A method and system for three-dimensional modeling of large-scale trees

Abstract

This invention relates to the field of tree modeling technology and discloses a method for 3D modeling of large-scale trees. The method includes: acquiring a multi-scale digital image set of the target tree using an image acquisition device; the multi-scale digital image set consists of image data captured by the user at different viewing distances; processing the multi-scale digital image set to identify and match visual feature points between images, and generating an initial 3D point cloud based on these visual feature points; performing density enhancement processing on the initial 3D point cloud to generate a dense 3D point cloud representing the surface morphology of the tree; performing meshing processing on the dense 3D point cloud to construct a 3D mesh model of the tree surface; and applying color information from the multi-scale digital image set to the 3D mesh model to establish a virtual 3D model of the trees in the park. This method achieves high-fidelity, refined, and efficient 3D modeling of large-scale trees through multi-scale image acquisition and a series of point cloud and mesh processing steps.

Description

Technical Field

[0001] This invention relates to the field of tree modeling technology, and specifically to a method and system for three-dimensional modeling of large-sized trees. Background Technology

[0002] Currently, existing methods for detailed tree modeling generally do not differentiate between specific trees and their dimensions, failing to realistically simulate the actual condition of large-scale trees. Therefore, designing a method for detailed tree modeling has become a pressing technical problem for those skilled in the art. Summary of the Invention

[0003] To address the aforementioned shortcomings, this invention discloses a method for three-dimensional modeling of large-sized trees, which enables rapid and accurate modeling of large-sized trees, facilitating subsequent analysis.

[0004] The first aspect of this invention discloses a method for three-dimensional modeling of large-sized trees, comprising: A multi-scale digital image set of the target tree is acquired through an image acquisition device; the multi-scale digital image set consists of image data taken by the user at different observation distances. The multi-scale digital image set is processed to identify and match visual feature points between images, and an initial three-dimensional point cloud is generated based on the visual feature points. The initial three-dimensional point cloud is subjected to density enhancement processing to generate a dense three-dimensional point cloud that characterizes the surface morphology of trees. The dense three-dimensional point cloud is meshed to construct a three-dimensional mesh model of the tree surface; Color information from the multi-scale digital image set is applied to the three-dimensional mesh model to create a virtual three-dimensional model of the trees in the park.

[0005] As an optional implementation, in a first aspect of the present invention, acquiring a multi-scale digital image set of the target tree through an image acquisition device includes: At the first observation distance, a global outline image of the target tree is acquired using an image acquisition device; At the second observation distance, an image of the main branch structure of the target tree is acquired by an image acquisition device, wherein the second observation distance is less than the first observation distance; At a third observation distance, a surface texture image of the target tree is acquired using an image acquisition device, wherein the third observation distance is less than the second observation distance; The global contour image, the main branch structure image, and the surface texture image constitute a multi-scale data image set.

[0006] As an optional implementation, in the first aspect of the present invention, after performing meshing processing on the dense three-dimensional point cloud to construct a three-dimensional mesh model of the tree surface, the method further includes: The three-dimensional mesh model is repaired, including filling data holes and smoothing surface noise. Assign wind resistance importance weights to the mesh elements of the three-dimensional mesh model; A mesh simplification algorithm is performed on the three-dimensional mesh model based on the aforementioned wind resistance importance weights; Assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the Gaussian curvature or average curvature of the mesh vertices on the surface of the three-dimensional mesh model, and assigning the wind resistance importance weights according to the calculated curvature values, wherein regions with absolute curvature values ​​higher than a first threshold are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: identifying ridges or edges representing branching in the three-dimensional mesh model, and assigning higher wind resistance importance weights to mesh elements near the identified ridges or edges. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the projected area of ​​each triangular facet of the three-dimensional mesh model on a plane perpendicular to the wind direction for at least one predetermined wind direction; assigning wind resistance importance weights to the corresponding mesh elements based on the projected area, wherein facets that contribute larger projected areas are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: performing a fluid dynamics simulation on the three-dimensional mesh model to obtain wind pressure distribution data on the model surface, and assigning wind resistance importance weights to the mesh elements based on the wind pressure distribution data, wherein regions with wind pressure absolute values ​​or wind pressure gradients higher than a second threshold are assigned higher wind resistance importance weights.

