Three-dimensional natural scene construction method and system, terminal and storage medium

Abstract

The application discloses a three-dimensional natural scene construction method and system, a terminal and a storage medium, relates to the technical field of geographic scene three-dimensional modeling, and comprises the following steps: acquiring input images of a target natural scene and performing semantic segmentation to obtain macro geographic entities, including at least one of preset entities; acquiring position information and geometric information of the macro geographic entities; acquiring a pre-constructed entity detail association table, determining target detail monomers corresponding to the macro geographic entities based on the table, acquiring density proportion ranges corresponding to the target detail monomers, and determining density proportion parameters corresponding to the target detail monomers according to the density proportion ranges; determining a detail monomer position point set of the target detail monomers according to the density proportion parameters and the position information and the geometric information of the macro geographic entities to which the target detail monomers belong; and constructing a corresponding three-dimensional model according to the macro geographic entities, the position information of the macro geographic entities, the target detail monomers and the detail monomer position point set. In this way, the authenticity of the generated three-dimensional natural scene can be improved.

Description

Technical Field

[0001] This application relates to the field of 3D modeling technology for geographic scenes, and in particular to a method, system, terminal and storage medium for constructing 3D natural scenes. Background Technology

[0002] With the development of 3D Geographic Information Systems (3D GIS) and computer graphics, the demand for high-fidelity representation of natural environments in digital space is increasing. The quality of natural scene modeling directly affects the immersion and credibility of virtual scenes. Currently, the modeling of artificial elements (such as buildings) is relatively mature, while the realistic representation of natural geographic elements such as trees, grasslands, and water bodies is insufficient.

[0003] In existing technologies, when constructing 3D natural scenes, macroscopic individual geometric structures and appearances of natural geographical elements such as trees are typically generated directly using procedural methods. The problem with these existing technologies is the lack of reasonable modeling methods for the microscopic details of natural geographical elements, which hinders the improvement of the realism of the generated 3D natural scenes.

[0004] Therefore, the relevant technologies still need to be improved and developed. Summary of the Invention

[0005] The main purpose of this application is to provide a method, system, terminal and storage medium for constructing three-dimensional natural scenes. It aims to solve the technical problem in the related technology that when constructing three-dimensional natural scenes, for natural geographical elements such as trees, the macroscopic individual geometric structure and appearance of the natural geographical elements are usually generated directly based on procedural methods, while there is a lack of reasonable modeling methods for the microscopic details of natural geographical elements, which is not conducive to improving the realism of the generated three-dimensional natural scenes.

[0006] To achieve the above objectives, the first aspect of this application provides a method for constructing a three-dimensional natural scene, wherein the method includes: Obtain the input image corresponding to the target natural scene; Semantic segmentation is performed on the above input image to obtain the macro-geographic entities corresponding to the above target natural scene, and the location and geometric information of the above macro-geographic entities are obtained; wherein, the macro-geographic entities corresponding to the above target natural scene include at least one of a variety of preset entities; Obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detail units corresponding to each of the above-mentioned preset entities, and the density ratio range corresponding to each of the above-mentioned detail units; search and determine the target detail units corresponding to each of the above-mentioned macro-geographic entities from the entity detail association table, and obtain the density ratio range corresponding to each of the above-mentioned target detail units; determine the density ratio parameter corresponding to the above-mentioned target detail units based on the density ratio range. Based on the density ratio parameters corresponding to each of the above target detail units, as well as the location and geometric information of the macro-geographic entities to which each of the above target detail units belongs, determine the set of detail unit location points corresponding to each of the above target detail units. Based on the aforementioned macro-geographic entities, their location information, the target details, and their location point sets, a 3D model corresponding to the target's natural scene is constructed.

[0007] Optionally, the aforementioned preset entities include lawns, water bodies, and ground cover layers that are planar entities, roads that are linear entities, and trees that are point entities. The above-mentioned semantic segmentation of the input image is performed to obtain the macro-geographic entities corresponding to the target natural scene, and the location and geometric information of the macro-geographic entities are obtained, including: Using a pre-defined semantic segmentation model, pixel-level semantic segmentation is performed on the input image to obtain at least one macro-geographic entity corresponding to the target natural scene. For macro-geographic entities that belong to the above-mentioned areal entities, the regional boundary coordinates of the macro-geographic entities are used as location information, and the regional area of ​​the macro-geographic entities is used as geometric information. For macro-geographic entities that belong to the aforementioned linear entities, the centerline coordinates of the aforementioned macro-geographic entities are used as location information, and the length and / or width of the aforementioned macro-geographic entities are used as geometric information. For macro-geographic entities that belong to the aforementioned point-like entities, the centroid coordinates of the macro-geographic entity are used as location information, and the minimum circumscribed circle radius of the macro-geographic entity is used as geometric information.

[0008] Optionally, determining the density ratio parameter corresponding to the target detail unit based on the aforementioned density ratio range includes: For each of the aforementioned target detail units, a random value is selected from the density ratio range corresponding to the aforementioned target detail unit, and this value is used as the density ratio parameter corresponding to the aforementioned target detail unit.

[0009] Optionally, determining the set of location points for each of the aforementioned target detail units based on the density ratio parameters corresponding to each of the aforementioned target detail units, and the location and geometric information of the macroscopic geographic entities to which each of the aforementioned target detail units belongs, includes: For each of the aforementioned target detail units, the detail distribution range is determined based on the location information of the macro-geographic entity to which the target detail unit belongs. The number of detail distributions is determined based on the geometric information of the macro-geographic entity to which the target detail unit belongs and the density ratio parameter corresponding to the target detail unit. Based on the number of detail distributions and the detail distribution range, a set of detail unit location points corresponding to the target detail unit is generated.

