textual descriptions of modeling scenarios, including query entities e q and query relation r q and the corresponding context description text C q ;

[0027] Step 3.2: Input the constructed thought chain prompt P into the pre-trained large language model for implicit association reasoning to generate the answer relation ra and the new entity e a , the new entity e a and new relationships a Add it to the progressive 3D scene modeling knowledge graph G, and update the representation of entities and relationships at the same time, so as to obtain the updated progressive 3D scene modeling knowledge graph.

[0028] A computer system includes a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of a large language model-driven geographic scene parameterized three-dimensional modeling method.

[0029] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a large language model-driven geographic scene parameterized three-dimensional modeling method.

[0030] A computer program product includes a computer program, which, when executed by a processor, implements the steps of a large language model-driven geographic scene parameterized three-dimensional modeling method.

[0031] Compared with the prior art, the advantages of the present invention are:

[0032] 1. The present invention proposes a progressive 3D scene modeling knowledge graph, which integrates the semantic information of geography, geometry and rendering to achieve a unified semantic expression from abstract to concrete. This multi-dimensional semantic fusion method improves the semantic richness and consistency of 3D geographic scene modeling, making the 3D modeled scene more realistic to the real environment.

[0033] Second, the present invention designs a reasoning context screening mechanism based on spatial autocorrelation and information entropy, so that the original data submitted to the large language model is compressed to 128k, allowing the large language model to implicitly associate reasoning in a larger knowledge graph;

[0034] 3. This invention introduces the large language model reasoning based on thought chain prompts, combined with the constructed knowledge graph, and improves the reasoning accuracy of the Qwen2-75b large language model in the modeling knowledge graph from 82% to 94%;

[0035] 4. The three-dimensional model generation method based on primitive models and modeling object semantics proposed in the present invention combines the reasoning results of the large language model with the parametric modeling technology, saving the time of manually constructing the knowledge graph and the 3D model modeling time in traditional modeling, greatly improving the degree of intelligence and modeling efficiency, and the generated models and scenes have semantic information, which is conducive to spatial query, spatial analysis, intelligent decision-making, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 Schematic diagram of three knowledge graphs and relationship definitions in the present invention;

[0038] Figure 2 A schematic diagram of the screening of context entities and subgraphs in the present invention;

[0039] Figure 3 A schematic diagram of a Prompt template for implicit relationship reasoning in the present invention;

[0040] Figure 4 A schematic diagram of a Prompt template for new entity reasoning in the present invention;

[0041] Figure 5 This is an example diagram of the primitive model in the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Constructing a knowledge graph of progressive 3D scene modeling for geographic scenes:

[0044] Three-dimensional geographic scene modeling is the process of abstracting the geographical environment of the real world into a digital model that can be recognized and rendered by computers. In order to realize the intelligent modeling of three-dimensional geographic scenes, a knowledge graph construction of progressive multi-dimensional semantic fusion is proposed. The core is to formally express and abstract the scene from the three dimensions of geographic semantics, geometric semantics and rendering semantics, and associate and transform different granularity semantics from the three dimensions of geography, geometry and rendering of the geographic scene through the semantic mapping mechanism, and finally obtain the associated multi-dimensional semantic graph, that is, the progressive three-dimensional scene modeling knowledge graph. Geographic semantics is used to abstract and conceptualize geographic information in the real world to form a geographic semantic knowledge graph. Then, geometric semantics further refines geographic semantics to generate a geometric semantic knowledge graph to support the specific geometric modeling process. Finally, rendering semantics is further abstracted on the basis of geometric semantics to form a rendering semantic knowledge graph to support high-realistic visualization rendering.

[0045] The geographic semantic knowledge graph focuses on the ontological definition and instantiation description of geographic entity types, attributes, and spatiotemporal relationships, forming a structured semantic network that reflects the geographic world. For example, in the geographic semantic knowledge graph, a tree can be represented as a geographic entity, and its attributes include tree species, height, and crown width. The geometric semantic knowledge graph focuses on the characterization of the scene's geometric elements, structural relationships, and constraint rules, forming a semantic representation that reflects the scene's geometric features. In the geometric semantic knowledge graph, the tree can be represented as a geometric model, whose geometric elements include the trunk (cylinder) and the crown (sphere or cone), the structural relationship is that the trunk supports the crown, and the constraint rule is that the crown height does not exceed the total height of the tree. The rendering semantic knowledge graph focuses on the semantic organization of the scene's visual attributes such as material, texture, and lighting, forming a parametric representation that supports high-realistic rendering. In the rendering semantic graph, the tree's geometric model can be given the material of the bark, the texture of the leaves, and the lighting parameters can be set to achieve realistic rendering effects.