[0007] As an optional implementation, in the first aspect of the present invention, after establishing the virtual three-dimensional model of the trees in the park, the method further includes: A virtual 3D model of the target tree is obtained, and the virtual 3D model is processed to obtain a structural skeleton, the structural skeleton including multiple nodes and branch components connecting the nodes; The structural skeleton is divided according to the set division logic to obtain the corresponding structural units; The stress points are determined on the structural units of the structural frame according to the set environmental load conditions.

[0008] As an optional implementation, in a first aspect of the present invention, dividing the structural skeleton according to a set partitioning logic to obtain corresponding structural units includes: Identify the main path from the structural skeleton, locate multiple branch points on the main path, and extract candidate branches from each branch point. One or more main branches are selected from the candidate branches based on an importance metric, wherein each selected main branch and its subtree are determined as a corresponding structural unit; wherein the importance metric is to calculate the subtree size of each candidate branch and select one or more candidate branches with the largest subtree size as the main branches.

[0009] As an optional implementation, in the first aspect of the present invention, after establishing the virtual three-dimensional model of the trees in the park, the method further includes: Configure material properties for the branch components in the virtual 3D model according to the tree species, including density, Young's modulus and bending strength; Configure hydrodynamic properties for the components representing the tree canopy in the virtual 3D model, including leaf area density and drag force coefficient, and define the tree canopy as a porous medium in the computational fluid dynamics simulation; Configure corresponding detection logic rules for the virtual 3D model, the detection logic rules including: Monitor health indicators and environmental conditions associated with the tree components in the virtual 3D model; In response to determining that the health indicator is below a first threshold and the environmental conditions exceed a second threshold, the state of the tree component is automatically switched from the first state to the second state, which is at a higher risk. Trigger the system response action associated with the second state; Alternatively, the detection logic rules include: The wind pressure acting on the tree branch components in the virtual three-dimensional model is calculated using fluid dynamics simulation. The calculated wind pressure value is compared with a predefined critical fracture strength value for the tree branch assembly; In response to determining that the wind pressure value exceeds the critical fracture strength value, the fracture of the tree branch component is simulated in the virtual three-dimensional model, and the hydrodynamic properties of the three-dimensional model are updated to reflect the geometric changes after fracture.

[0010] As an optional implementation, in the first aspect of the present invention, the three-dimensional modeling method further includes: acquiring UAV images and calculating a canopy density index based on the UAV images, wherein the canopy density index is a dimensionless parameter characterizing the density of tree canopies; The calculated canopy density index is input into a transformation mapping function to obtain the output value of the transformation mapping function, and this output value is assigned as the updated leaf area density value to the corresponding canopy component of the virtual 3D model. The transformation mapping function is constructed as follows: measured data pairs of canopy density index and leaf area density of multiple sample trees are obtained; based on the measured data pairs, the mapping function is trained by a regression algorithm, wherein the input of the regression algorithm includes the canopy density index, and the output is the predicted leaf area density value.

[0011] As an optional implementation, in the first aspect of the present invention, the system response action includes: Based on the calculated bending stress and the set bending strength, the additional support force required to reduce the stress of the structural unit to a safe level is calculated; the bending stress is obtained using a wind force calculation formula, which is: Where F represents the wind force experienced by the corresponding structural unit. Let be the air density, v be the wind speed at the corresponding point, R be the characteristic radius of the corresponding structural unit, and L be the length of the corresponding structural unit. Based on the additional support force, the installation position and orientation of the support component are simulated in the virtual 3D model; a suggested support scheme including the parameters of the support component is output in the user interface; the support component is an elastic cable; The proposed support scheme was determined in the following manner: In the virtual three-dimensional model, the connection path of the elastic cable between the structural unit and the anchor point is determined; The required tension value of the elastic cable is calculated based on the connection path to ensure that its tension component can offset part of the wind load, thereby making the combined bending stress of the structural unit lower than the set bending strength. Based on the stated tension value and cable specifications, determine whether the load is within a safe operating range. If so, output the corresponding recommended support scheme.