[0010] Optionally, the above-mentioned construction of a 3D model corresponding to the natural scene of the target based on the macro-geographic entity, the location information of the macro-geographic entity, the target detail unit, and the location point set of the detail unit includes: Using the aforementioned macro-geographic entities as parent nodes and the target detail units corresponding to the aforementioned macro-geographic entities as child nodes corresponding to the aforementioned parent nodes, a semantic tree is constructed. The location information of the aforementioned macro-geographic entities is used as the attribute information of the aforementioned parent nodes, and the location point set of the aforementioned detail units is used as the attribute information of the aforementioned child nodes. Based on the semantic tree described above, construct a 3D model corresponding to the target natural scene.

[0011] Optionally, the above-mentioned construction of the 3D model corresponding to the target natural scene based on the semantic tree includes: For each of the above parent nodes, select a target entity base model from at least one preset entity base model associated with the macro-geographic entity corresponding to the above parent node, and perform 3D rendering based on the target entity base model and the attribute information corresponding to the above parent node. For each of the above-mentioned sub-nodes, a target detail base model is determined for each point in the set of position points of the detail unit corresponding to the above-mentioned sub-node from at least one preset detail base model associated with the target detail unit corresponding to the above-mentioned sub-node. 3D rendering is performed based on the target detail base model and the attribute information corresponding to the above-mentioned sub-node. After completing the 3D rendering of the aforementioned parent node and child node, a 3D model representing the aforementioned target natural scene is obtained.

[0012] A second aspect of this application provides a three-dimensional natural scene construction system, wherein the system includes: The image acquisition module is used to acquire the input image corresponding to the target natural scene; The semantic segmentation module is used to perform semantic segmentation on the input image to obtain the macro-geographic entities corresponding to the target natural scene, and to acquire the location and geometric information of the macro-geographic entities; wherein, the macro-geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; The detailed unit determination module is used to obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detailed units corresponding to each of the above-mentioned preset entities, and the density ratio range corresponding to each of the above-mentioned detailed units; from the entity detail association table, the module searches and determines the target detailed units corresponding to each of the above-mentioned macro-geographic entities, and obtains the density ratio range corresponding to each of the above-mentioned target detailed units; and determines the density ratio parameter corresponding to the above-mentioned target detailed units based on the density ratio range. The detailed location determination module is used to determine the set of detailed unit location points corresponding to each of the above-mentioned target detailed units based on the density ratio parameters corresponding to each of the above-mentioned target detailed units, as well as the location information and geometric information of the macro-geographic entities to which each of the above-mentioned target detailed units belongs. The scene construction module is used to construct a 3D model corresponding to the natural scene of the target based on the macro-geographic entities, the location information of the macro-geographic entities, the target detail units, and the location point set of the detail units.

[0013] A third aspect of this application provides a terminal, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of any of the three-dimensional natural scene construction methods described above.

[0014] A fourth aspect of this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the above-described three-dimensional natural scene construction methods.

[0015] As can be seen from the above, the present application provides a method for constructing a three-dimensional natural scene. The method includes: acquiring an input image corresponding to a target natural scene; performing semantic segmentation on the input image to obtain macroscopic geographic entities corresponding to the target natural scene, and acquiring the location information and geometric information of the macroscopic geographic entities; wherein the macroscopic geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; acquiring a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detail units corresponding to each of the preset entities, and the density ratio range corresponding to each of the detail units; searching and determining the target detail units corresponding to each of the macroscopic geographic entities from the entity detail association table, and acquiring the density ratio range corresponding to each of the target detail units; determining the density ratio parameter corresponding to the target detail unit based on the density ratio range; determining the detail unit location point set corresponding to each of the target detail units based on the density ratio parameter corresponding to each of the target detail units, and the location information and geometric information of the macroscopic geographic entities to which each of the target detail units belongs; and constructing a three-dimensional model corresponding to the target natural scene based on the macroscopic geographic entities, the location information of the macroscopic geographic entities, the target detail units, and the detail unit location point set.

[0016] Thus, when constructing a 3D natural scene, the input image corresponding to the target natural scene is first acquired and semantically segmented to identify macroscopic geographic entities within the target natural scene, and the location and geometric information of these macroscopic geographic entities are obtained. Further, based on a pre-constructed entity detail association table, the target detail units corresponding to the macroscopic geographic entities and their density ratio parameters are determined. Then, based on the density ratio parameters, the location and geometric information of the macroscopic geographic entities to which the target detail units belong, the set of detail unit location points corresponding to each target detail unit is determined, and the above data is then used to construct a 3D model corresponding to the target natural scene. In this way, the corresponding target detail entities are determined for each macroscopic geographic entity. Furthermore, based on the actual information in the target natural scene, the set of detail unit location points storing the locations where target detail entities need to be placed is dynamically determined. For macroscopic geographic entities, appropriate categories of details are added at appropriate locations; for the microscopic details of natural geographic elements, a reasonable modeling method can be provided, which helps improve the realism of the generated 3D natural scene. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for constructing a three-dimensional natural scene according to an embodiment of this application; Figure 2 This is a schematic diagram of the constituent modules of a three-dimensional natural scene construction system provided in an embodiment of this application; Figure 3 This is a block diagram illustrating the internal structure of a terminal provided in an embodiment of this application. Detailed Implementation

[0019] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0020] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to classification to [the described condition or event]."

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0026] Currently, 3D modeling of natural scenes is receiving increasing attention. Existing technologies are relatively mature in modeling artificial elements such as buildings, but modeling natural elements such as trees and grasslands still needs optimization.

[0027] Specifically, existing technologies mainly focus on representing the macroscopic morphology of geographic entities, such as generating the individual geometric structures and appearances of trees based on procedural or data-driven methods. They lack reasonable modeling methods for the microscopic details of natural geographic elements, thus hindering the improvement of the realism of the generated 3D natural scenes.