[0046] These three semantic knowledge graphs are relatively independent and logically related. From geographic semantics to geometric semantics to rendering semantics, they reflect the hierarchical expression from abstract to concrete, from concept to instance. At the same time, based on the semantic mapping mechanism, any semantic knowledge graph can establish associations with other graphs to achieve cross-dimensional semantic conversion and reasoning. The associations between multi-dimensional semantic graphs include explicit associations and implicit associations. Explicit associations directly establish deterministic mappings between semantics of different dimensions through the entity connection of the knowledge graph, such as the one-to-one correspondence between geographic entities and geometric models. Implicit associations are potential uncertain associations between semantics of different dimensions that may be established by intelligent reasoning mechanisms such as rule reasoning, path reasoning, or the large language model used in this case, such as the dependency between geographic attributes and rendering parameters. Through the combination of explicit and implicit associations, multi-dimensional semantic graphs can dynamically evolve together to continuously enrich and improve the semantic representation of scene models. For example, given a geographic entity and its attributes, its geometric model can be directly generated through explicit associations, and appropriate rendering parameters can be inferred through implicit associations. Conversely, given a geometric model, its geographic entity type can be determined through explicit associations, and its geographic attributes can be inferred through implicit associations. The interaction and collaboration between multi-dimensional semantic graphs enhance the semantic consistency and completeness of the scene model.

[0047] It is ideal if the three semantic knowledge graphs can be converted to each other. Considering that the final visual effect is determined by the rendering semantics, the rendering semantics need to be strictly set according to different 3D engines and cannot be customized by users or developers. Each entity, attribute, and relationship should have a corresponding implementation method in the 3D engine (the present invention provides a new implementation method). In contrast, the geographic semantic knowledge graph can be as rich and diverse as possible and is not restricted. A richer and larger geographic semantic knowledge graph helps LLMs capture more detailed details, which makes the modeling results more accurate. The geometric semantic knowledge graph is in between the two. The geometric semantic knowledge graph is a feature extraction of the geographic knowledge graph, and the rendering semantic knowledge graph is an implementation of the geometric semantic knowledge graph. This also means that the semantic knowledge graph is independent of the specific 3D engine, which facilitates the implementation of the method on different platforms and in different environments. The entity and relationship definitions of the three knowledge graphs are as follows: Figure 1 shown.

[0048] In addition, for explicitly defined relationships, corresponding mapping methods need to be provided during the modeling process. For example, if the green attribute of grass in geographic semantics is connected to the ARGB value in rendering semantics through explicit definition, corresponding tools or methods need to be provided. The reasoning of implicit relationships is based on the LLMs reasoning prompt text to build a chain of thinking guided by the implicit association reasoning rules between entities and entities to update the progressive 3D scene modeling knowledge graph. Therefore, the initial construction of the progressive 3D scene modeling knowledge graph for geographic scenes only has explicit associations, and implicit associations are added through subsequent mining.

[0049] Based on the geographic entity space and progressive 3D scene modeling knowledge graph, a reasoning context screening mechanism based on spatial autocorrelation and information entropy is constructed to obtain the prompt text for LLMs reasoning:

[0050] When using large language models (LLMs) for knowledge reasoning, it is crucial to choose appropriate contextual information. The quality and quantity of contextual information directly affect the accuracy and efficiency of reasoning. Too much information will lead to redundancy and interference, making it difficult for LLMs to grasp the key points; insufficient information may lead to inaccurate or incomplete reasoning results. Therefore, a method is needed to balance the breadth and depth of contextual information to optimize the reasoning performance of LLMs. Considering that the inferred objects have corresponding instances in the knowledge graph and corresponding objects in the geographic entity space, this case evaluates whether other objects and their relationships need to be included in the reasoning context from two dimensions: geographic entity space and three-dimensional scene modeling knowledge graph space.

[0051] Spatial autocorrelation refers to the correlation between the similarity of objects in space in a certain attribute and their spatial position. Spatial autocorrelation follows the "first law of distance" (Tobler's first law of geography), that is, the closer the objects are in space, the more similar their attributes are; conversely, the farther the objects are in space, the greater the difference in their attributes. Spatial autocorrelation is used to measure the degree of correlation between the object being inferred and its surrounding objects in the selection of reasoning context. When modeling three-dimensional geographic scenes, the size range of the space is uncertain. In order to select the most appropriate objects in the spatial dimension of geographic entities to be included in the reasoning basis, the spatial autocorrelation under different spatial distance thresholds is calculated with the object being inferred (such as the RGB value of the leaf color of a tree) as the center, and the distance with the largest absolute value of the spatial autocorrelation is selected as the final spatial reasoning range.