[0012] A second aspect of this invention discloses a system for three-dimensional modeling of large-sized trees, comprising: Acquisition module: used to acquire a multi-scale digital image set of the target tree through an image acquisition device; the multi-scale digital image set is image data taken by the user at different observation distances; Matching module: used to process the multi-scale digital image set to identify and match visual feature points between images, and generate an initial three-dimensional point cloud based on the visual feature points; Generation module: used to perform density enhancement processing on the initial three-dimensional point cloud to generate a dense three-dimensional point cloud that characterizes the surface morphology of trees; Processing module: used to perform meshing processing on the dense 3D point cloud to construct a 3D mesh model of the tree surface; Model building module: used to apply color information from the multi-scale digital image set to the three-dimensional mesh model to build a virtual three-dimensional model of the park trees.

[0013] A third aspect of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calling the executable program code stored in the memory to execute the method for three-dimensional modeling of large-sized trees disclosed in the first aspect of the present invention.

[0014] A fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program causes a computer to execute the method for three-dimensional modeling of large-sized trees disclosed in the first aspect of the present invention.

[0015] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The method in this embodiment of the invention, based on image acquisition, eliminates the need for complex laser scanning equipment. Data acquisition can be completed using only conventional image acquisition devices, reducing equipment costs and operational barriers for 3D modeling of large-scale trees. Through multi-scale image acquisition and a series of point cloud and mesh processing procedures, high-fidelity, detailed, and efficient 3D modeling of large-scale trees is achieved, while also possessing both visualization effects and practical application value. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the method for three-dimensional modeling of large-sized trees disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the specific process of multi-scale image capture disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the specific process of image capture as disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram showing the three-dimensional model disclosed in the embodiment of the present invention; Figure 5This is a schematic diagram of the structure of a system for three-dimensional modeling of large-sized trees provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that the terms first, second, third, fourth, etc., in the specification and claims of this invention are used to distinguish different objects, not to describe a specific order. The terms used in the embodiments of this invention include and have, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0020] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for 3D modeling of large-scale trees disclosed in an embodiment of the present invention. The execution entity of the method described in this embodiment is an execution entity composed of software and / or hardware. This execution entity can receive relevant information and send certain instructions via wired and / or wireless means. It may also have certain processing and storage functions. This execution entity can control multiple devices, such as remote physical servers or cloud servers and related software, or local hosts or servers and related software that perform related operations on devices located in a certain location. In some scenarios, multiple storage devices can also be controlled; these storage devices may be placed in the same location as the devices or in different locations. Figure 1 As shown, the method for 3D modeling of large-sized trees includes the following steps: S101: Acquire a multi-scale digital image set of the target tree through an image acquisition device; the multi-scale digital image set consists of image data captured by the user at different observation distances. S102: Process the multi-scale digital image set to identify and match visual feature points between images, and generate an initial three-dimensional point cloud based on the visual feature points; S103: Perform density enhancement processing on the initial three-dimensional point cloud to generate a dense three-dimensional point cloud that characterizes the surface morphology of the tree. S104: The dense three-dimensional point cloud is meshed to construct a three-dimensional mesh model of the tree surface; S105: Apply color information from the multi-scale digital image set to the three-dimensional mesh model to establish a virtual three-dimensional model of the trees in the park.

[0021] The multi-scale digital image set in this invention includes tree image data at different observation distances. This captures both the overall outline of the tree and detailed features such as bark texture and leaf details, providing rich visual information for feature point matching and ensuring that the initial 3D point cloud accurately reflects the true shape of the tree. Then, density enhancement processing of the initial point cloud generates a dense 3D point cloud, filling in the gaps in feature points on the tree surface. This solves the morphological distortion problem caused by sparse point clouds in traditional modeling, enabling the model to finely represent the surface texture and other morphological details of the tree, as well as the morphological features of the canopy of large-scale trees. This is crucial for subsequent surface wind pressure analysis.

[0022] An initial 3D point cloud is generated by matching visual feature points. The spatial structural framework of the tree is constructed based on the correlation features between images, avoiding the one-sidedness of single-view modeling and ensuring the integrity of the overall structure of large-scale trees (such as the spatial distribution of trunks, main branches, and lateral branches). Meshing processing transforms the dense point cloud into a 3D mesh model, converting discrete point cloud data into a continuous surface structure, giving the tree model a clear topological relationship.