[0028] Meanwhile, existing technologies typically treat scenes as a collection of independent objects, lacking a systematic expression of the microscopic details derived from entities and their spatial relationships. Real natural scenes, in addition to macroscopic geographical entities, also include microscopic details formed by environmental processes (such as fallen leaves, broken branches, surface deposits, and attachments), which are crucial for visual realism during close-up observation. The lack of a unified modeling mechanism in existing technologies results in scenes with similar overall shapes but severely insufficient local realism. The lack of a unified multi-scale framework for macroscopic geographical entities and microscopic details makes continuous expression difficult. In some application scenarios, detail generation relies on manual modeling and placement, which is inefficient, lacks scalability, and fails to accurately reflect the spatial relationships between details and entities, affecting geographical plausibility.

[0029] To address at least one of the aforementioned technical problems, this application proposes a method for constructing a three-dimensional natural scene. The method includes: acquiring an input image corresponding to a target natural scene; performing semantic segmentation on the input image to obtain macroscopic geographic entities corresponding to the target natural scene, and acquiring the location and geometric information of the macroscopic geographic entities; wherein the macroscopic geographic entities corresponding to the target natural scene include at least one of multiple preset entities; acquiring a pre-constructed entity detail association table, wherein the entity detail association table stores detail units corresponding to each preset entity, and density ratio ranges corresponding to each detail unit; searching and determining target detail units corresponding to each macroscopic geographic entity from the entity detail association table, and acquiring the density ratio ranges corresponding to each target detail unit; determining density ratio parameters corresponding to the target detail units based on the density ratio ranges; determining the detail unit location point set corresponding to each target detail unit based on the density ratio parameters corresponding to each target detail unit, and the location and geometric information of the macroscopic geographic entity to which each target detail unit belongs; and constructing a three-dimensional model corresponding to the target natural scene based on the macroscopic geographic entities, the location information of the macroscopic geographic entities, the target detail units, and the detail unit location point set.

[0030] Thus, when constructing a 3D natural scene, the input image corresponding to the target natural scene is first acquired and semantically segmented to identify macroscopic geographic entities within the target natural scene, and the location and geometric information of these macroscopic geographic entities are obtained. Further, based on a pre-constructed entity detail association table, the target detail units corresponding to the macroscopic geographic entities and their density ratio parameters are determined. Then, based on the density ratio parameters, the location and geometric information of the macroscopic geographic entities to which the target detail units belong, the set of detail unit location points corresponding to each target detail unit is determined, and the above data is then used to construct a 3D model corresponding to the target natural scene. In this way, the corresponding target detail entities are determined for each macroscopic geographic entity. Furthermore, based on the actual information in the target natural scene, the set of detail unit location points storing the locations where target detail entities need to be placed is dynamically determined. For macroscopic geographic entities, appropriate categories of details are added at appropriate locations; for the microscopic details of natural geographic elements, a reasonable modeling method can be provided, which helps improve the realism of the generated 3D natural scene.

[0031] like Figure 1 As shown in the embodiments of this application, a method for constructing a three-dimensional natural scene is provided. Specifically, the method includes the following steps: Step S100: Obtain the input image corresponding to the target natural scene; Step S200: Semantic segmentation is performed on the input image to obtain the macro-geographic entities corresponding to the target natural scene, and the location and geometric information of the macro-geographic entities are obtained; wherein, the macro-geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; Step S300: Obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detail units corresponding to each of the above-mentioned preset entities, and the density ratio range corresponding to each of the above-mentioned detail units; search and determine the target detail units corresponding to each of the above-mentioned macro-geographic entities from the entity detail association table, and obtain the density ratio range corresponding to each of the above-mentioned target detail units; determine the density ratio parameter corresponding to the above-mentioned target detail units based on the density ratio range. Step S400: Based on the density ratio parameters corresponding to each of the above target detail units, and the location and geometric information of the macro-geographic entities to which each of the above target detail units belongs, determine the set of detail unit location points corresponding to each of the above target detail units. Step S500: Based on the above macro-geographic entities, the location information of the above macro-geographic entities, the above target detail units, and the set of location points of the above detail units, construct a three-dimensional model corresponding to the above target natural scene.

[0032] The aforementioned target natural scene refers to a scene that requires 3D model construction. This scene includes natural geographical elements and may also include artificial elements. It should be noted that in some application scenarios, the above-mentioned 3D natural scene construction method can also be applied to the 3D model construction process of artificial scenes that only include artificial elements; no specific limitations are made here.

[0033] The above-mentioned input images are used to characterize the elements contained in the target natural scene. The input images may include real-scene images and / or remote sensing images obtained by image acquisition of the target natural scene, and may also include other types of images, which are not specifically limited here.

[0034] After acquiring the input image, semantic segmentation is performed to obtain information corresponding to the target natural scene, including macroscopic geographic entities included in the target natural scene and their location and geometric information. The location information of macroscopic geographic entities can be used to limit the position of each macroscopic geographic entity and / or the relative positional relationship between different macroscopic geographic entities when constructing the 3D model, while the geometric information of macroscopic geographic entities can be used to control the shape of the macroscopic geographic entity and / or the number of target detail units corresponding to that macroscopic geographic entity.

[0035] It should be noted that, in this embodiment, the aforementioned macro-geographic entities are independent geographic objects with clear spatial boundaries and geometric shapes, directly identified from the input image through semantic segmentation, and serve as carriers for detail generation. Detail entities are microscopic visual elements attached to macro-geographic entities, generated by the entity's own processes or external environmental effects. Their locations can be confirmed through constrained procedural sampling, and pre-built basic models can be invoked for rendering, thereby enhancing the realism of the entire scene.