[0052] Moran's index is used to quantify spatial autocorrelation. Among them, the definition formula of Moran'sl index is:

[0053]

[0054] in, represents the Moran's l index value, n represents the distance from the geographic entity of the inference object to the object less than D k The total number of geographic entity spatial objects, D k represents the k-th geographic entity spatial distance threshold, x i and x j are the attribute values ​​of the inferred objects i and j respectively, is the average value of the attribute value, wij is the spatial weight of the geographic entity between the inferred objects i and j, d is the interval distance between the geographic entity spaces, is a non-zero positive integer.

[0055] The value range of Moran'sl index is [-1, 1], corresponding to complete negative correlation, no correlation and complete positive correlation respectively. Considering that both negative correlation and positive correlation may be helpful for LLMs reasoning, we choose the D corresponding to the Moran'sl index with the largest absolute value. k As the best. By calculating the Moran'sl index at different spatial distances, the spatial autocorrelation between the inferred object and the objects in different ranges around it can be quantified, but this is only a geographical spatial correlation, and these screened objects also correspond to other related objects or attributes in the knowledge graph, and these objects and attributes may also affect the inference object. In the network structure of the knowledge graph, the concept of "jump" is also applicable. The number of hops in the knowledge graph refers to the number of edges required to reach other entities from the central entity. Through different hops, starting from a central entity, subnetworks of different sizes can be constructed. This gradual expansion process will include more and more entities and relationships, thus forming neighborhoods of different ranges. The information content of networks composed of different neighborhoods is different. Considering that the information content of entities in the knowledge graph is different, in order to balance this difference, a subnetwork is formed with a certain entity in the knowledge graph as the center and different hops, and the number of relationships and attributes in the network is used as the information richness. Information entropy is used to calculate the information richness of each subnetwork. Information entropy is an indicator of information uncertainty or information volume, reflecting the degree of disorder or uncertainty of a random variable. As Figure 2 As shown, taking a certain entity as the center, the information entropy H of the sub-network composed of different jumps is calculated. k (v), when the information entropy reaches the threshold H s When the text description of the parameters of all entities within the hop number and the relationship between entities is used as the prompt text of LLMs reasoning, it is represented as a local subgraph G in the form of a graph. q The formula for calculating the information entropy of a certain entity within different hops is:.

[0056]

[0057] Among them, H k (v) represents the information entropy of sub-networks with different hop numbers when a certain entity is the center, N h (v) represents the set of all nodes with a hop count of h when the entity in the knowledge graph to which the object to be inferred belongs is the central node v. A(N h (v)) represents the attribute set in the sub-network with a central node of v and a hop count of h. h (v) represents the set of relations in a subnetwork with a central node of v and a hop count of h. i ) and P(r i ) are attributes a i and the relationship i In the set A(N h (v)) and R(N h (v) The complete context screening process is as follows Figure 2 shown.

[0058] D k The value of determines the sampling accuracy of the spatial distance, D k The smaller the value, the more reasonable the selected spatial object is, but too large D k The value is the opposite. s If the value is too large, the reasonable context advantage will be small, while if it is too small, less useful information will be selected. Therefore, the setting of these two values ​​depends on the needs in practice and needs to be adjusted according to parameters such as the density of geographic objects, spatial scale, and the number of knowledge graph entities.

[0059] Based on the prompt text of LLMs reasoning, the chain of thought guides the entities and the implicit association reasoning rules between entities to update the progressive three-dimensional scene modeling knowledge graph: Chain-of-Thought (CoT) is an emerging large language model reasoning paradigm that aims to improve the interpretability and accuracy of reasoning by decomposing complex reasoning tasks into a series of intermediate steps and guiding the model to perform step-by-step reasoning. Unlike traditional end-to-end reasoning methods, CoT reasoning introduces a structured reasoning path, simulates the step-by-step thinking process of humans, and enables the model to use existing knowledge more effectively for reasoning. In this case, the content of reasoning includes implicit relational reasoning and new entity reasoning. Implicit relational reasoning is to infer potential relationships based on the entities and relationships that already exist in the knowledge graph, while new entity reasoning is to insert new entities into the knowledge graph based on the external knowledge base. Specifically:

[0060] The local subgraph G q Context description text C converted into natural language formq , using Embedding technology and external knowledge base K to describe the context text C q Convert it into a vector, and then calculate the distance between the vectors by cosine similarity, and use the context description text corresponding to the vector with the closest distance as the external knowledge base K q ,Finally, according to the given query entity e q and query relation r q Retrieve external knowledge base K q The target context description text is obtained to construct a thinking chain prompt P, wherein the thinking chain prompt P includes a query description, a knowledge description and a reasoning step prompt. For example, when reasoning about the color of tree leaves, the query description is: "Please reason about the color of tree leaves in winter." Knowledge description: "Leaves contain chlorophyll and can carry out photosynthesis. In autumn, the temperature drops, the sunlight decreases, and the trees stop producing chlorophyll." Reasoning step prompts: Step 1: Analyze the main influencing factors of leaf color; Step 2: Consider the color of leaves in winter and the reasons; Step 3: Analyze the principle of leaf color change in winter; Step 4: Infer the color difference between deciduous trees and evergreen trees in winter; Step 5: Summarize the RGB values ​​of winter colors.

[0061] Among them, the external knowledge base K q is a knowledge base of textual descriptions of modeling scenarios, including query entities e q and query relation r q and the corresponding context description text C q ;

[0062] Input the constructed thought chain prompt P into the pre-trained large language model for implicit association reasoning to generate the answer relation r a and the new entity e a , the new entity e a and new relationships a Add it to the progressive 3D scene modeling knowledge graph G, and update the representation of entities and relationships at the same time, so as to obtain the updated progressive 3D scene modeling knowledge graph.

[0063] The new knowledge graph is based on the external knowledge base K multi-step guidance pre-trained large language model LLM to infer the new entity e a The names, attributes, and explicit relationships of a The implicit relationship of a . And insert it into the constructed knowledge graph of progressive 3D scene modeling of geographic scenes, such as Figure 4 .

[0064] Parametric geographic scene 3D modeling based on the 3D model generation method based on primitive models and modeling object semantics and the updated progressive 3D scene modeling knowledge graph:

[0065] All 3D models in this case are implemented by modifying the properties of primitive models (cubes, irregular geometric shapes, roads, trees). A simple example is Figure 5 shown.

[0066] The attributes of primitive models include: position, size, color, texture, etc. These attributes are also mapped to the geometric semantic knowledge graph, and can also be associated with other nodes or attributes through explicit or implicit relationships. Some of the primitive models are complete three-dimensional models imported from outside, such as models composed of multiple components. The parameters of different components can be controlled by adding appropriate modification functions to the model. Some primitive models do not rely on external modeling software and are directly generated by control functions, such as two-way roads and irregular buildings. A series of three-dimensional points are input into the two-way road, and the road is generated according to attributes such as width, length, and color. Irregular buildings also input a series of three-dimensional points as the shape of the building, and then the appearance of the three-dimensional building is constructed through floor height, number of floors, etc.

[0067] This case proposes a parametric 3D modeling method driven by a large language model for 3D geographic scene modeling. First, features are extracted from the perspectives of geographic entities, geometric information, and rendering engines, and three knowledge graphs are designed. A progressive 3D scene modeling knowledge graph expression from geography to rendering is proposed. The semantics and parameters of modeling objects are managed through the progressive 3D scene modeling knowledge graph. Then, a large language model based on Chain-of-Thought (CoT) prompts is introduced to perform common sense and multi-step reasoning on the knowledge graph. In order to avoid the problem of loss of reasoning focus due to the context restriction of the large language model, the reasoning context screening mechanism based on spatial autocorrelation and information entropy is used to constrain the context of the reasoning content. Finally, the method of parameter-controlled primitive model (i.e., the 3D model generation method referring to the semantics of the modeling object) is adopted to map the spatial knowledge obtained by reasoning into the geometry, texture, and semantic information of the 3D geographic scene model (i.e., the updated progressive 3D scene modeling knowledge graph), realizing the automatic construction from knowledge to model. . Compared with traditional methods, this case significantly improves the intelligence and efficiency of the modeling process, reduces manual intervention, and has advantages in semantic completeness and realism.

Claims

1. A parameterized 3D modeling method for geographic scenes driven by a large language model, characterized in that: The steps include: Step 1: Construct a knowledge graph of progressive 3D scene modeling of geographic scenes; Step 2: Based on the geographic entity space and progressive 3D scene modeling knowledge graph, a reasoning context screening mechanism based on spatial autocorrelation and information entropy is constructed to obtain prompt text for LLMs reasoning; Step 3: Based on the prompt text of LLMs reasoning, the implicit association reasoning rules between entities guided by the thinking chain are constructed to update the progressive 3D scene modeling knowledge graph; Step 4: Perform parametric geographic scene 3D modeling based on the 3D model generation method based on the primitive model and modeling object semantics and the updated progressive 3D scene modeling knowledge graph.