[0023] By mapping the color information of multi-scale images onto a 3D mesh model, the virtual 3D model can reproduce the real colors and textures of trees, such as the color of leaves and the color gradient of bark, which greatly improves the visual realism of the model and can meet the visualization needs of scenarios such as virtual parks and digital landscapes.

[0024] Image-based modeling eliminates the need for complex laser scanning equipment; data acquisition can be completed using only conventional image acquisition devices, reducing equipment costs and operational barriers for 3D modeling of large-scale trees. The generated virtual 3D models can be directly applied to the digital management of trees in parks and specific scene simulations, which is particularly significant for tree protection; they enable proactive preparedness for impending disasters through realistic simulations.

[0025] Specifically, using laser point clouds presents challenges such as occlusion leading to data control issues and uneven sampling density, resulting in loss of detail. This can cause significant deviations in subsequent fluid dynamics calculations. Furthermore, laser point cloud acquisition and processing are costly, making it difficult to quickly update the model to reflect morphological changes such as tree pruning, leaf drop, and branch breakage. If continuous CFD simulations of wind redistribution after branch breakage are required, the laser point cloud model cannot quickly adjust canopy geometry and LAD parameters, causing simulation interruptions.

[0026] The image acquisition point cloud provided in this embodiment of the invention, through multi-level and multi-angle shooting of global, branch, and texture layers, can completely restore the spatial distribution of the outer branches and leaves, inner main branches, and twigs of large-sized tree canopies. This avoids the problem of insufficient penetration of laser point clouds, enabling CFD simulation to accurately simulate the process of airflow obstruction, flow around, and eddy generation in the canopy, such as the acceleration effect of airflow passing through the edge of the canopy, and the deceleration and pressure accumulation of airflow in the middle and lower parts of the canopy, significantly improving the calculation accuracy of wind resistance coefficient and wind pressure distribution. After meshing, the dense point cloud generated from the image acquisition point cloud has a continuous and smooth surface, which can directly generate a high-quality computational mesh with low mesh distortion rate. This ensures the convergence of CFD numerical calculations and reduces the time cost of mesh repair. The acquired point cloud can be combined with UAV imagery to calculate the canopy density index. Through a mapping function trained on the canopy density index-LAD measured data, the LAD can be accurately assigned, reflecting the LAD differences in different regions of the canopy (e.g., high LAD at the top of the canopy and low LAD at the edges), making the airflow penetration characteristics in CFD simulation consistent with the real canopy.

[0027] More preferably, such as Figure 2 As shown, the acquisition of a multi-scale digital image set of the target tree through the image acquisition device includes: S1011: At the first observation distance, acquire a global outline image of the target tree using an image acquisition device; S1012: At the second observation distance, an image of the main branch structure of the target tree is acquired by an image acquisition device, wherein the second observation distance is less than the first observation distance; S1013: At the third observation distance, the surface texture image of the target tree is acquired by an image acquisition device, wherein the third observation distance is less than the second observation distance; S1014: A multi-scale data image set is constructed based on the global contour image, the main branch structure image, and the surface texture image.

[0028] In specific implementation, such as Figure 3As shown, the first observation distance focuses on global contour acquisition, specifically capturing the overall shape, crown range, trunk height, and other macroscopic features of large-sized trees, avoiding the loss of global structure due to close-up shooting, and providing an accurate basis for the subsequent construction of the overall framework of the 3D model; the second observation distance specifically acquires the main branch structure, clearly restoring the connection relationship, spatial direction, and thickness distribution of the trunk and main and lateral branches, filling the gaps in the details of branches in the global image, and providing support for the structural accuracy of the initial 3D point cloud; the third observation distance focuses on surface texture acquisition, accurately capturing the cracks, wrinkles, and color gradations of the bark, as well as the shape and arrangement of the leaves, solving the problem of blurred textures in distant images, and laying the foundation for subsequent color mapping and detail restoration.

[0029] The solution in this invention employs a three-tiered acquisition approach—global, branch, and texture detail—to create a comprehensive image set encompassing macroscopic morphology, mesoscopic structure, and microscopic texture. This avoids the shortcomings of single-distance acquisition, which often results in either a lack of global data or insufficient detail. It ensures that subsequent 3D point clouds can accurately reconstruct the overall outline of the tree while precisely representing the branch structure and surface texture, resulting in a final model that is both complete and detailed. The solution in this invention utilizes complementary image data from different observation distances: global images ensure accurate model shape, branch images ensure robust model structure, and texture images ensure lifelike model representation. The combination of these three ensures high-quality data support for every step of the modeling process, from feature recognition and point cloud generation to mesh construction and color mapping, significantly reducing model distortion and detail loss caused by missing data.