[0036] Specifically, the aforementioned preset entities include lawns, water bodies, and ground cover layers, which are planar entities; roads, which are linear entities; and trees, which are point entities. The above-mentioned semantic segmentation of the input image is performed to obtain the macro-geographic entities corresponding to the target natural scene, and the location and geometric information of the macro-geographic entities are obtained, including: Using a pre-defined semantic segmentation model, pixel-level semantic segmentation is performed on the input image to obtain at least one macro-geographic entity corresponding to the target natural scene. For macro-geographic entities that belong to the above-mentioned areal entities, the regional boundary coordinates of the macro-geographic entities are used as location information, and the regional area of ​​the macro-geographic entities is used as geometric information. For macro-geographic entities that belong to the aforementioned linear entities, the centerline coordinates of the aforementioned macro-geographic entities are used as location information, and the length and / or width of the aforementioned macro-geographic entities are used as geometric information. For macro-geographic entities that belong to the aforementioned point-like entities, the centroid coordinates of the macro-geographic entity are used as location information, and the minimum circumscribed circle radius of the macro-geographic entity is used as geometric information.

[0037] The aforementioned preset semantic segmentation model is a pre-trained model. Models such as Semantic Segmentation Transformer (SegFormer), Segment Anything Model (SAM), or other models with image semantic segmentation capabilities can be used, without specific limitations.

[0038] In this embodiment of the application, the input image is segmented at the pixel level based on a preset semantic segmentation model to identify the category of each macro-geographic entity and simultaneously extract its location information and geometric information.

[0039] It should be noted that the aforementioned preset entities may also include other entities, such as forests, rocks, buildings, etc., and can be associated with their respective categories according to actual needs. For example, rocks correspond to planar entities, which are not specifically limited here.

[0040] It should be further explained that, for macroscopic geographic entities belonging to the aforementioned linear entities, the centerline coordinates (i.e., the polyline coordinates corresponding to the centerline) of the macroscopic geographic entity are used as location information, and its length and / or width are used as geometric information. In this embodiment, the length and width of the linear entity are extracted simultaneously as geometric information. In some application scenarios, the length and width can be determined according to the preset or specified extension direction of the linear entity. For example, if the extension direction of a road is specified, the length along the road extension direction is used as the length, and the length perpendicular to the road extension direction is used as the width. In other application scenarios, for roads, the length values ​​of two mutually perpendicular directions can also be identified, and the larger of the two length values ​​is used as the road length, and the other is used as the road width. Other methods can also be used to determine the corresponding length and width, which are not specifically limited here.

[0041] In some application scenarios, the relative positions and topological relationships between macro-geographic entities can also be extracted and used as location information of macro-geographic entities, without specific limitations here.

[0042] Specifically, the aforementioned entity detail association table is a pre-built and stored table used to store the detailed individual units corresponding to each of the aforementioned preset entities, as well as the density ratio range corresponding to each of the aforementioned detailed individual units.

[0043] In some application scenarios, various detailed units are pre-determined based on actual needs, as shown in Table 1 below: Table 1

[0044] Table 1 is a detail monomer definition table provided in the embodiments of this application, showing various predefined detail monomers, the categories and subcategories of each detail monomer, and the function of the detail monomer. The function of the detail monomer is used to describe the information that the detail monomer can represent.

[0045] In this embodiment, the pre-defined entities include trees, lawns, water bodies, roads, and surface cover. The corresponding detail monomers are divided into two categories: biological details and physical details. Biological details originate from life processes such as growth and decay; physical details originate from structural changes or material deposition caused by external physical or chemical effects. The specific detail monomers can be referred to Table 1 above.

[0046] In some application scenarios, a detailed unit definition table can be pre-built and maintained, and the detailed units within it can be updated in real time according to actual needs. Based on this detailed unit definition table, the corresponding entity detail association table can also be updated in real time. An example of an entity detail association table is shown in Table 2 below: Table 2

[0047] Table 2 stores the detailed units corresponding to each macro-geographic entity, as well as the density ratio range corresponding to each detailed unit. The density ratio range is used to limit the value range of the density ratio parameter.

[0048] In this embodiment, the aforementioned entity detail association table also stores the association relationships between macro-geographic entities and detail individuals, but this is not intended as a specific limitation. Specifically, the association relationships between macro-geographic entities and detail individuals include two types: part-of and attached-on. Part-of indicates that the detail is generated by the entity's own life or physical processes and is attached to the entity, such as fallen leaves and broken branches caused by the decay of a tree; attached indicates that the detail originates from external environmental factors and is attached to the entity's surface, such as cracks and oil stains caused by weathering and pollution.

[0049] When the corresponding density ratio range is a preset interval, values ​​can be taken within that range to determine the corresponding density ratio. The density ratio range corresponds to discrete values ​​of 0 or 1, indicating whether the detail appears, with the corresponding density ratio being either 0 or 1. The above-mentioned relationships and density ratio ranges can be used as attribute information corresponding to the detail unit and associated with the determined target detail unit.

[0050] In this embodiment of the application, the aforementioned surface cover layer is used to characterize the surface cover material area in the scene, including discrete material blocks such as sand, soil, rock, gravel road surface, and stone slab paving, as a basic cover layer to support other entities.

[0051] It should be noted that the specific content stored in Tables 1 and 2 above is only an example and not a specific limitation. Furthermore, the use of tables as a storage method is also only an example and not a specific limitation. In practical applications, other data storage methods can be used to store the above information, such as directly storing the relationships between the data.

[0052] Furthermore, the determination of the density ratio parameters corresponding to the target detail unit based on the aforementioned density ratio range includes: For each of the aforementioned target detail units, a random value is selected from the density ratio range corresponding to the aforementioned target detail unit, and this value is used as the density ratio parameter corresponding to the aforementioned target detail unit.

[0053] In some application scenarios, random values ​​can be generated based on the aforementioned density ratio range and used as density ratio parameters. In other application scenarios, values ​​within the aforementioned density ratio range, either pre-set by the user or input in real-time, can also be used as density ratio parameters. In still other application scenarios, current geographic information can be obtained, and based on the aforementioned geographic information and density ratio range, the density ratio parameters can be calculated and determined using a pre-trained parameter mapping model or pre-defined parameter mapping rules. The aforementioned geographic information includes at least one of seasonal information, climate information corresponding to the target natural scene, and latitude and longitude information. This allows for better adjustment of the number of individual detail units based on the actual geographic information of the target natural scene.