2. The method for parameterized 3D modeling of geographic scenes driven by a large language model according to claim 1, characterized in that: The specific steps of step 1 are: The scene is expressed and abstracted from the three dimensions of geography, geometry and rendering to realize the construction of a progressive knowledge graph from geographic semantics to rendering data; that is, through the semantic mapping mechanism, different granularity semantics are associated and converted from the three dimensions of geography, geometry and rendering of the geographic scene, and finally an associated multi-dimensional semantic graph is obtained, that is, a progressive three-dimensional scene modeling knowledge graph is obtained, wherein the associated multi-dimensional semantic graph includes an explicitly associated geographic semantic knowledge graph, a geometric semantic knowledge graph and a rendering semantic knowledge graph. The geographic knowledge graph is used to abstract and conceptualize geographic information in the real world, the geometric semantic knowledge graph is used to further refine the geographic semantics to support the specific geometric modeling process, and the rendering semantic knowledge graph is used to further abstract on the basis of geometric semantics to support high-realistic visual rendering. Explicit association refers to the direct establishment of deterministic mappings between semantics of different dimensions through the entity connection of the knowledge graph.

3. A large language model driven parametric outdoor 3D modeling method according to claim 1 or 2, characterized in that: The specific steps of step 2 are: Taking the spatial location of the geographic entity to which the inferred object belongs as the center, the Moran's I index is used to calculate the spatial autocorrelation under the spatial distance threshold of the geographic entity, and the distance with the largest absolute value of the spatial autocorrelation is selected as the final spatial reasoning range. The formula of the Moran's I index is: in, Represents the Moran's I index value, n means that the spatial distance from the geographic entity of the inference object is less than D k The total number of geographic entity spatial objects, D k represents the k-th geographic entity spatial distance threshold, x i and x j are the attribute values ​​of the inferred objects i and j, is the average value of the attribute, w ij is the spatial weight of the geographic entity between the inferred objects i and j, d is the interval distance between the geographic entity spaces, is a non-zero positive integer; Based on the obtained spatial reasoning scope, the entity in the knowledge graph to which the inferred object belongs in the progressive 3D scene modeling knowledge graph is taken as the center, and sub-networks are formed with different hop numbers. The number of relationships and attributes in the network is used as the information richness. Then, the information entropy is used to calculate the information richness of each sub-network. When the information entropy reaches the threshold H s When the text description of the parameters of all entities within the hop number and the relationship between entities is used as the prompt text of LLMs reasoning, it is represented as a local subgraph G in the form of a graph. q , the formula for calculating the information entropy in different hops with a certain entity as the center is: Among them, H k (v) represents the information entropy of sub-networks with different hop numbers when a certain entity is the center, N h (v) represents the set of all nodes with a hop count of h when the entity in the knowledge graph to which the object to be inferred belongs is the central node v. A(N h (v)) represents the attribute set in the sub-network with a central node of v and a hop count of h. h (v) represents the set of relations in a subnetwork with a central node of v and a hop count of h. i ) and P(r i ) are attributes a i and the relationship i In the set A(N h (v)) and R(N h The probability in (v).

4. The large language model driven parameterized outdoor 3D modeling method according to claim 3, characterized in that: The specific steps of step 3 are: Step 3.1: Local subgraph C q Context description text C converted into natural language form q , using Embedding technology and external knowledge base K to describe the context text C q Convert it into a vector, and then calculate the distance between the vectors by cosine similarity, and use the context description text corresponding to the vector with the closest distance as the external knowledge base K q ,Finally, according to the given query entity e q and query relation r q Retrieve external knowledge base K q Get the target context description text to construct a thought chain prompt P, where the thought chain prompt P includes query description, knowledge description and reasoning step prompts, where the external knowledge base K q is a knowledge base of textual descriptions of modeling scenarios, including query entities e q and query relation r q and the corresponding context description text C q ; Step 3.2: Input the constructed thought chain prompt P into the pre-trained large language model for implicit association reasoning to generate the answer relation r a and the new entity e a , the new entity e a and new relationships a Add it to the progressive 3D scene modeling knowledge graph G, and update the representation of entities and relationships at the same time, so as to obtain the updated progressive 3D scene modeling knowledge graph.

5. A computer system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer program product comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images