[0030] Targeted image data acquisition reduces redundant and invalid information, improving subsequent visual feature point recognition and matching: global images are used to match overall tree spatial features, branch images are used to match structural correlation features, and texture images are used to match surface detail features. This avoids the problems of feature point interference and low matching efficiency in mixed-scale images, improving the speed and accuracy of initial 3D point cloud generation. The specific construction and display effects are as follows... Figure 4 As shown.

[0031] The solution of this invention does not rely on expensive equipment such as laser scanning. It can complete the hierarchical acquisition using only conventional image acquisition devices. While ensuring data quality, it reduces the equipment threshold and acquisition cost for 3D modeling of large-scale trees. Moreover, the acquisition process is simple and easy to operate, and it is suitable for the actual application needs of parks and other scenarios.

[0032] Specifically, before performing meshing on the dense 3D point cloud, the process also includes point cloud structure layering: based on the multi-view semantic segmentation results, the point cloud is classified into trunk point cloud, large branch point cloud, small branch point cloud, and leaf cluster point cloud; the central skeleton structure is extracted by applying a cylindrical fitting algorithm to the trunk and large branch point clouds, and multiple independent leaf cluster units are formed by applying a density clustering algorithm to the leaf cluster point cloud; a hierarchical topological relationship is established from the trunk to the leaf cluster units.

[0033] In the point cloud structure layering, a Bayesian fusion method is used to determine the category of each 3D point: Project the point cloud onto all visible digital images; obtain the semantic segmentation label of each projected point in the corresponding image; calculate the posterior probability based on the label distribution, and assign the point to the category with the highest probability.

[0034] The application of color information from the multi-scale digital image set to the three-dimensional mesh model specifically includes: Map high-resolution textures from close-up images to the branch and trunk regions; Map textures from mid-range images to the canopy region; A multi-level texture system is established, dynamically switching texture resolution based on observation distance. Specifically, point cloud analysis combined with image recognition is used to obtain clearer layers in the canopy space and branches, enhancing the overall visual effect.

[0035] More preferably, after performing meshing processing on the dense three-dimensional point cloud to construct a three-dimensional mesh model of the tree surface, the method further includes: The three-dimensional mesh model is repaired, including filling data holes and smoothing surface noise. Assign wind resistance importance weights to the mesh elements of the three-dimensional mesh model; A mesh simplification algorithm is performed on the three-dimensional mesh model based on the aforementioned wind resistance importance weights; Assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the Gaussian curvature or average curvature of the mesh vertices on the surface of the three-dimensional mesh model, and assigning the wind resistance importance weights according to the calculated curvature values, wherein regions with absolute curvature values ​​higher than a first threshold are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: identifying ridges or edges representing branching in the three-dimensional mesh model, and assigning higher wind resistance importance weights to mesh elements near the identified ridges or edges. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the projected area of ​​each triangular facet of the three-dimensional mesh model on a plane perpendicular to the wind direction for at least one predetermined wind direction; assigning wind resistance importance weights to the corresponding mesh elements based on the projected area, wherein facets that contribute larger projected areas are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: performing a fluid dynamics simulation on the three-dimensional mesh model to obtain wind pressure distribution data on the model surface, and assigning wind resistance importance weights to the mesh elements based on the wind pressure distribution data, wherein regions with wind pressure absolute values ​​or wind pressure gradients higher than a second threshold are assigned higher wind resistance importance weights.

[0036] This invention addresses the mesh voids caused by branch occlusion and blind spots in large-scale tree modeling. By repairing these voids, it ensures the surface continuity and structural integrity of the 3D mesh model, preventing computational errors (such as airflow penetration and distorted wind resistance assessment) caused by voids in subsequent wind resistance analysis and simulation applications. It also eliminates surface burrs and irregular protrusions introduced during point cloud acquisition and meshing, making the mesh model surface more closely resemble the actual tree shape. Furthermore, it reduces noise interference in subsequent weight allocation steps such as curvature calculation and ridgeline identification, ensuring the accuracy of wind resistance importance weight allocation.