[0054] Furthermore, the determination of the set of location points for each of the aforementioned target detail units, based on the density ratio parameters corresponding to each of the aforementioned target detail units and the location and geometric information of the macroscopic geographic entities to which each of the aforementioned target detail units belongs, includes: For each of the aforementioned target detail units, the detail distribution range is determined based on the location information of the macro-geographic entity to which the target detail unit belongs. The number of detail distributions is determined based on the geometric information of the macro-geographic entity to which the target detail unit belongs and the density ratio parameter corresponding to the target detail unit. Based on the number of detail distributions and the detail distribution range, a set of detail unit location points corresponding to the target detail unit is generated.

[0055] Specifically, for a target detail unit, if its macro-geographic entity is an area entity, the detail distribution range corresponding to the target detail unit is determined based on its regional boundary coordinates. This detail distribution range is set near the regional boundary, and the specific area can be determined based on a preset first distance. For example, the first distance can be extended inwards and / or outwards along the aforementioned regional boundary (the specific value can be set and adjusted according to actual needs), and the obtained range is used as the detail distribution range. The number of detail distributions is calculated based on the area corresponding to the macro-geographic entity and the aforementioned density ratio parameter. In some application scenarios, the number of detail distributions is determined by the product of the aforementioned regional area, the aforementioned density ratio parameter, and a preset unit area control parameter. The preset unit area control parameter is a pre-set maximum number of instances per unit area, and its specific value can be set and adjusted according to actual needs.

[0056] For a target detail unit, if its macro-geographic entity is a linear entity, the detail distribution range corresponding to the target detail unit is determined based on its centerline coordinates. This detail distribution range is set near the centerline, and the specific area can be determined based on a preset second distance (the specific value can be set and adjusted according to actual needs). For example, the second distance can be extended along the centerline in a direction perpendicular to the centerline, and the obtained range can be used as the detail distribution range. The number of detail distributions is calculated based on the length and / or width of the macro-geographic entity and the aforementioned density ratio parameter. In some application scenarios, the length, width, or the average of the length and width of the macro-geographic entity is used as the entity's actual control parameter. The number of detail distributions is determined by multiplying the entity's actual control parameter, the density ratio parameter, and a preset unit length control parameter. The preset unit length control parameter is the maximum number of instances within a pre-set unit length, and its specific value can be set and adjusted according to actual needs.

[0057] For a target detail unit, if its macro-geographic entity is a point entity, the detail distribution range corresponding to the target detail unit is determined based on its centroid coordinates. This detail distribution range is set near the centroid coordinates, and the specific area can be determined based on a preset third distance (the specific value can be set and adjusted according to actual needs). For example, the centroid coordinates can be used as the center point, and the preset third distance as the radius to determine the corresponding detail distribution range. The number of detail distributions is then calculated based on the minimum circumscribed circle radius of the macro-geographic entity and the aforementioned density ratio parameter. In some application scenarios, the minimum circumscribed circle area is calculated based on the minimum circumscribed circle radius. The number of detail distributions is determined by multiplying the minimum circumscribed circle area, the aforementioned density ratio parameter, and a preset unit area control parameter. The preset unit area control parameter is the maximum number of instances per unit area, and its specific value can be set and adjusted according to actual needs. It should be noted that the unit area control parameters for areal entities and point entities can be the same or different; no specific limitations are imposed here. Furthermore, for each target detail unit, based on the number of detail distributions and the range of detail distributions mentioned above, a set of detail unit location points matching the number of detail distributions is generated within the range of detail distributions using a preset sampling method for the target detail unit.

[0058] It should be noted that a corresponding sampling method can be set for each target detail unit corresponding to each macro-geographic entity. When the same detail unit belongs to different macro-geographic entities, different sampling methods can be used. The specific sampling method can be set and adjusted according to actual needs, and no specific limitation is made here.

[0059] In some application scenarios, for a target detail unit, if its association with its corresponding macro-geographic entity is attachment, and the target detail unit is distributed on the geometric surface of the entity (e.g., potholes, cracks, oil stains, puddles, gravel, and dust on the ground), the corresponding sampling method can be surface sampling. Within the surface space defined by the aforementioned detail distribution range, candidate sampling points are uniformly generated. The surface normal vector at each point is calculated as the placement direction of the detail. Random perturbations are applied to the sampling point positions within a preset offset range. This preset offset range is a fixed parameter predefined offline to control the naturalness of the detail distribution and avoid excessive regularity. In this way, the set of detail unit locations and their corresponding normal vectors attached to the ground surface are obtained.

[0060] In other application scenarios, for a specific target detail, if its association with its corresponding macro-geographic entity is intrinsic, and the detail is distributed within the entity's defined area (e.g., grass and flowers within a lawn), then region sampling can be used. Within the region boundary defined by the aforementioned detail distribution range, candidate points are randomly generated. The distance between each candidate point and existing sampled points is checked to see if it exceeds a preset minimum distance. If it does, it is retained; otherwise, it is discarded. This process is repeated until no new points can be added (or the required number of detail distributions is reached). In this way, a set of location points for detail units distributed within the lawn area is obtained.

[0061] In other application scenarios, for a target detail unit, if its association with its corresponding macro-geographic entity is attachment, and the detail is distributed along a linear path (e.g., potholes, cracks, oil stains, puddles, gravel, and dust on a road), path sampling can be used. Based on the aforementioned detail distribution range, sampling points are generated sequentially along the broken line corresponding to the road centerline at longitudinal sampling intervals. For each sampling point, lateral displacement and phase offset can be applied simultaneously to simulate uneven distribution. It should be noted that the sampling interval can be determined based on the road length and the number of detail distributions, and can also be determined in conjunction with the curvature of each road segment. For example, the sampling interval can be reduced for road segments with higher curvature to increase detail density. Furthermore, for each sampling point, the lateral displacement can be a random value within a preset lateral offset range, and the phase offset can be a random value within a preset phase offset range. Alternatively, it can be set and adjusted according to actual needs, without specific limitations here. In this way, a set of detail unit location points distributed near the road centerline is obtained.

[0062] In other application scenarios, for a target detail unit, if its association with its corresponding macro-geographic entity is intrinsic, and the details are distributed within the entity's vertical projection area (e.g., leaves, twigs, moss, and lichen under a tree), projection sampling can be used. For the detail distribution range determined by the centroid coordinates of the macro-geographic entity, sampling points are randomly generated based on the number of detail distributions to determine the set of detail unit location points. It should be noted that in some application scenarios, these sampling points are used as the set of detail unit location points, and a projection attribute is added to indicate the projection processing of the location points. That is, in the actual rendering process, for each location point, based on the base model corresponding to the macro-geographic entity, each location point is projected onto the surface of the base model corresponding to the macro-geographic entity (e.g., the base model corresponding to the land surface or water body) to obtain the final placement position of the detail unit.

[0063] In a specific application scenario, for the entity of the ground, details such as potholes, cracks, oil stains, puddles, gravel, and dust are sampled using surface sampling; for the entity of the lawn, details such as flowers and grass are sampled using region sampling; for the entity of the road, details such as potholes, cracks, oil stains, puddles, gravel, and dust are sampled using path sampling; and for the entities of trees and water bodies, details such as leaves, twigs, moss, and lichen are sampled using projection sampling.

[0064] Furthermore, based on the aforementioned macro-geographic entities, the location information of the aforementioned macro-geographic entities, the aforementioned target detail units, and the aforementioned set of location points for the detail units, the construction of the 3D model corresponding to the natural scene of the aforementioned target includes: Using the aforementioned macro-geographic entities as parent nodes and the target detail units corresponding to the aforementioned macro-geographic entities as child nodes corresponding to the aforementioned parent nodes, a semantic tree is constructed. The location information of the aforementioned macro-geographic entities is used as the attribute information of the aforementioned parent nodes, and the location point set of the aforementioned detail units is used as the attribute information of the aforementioned child nodes. Based on the semantic tree described above, construct a 3D model corresponding to the target natural scene.

[0065] In this embodiment, a semantic tree is constructed and a 3D model corresponding to the target natural scene is built based on the semantic tree. The semantic tree is composed of nodes and is divided into entity level and detail level according to semantic granularity. The parent node is the macro-geographic entity and the child node is the subordinate detail element (i.e. the target detail unit corresponding to the macro-geographic entity represented by the parent node).

[0066] Specifically, macroscopic geographic entities are used as parent nodes, and child nodes are constructed based on the target detailed units. These child nodes are then attached to their corresponding parent nodes. For each node, corresponding attribute information can be added, which is used to guide the subsequent rendering process. For parent nodes, the category, name, and location information of the macroscopic geographic entity can be used as attribute information; for child nodes, the category, name, location set of the detailed unit, and the relationship between the detailed unit and the parent node can be used as attribute information. For example, after identifying trees, child nodes such as leaves, twigs, moss, and lichen are automatically bound; after identifying roads, child nodes such as cracks, oil stains, puddles, gravel, dust, and potholes are automatically bound.

[0067] Furthermore, the construction of the 3D model corresponding to the target natural scene based on the semantic tree includes: For each of the above parent nodes, select a target entity base model from at least one preset entity base model associated with the macro-geographic entity corresponding to the above parent node, and perform 3D rendering based on the target entity base model and the attribute information corresponding to the above parent node. For each of the above-mentioned sub-nodes, a target detail base model is determined for each point in the set of position points of the detail unit corresponding to the above-mentioned sub-node from at least one preset detail base model associated with the target detail unit corresponding to the above-mentioned sub-node. 3D rendering is performed based on the target detail base model and the attribute information corresponding to the above-mentioned sub-node. After completing the 3D rendering of the aforementioned parent node and child node, a 3D model representing the aforementioned target natural scene is obtained.

[0068] In this embodiment, a basic model library is pre-built. For each macro-geographic entity, at least one preset entity basic model is pre-set; for each detailed entity, at least one preset detail basic model is pre-set. The preset detail basic models are divided into two categories: models containing geometry and materials, used for details with independent shapes such as grass, leaves, and twigs; and models containing only materials, used for details such as cracks, oil stains, and potholes that are attached to the entity surface through textures. The basic models can be modified through parameter attributes, are stored offline in the model library, and are called on demand during real-time processing. For a macro-geographic entity, one of its associated preset entity basic models is randomly selected (or specified by the user) as the target entity basic model. The target entity basic model is adjusted according to the preset affine transformation parameters corresponding to the entity, and a 3D model is rendered in conjunction with the attribute parameters corresponding to the macro-geographic entity. The aforementioned affine transformation parameters include displacement parameters, rotation parameters, scaling parameters, etc. These parameters can be randomly selected from a preset range according to actual needs and are not specifically limited here.

[0069] For a target detail unit, similar processing is performed on each location point in the set of location points of its associated detail units to determine the model corresponding to that point for rendering. Specifically, for each location point, the positional relationship between it and the rendering model corresponding to the entity is determined during the rendering process to identify the target point. For each target point, one of the various preset detail base models associated with it is randomly selected (or one is specified by the user) as the target detail base model corresponding to that target point. The target detail base model is adjusted according to the affine transformation parameters corresponding to the target point, and the 3D model is rendered by combining the attribute parameters of the child nodes to which the target point belongs.

[0070] It should be noted that when rendering the base model, attributes such as orientation angle, scaling ratio, and color value can be set for each base model. In one application scenario, the orientation angle is randomly generated within a preset range; the scaling ratio is randomly generated within a preset range and applied evenly in the 3D direction to maintain the correct proportions; the color value is randomly generated within the preset color range for that detail type. Since the base model does not contain finer substructures, color modifications affect the entire instance and do not affect other types of details or entities.

[0071] It should be further noted that when rendering 3D models, you can use preset 3D rendering software or a preset 3D rendering platform; no specific restrictions are imposed here.

[0072] This application provides a method for constructing a three-dimensional natural scene. The method includes: acquiring an input image corresponding to a target natural scene; performing semantic segmentation on the input image to obtain macroscopic geographic entities corresponding to the target natural scene, and acquiring the location and geometric information of the macroscopic geographic entities; wherein the macroscopic geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; acquiring a pre-constructed entity detail association table, wherein the entity detail association table is used to store detail units corresponding to each preset entity, and the density ratio range corresponding to each detail unit; searching and determining target detail units corresponding to each macroscopic geographic entity from the entity detail association table, and acquiring the density ratio range corresponding to each target detail unit; determining the density ratio parameter corresponding to the target detail unit based on the density ratio range; determining the detail unit location point set corresponding to each target detail unit based on the density ratio parameter corresponding to each target detail unit, and the location and geometric information of the macroscopic geographic entity to which each target detail unit belongs; and constructing a three-dimensional model corresponding to the target natural scene based on the macroscopic geographic entities, the location information of the macroscopic geographic entities, the target detail units, and the detail unit location point set.

[0073] Thus, when constructing a 3D natural scene, the input image corresponding to the target natural scene is first acquired and semantically segmented to identify macroscopic geographic entities within the target natural scene, and the location and geometric information of these macroscopic geographic entities are obtained. Further, based on a pre-constructed entity detail association table, the target detail units corresponding to the macroscopic geographic entities and their density ratio parameters are determined. Then, based on the density ratio parameters, the location and geometric information of the macroscopic geographic entities to which the target detail units belong, the set of detail unit location points corresponding to each target detail unit is determined, and the above data is then used to construct a 3D model corresponding to the target natural scene. In this way, the corresponding target detail entities are determined for each macroscopic geographic entity. Furthermore, based on the actual information in the target natural scene, the set of detail unit location points storing the locations where target detail entities need to be placed is dynamically determined. For macroscopic geographic entities, appropriate categories of details are added at appropriate locations; for the microscopic details of natural geographic elements, a reasonable modeling method can be provided, which helps improve the realism of the generated 3D natural scene.

[0074] Specifically, the above-mentioned three-dimensional natural scene construction method can be applied not only to scenes containing typical natural elements such as trees, lawns, and water bodies, but also to detailed modeling of other complex natural scenes, such as rock weathering, soil erosion, and leaf litter accumulation, which is conducive to improving the overall coordination and visual coherence of large-scale natural landscapes.

[0075] like Figure 2 As shown, corresponding to the above-described three-dimensional natural scene construction method, this application embodiment also provides a three-dimensional natural scene construction system, which includes: Image acquisition module 210 is used to acquire the input image corresponding to the target natural scene; The semantic segmentation module 220 is used to perform semantic segmentation on the input image to obtain the macro-geographic entities corresponding to the target natural scene, and to obtain the location information and geometric information of the macro-geographic entities; wherein, the macro-geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; The detailed unit determination module 230 is used to obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detailed units corresponding to each of the preset entities, and the density ratio range corresponding to each of the detailed units; from the entity detail association table, the target detailed units corresponding to each of the macro-geographic entities are searched and determined, and the density ratio range corresponding to each of the target detailed units is obtained; based on the density ratio range, the density ratio parameter corresponding to the target detailed unit is determined. The detailed location determination module 240 is used to determine the set of detailed unit location points corresponding to each of the above-mentioned target detailed units based on the density ratio parameters corresponding to each of the above-mentioned target detailed units, as well as the location information and geometric information of the macro-geographic entities to which each of the above-mentioned target detailed units belongs. The scene construction module 250 is used to construct a 3D model corresponding to the natural scene of the target based on the macro-geographic entities, the location information of the macro-geographic entities, the target detail units, and the location point set of the detail units.

[0076] Thus, when constructing 3D natural scenes, for macro-geographic entities, adding appropriate categories of details in reasonable locations, and for micro-details of natural geographic elements, we can provide reasonable modeling methods, which helps to improve the realism of the generated 3D natural scenes.

[0077] It should be noted that the specific structure and implementation of the above-mentioned three-dimensional natural scene construction system and its various modules or units can be referred to the corresponding descriptions in the above method embodiments, and will not be repeated here.

[0078] It should be noted that the division of the various modules in the above-mentioned 3D natural scene construction system is not unique and is not intended as a specific limitation.

[0079] Based on the above embodiments, this application also provides a terminal, the principle block diagram of which can be as follows: Figure 3 As shown. The terminal includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps of any of the above-described three-dimensional natural scene construction methods. The display screen can be a liquid crystal display (LCD) or an e-ink display.

[0080] Those skilled in the art will understand that Figure 3 The block diagram shown is only a partial structural diagram related to the solution of this application and does not constitute a limitation on the terminal on which the solution of this application is applied. The specific terminal may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0081] In one embodiment, a terminal is provided, the terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of any of the three-dimensional natural scene construction methods provided in the embodiments of this application.

[0082] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of any of the three-dimensional natural scene construction methods provided in this application.

[0083] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0084] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the above device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments 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. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0085] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0086] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0087] In the embodiments provided in this application, it should be understood that the disclosed systems / terminal devices and methods can be implemented in other ways. For example, the system / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units described above is merely a logical functional division, and in actual implementation, it can be divided in other ways. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0088] If the integrated modules / units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, and software distribution media, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0089] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions are not in essence a departure from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for constructing a three-dimensional natural scene, characterized in that, The method includes: Obtain the input image corresponding to the target natural scene; The input image is semantically segmented to obtain the macro-geographic entities corresponding to the target natural scene, and the location and geometric information of the macro-geographic entities are obtained; wherein, the macro-geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; Obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detail units corresponding to each preset entity and the density ratio range corresponding to each detail unit; search and determine the target detail units corresponding to each macro-geographic entity from the entity detail association table, and obtain the density ratio range corresponding to each target detail unit; determine the density ratio parameter corresponding to the target detail unit based on the density ratio range. Based on the density ratio parameters corresponding to each target detail unit, and the location and geometric information of the macro-geographic entity to which each target detail unit belongs, determine the set of detail unit location points corresponding to each target detail unit; Based on the macro-geographic entity, the location information of the macro-geographic entity, the target detail unit, and the location point set of the detail unit, a 3D model corresponding to the target natural scene is constructed.

2. The method for constructing a three-dimensional natural scene according to claim 1, characterized in that, The various preset entities include lawns, water bodies, and ground cover layers, which are planar entities; roads, which are linear entities; and trees, which are point entities. The step of performing semantic segmentation on the input image to obtain the macro-geographic entities corresponding to the target natural scene, and acquiring the location and geometric information of the macro-geographic entities, includes: The input image is segmented at the pixel level using a preset semantic segmentation model to obtain at least one macro-geographic entity corresponding to the target natural scene. For macro-geographic entities belonging to the areal entities, the regional boundary coordinates of the macro-geographic entities are used as location information, and the regional area of ​​the macro-geographic entities is used as geometric information. For macro-geographic entities that belong to the linear entity, the centerline coordinates of the macro-geographic entity are used as location information, and the length and / or width of the macro-geographic entity are used as geometric information. For macroscopic geographic entities that belong to the point-like entities, the centroid coordinates of the macroscopic geographic entity are used as location information, and the minimum circumscribed circle radius of the macroscopic geographic entity is used as geometric information.

3. The method for constructing a three-dimensional natural scene according to claim 1, characterized in that, Determining the density ratio parameter corresponding to the target detail unit based on the density ratio range includes: For each target detail unit, a random value is taken from the density ratio range corresponding to the target detail unit, and used as the density ratio parameter corresponding to the target detail unit.

4. The method for constructing a three-dimensional natural scene according to claim 1, characterized in that, The step of determining the set of location points for each target detail unit based on the density ratio parameter corresponding to each target detail unit, and the location and geometric information of the macroscopic geographic entity to which each target detail unit belongs, includes: For each target detail unit, the detail distribution range is determined based on the location information of the macro-geographic entity to which the target detail unit belongs. The number of detail distributions is determined based on the geometric information of the macro-geographic entity to which the target detail unit belongs and the density ratio parameter corresponding to the target detail unit. Based on the number of detail distributions and the detail distribution range, a set of detail unit location points corresponding to the target detail unit is generated.

5. The method for constructing a three-dimensional natural scene according to claim 1, characterized in that, The step of constructing a 3D model corresponding to the target natural scene based on the macro-geographic entity, the location information of the macro-geographic entity, the target detailed unit, and the set of location points of the detailed unit includes: A semantic tree is constructed by taking the macro-geographic entity as the parent node and the target detail unit corresponding to the macro-geographic entity as the child node corresponding to the parent node. The location information of the macro-geographic entity is used as the attribute information of the parent node, and the location point set of the detail unit is used as the attribute information of the child node. A 3D model corresponding to the target natural scene is constructed based on the semantic tree.

6. The method for constructing a three-dimensional natural scene according to claim 5, characterized in that, The step of constructing a 3D model corresponding to the target natural scene based on the semantic tree includes: For each parent node, a target entity base model is selected from at least one preset entity base model associated with the macro-geographic entity corresponding to the parent node, and 3D rendering is performed based on the target entity base model and the attribute information corresponding to the parent node. For each of the child nodes, a target detail base model is determined for each point in the set of position points of the detail unit corresponding to the child node from at least one preset detail base model associated with the target detail unit corresponding to the child node, and 3D rendering is performed based on the target detail base model and the attribute information corresponding to the child node. After completing the 3D rendering of the parent node and the child node, a 3D model representing the target natural scene is obtained.

7. A three-dimensional natural scene construction system, characterized in that, The system includes: The image acquisition module is used to acquire the input image corresponding to the target natural scene; The semantic segmentation module is used to perform semantic segmentation on the input image to obtain macro-geographic entities corresponding to the target natural scene, and to acquire the location and geometric information of the macro-geographic entities; wherein, the macro-geographic entities corresponding to the target natural scene include at least one of a variety of preset entities; The detailed unit determination module is used to obtain a pre-constructed entity detail association table, wherein the entity detail association table is used to store the detailed units corresponding to each preset entity and the density ratio range corresponding to each detailed unit; from the entity detail association table, the module searches and determines the target detailed units corresponding to each macro-geographic entity and obtains the density ratio range corresponding to each target detailed unit; and determines the density ratio parameter corresponding to the target detailed unit based on the density ratio range. The detailed location determination module is used to determine the set of detailed unit location points corresponding to each of the target detailed units based on the density ratio parameters corresponding to each of the target detailed units, as well as the location information and geometric information of the macro-geographic entities to which each of the target detailed units belongs; The scene construction module is used to construct a 3D model corresponding to the target natural scene based on the macro-geographic entity, the location information of the macro-geographic entity, the target detail unit, and the location point set of the detail unit.

8. A terminal, characterized in that, The terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the steps of the three-dimensional natural scene construction method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the three-dimensional natural scene construction method as described in any one of claims 1 to 6.

Images