[0037] Specifically, by employing four targeted weight allocation methods, core regions that significantly impact wind resistance in the tree model are identified from different dimensions, providing a clear priority basis for subsequent mesh simplification. The technical effects of each method are complementary and have distinct focuses: Regions with high absolute curvature (such as branch bends, bark protrusions, and twig tips) are typically key locations for airflow separation and vortex generation, significantly contributing to wind resistance. By using curvature-related weights, these wind resistance-sensitive regions can be precisely located, avoiding the loss of key wind resistance features during simplification.

[0038] The ridges and edges of branch forks are the main stress-bearing areas on the windward side of trees and the core paths for wind resistance transmission. Assigning high weight to this area allows for the priority preservation of mesh details of key structures such as branch forks, ensuring the accuracy of force transmission path simulation in wind resistance analysis.

[0039] The projected area of ​​the triangular facet on the vertical wind plane is directly related to the magnitude of wind resistance (the larger the projected area, the stronger the wind resistance contribution). This method can be specifically adapted to specific wind direction scenarios (such as the prevailing wind direction in a park), accurately identify key windward areas, and ensure that the simplified model can still accurately reflect the actual wind resistance characteristics of the trees in the diagram under that wind direction.

[0040] By assigning weights to wind pressure distribution data, the actual airflow effect is directly correlated (areas with high absolute wind pressure / gradient are the core stress areas of wind resistance). The weight allocation is more in line with the real physical scenario, and is especially suitable for engineering needs such as high-precision wind resistance assessment and wind-resistant design, thereby enhancing the practical application value of the model.

[0041] The grid simplification algorithm of this invention prioritizes retaining grid elements with high wind resistance importance weights, such as wind resistance sensitive areas and core stress areas, and only simplifies redundant areas with low weights (such as smooth surfaces of tree trunks and large continuous areas on non-windward surfaces). This solves the problem that traditional uniform simplification either loses key features or has insufficient simplification effect, ensuring that the simplified model can still accurately reflect the wind resistance characteristics of trees.

[0042] While retaining key details of wind resistance, the number of mesh elements is reduced, decreasing the model's data volume and facilitating subsequent calculations such as wind resistance simulation and wind resistance performance analysis, thereby improving computational speed. Simultaneously, the model's storage footprint is reduced, making it suitable for practical applications in scenarios such as digital management of trees in industrial parks and wind resistance planning. The simplified model meets the engineering requirements of high-precision wind resistance assessment and wind resistance design, while also possessing lightweight advantages. It can be adapted to scenarios with varying accuracy requirements, such as wind resistance risk early warning for trees in smart parks and wind environment simulation in landscape planning, improving the method's scenario adaptability.

[0043] The solution of this invention provides a more solid model foundation through repair, locks the key wind resistance areas through weight allocation to ensure that the wind resistance-related features of the simplified model are not distorted, simplifies redundant areas in a targeted manner to reduce model complexity and computational cost, and solves the problems of large data volume and low computational efficiency of large-scale tree models; four weight allocation methods can be selected as needed to adapt to wind resistance analysis requirements of different precision and different scenarios.

[0044] More preferably, after establishing the virtual 3D model of the trees in the park, the method further includes: A virtual 3D model of the target tree is obtained, and the virtual 3D model is processed to obtain a structural skeleton, the structural skeleton including multiple nodes and branch components connecting the nodes; The structural skeleton is divided according to the set division logic to obtain the corresponding structural units; The stress points are determined on the structural units of the structural frame according to the set environmental load conditions.

[0045] Specifically, by extracting the structural skeleton containing nodes and branches, non-mechanically relevant information such as surface textures and redundant meshes are stripped away, directly identifying the core mechanics of the tree: the connection relationships, spatial morphology, and dimensional characteristics (such as branch thickness, length, and forking angles) of the trunk, main branches, and lateral branches. This provides a precise, skeleton-level analysis object for subsequent mechanical analysis, avoiding interference from redundant data in the visualization model. The node-branch topology of the structural skeleton corresponds to the mechanical transmission path of the tree (such as wind and snow loads being transferred from leaf nodes to main branches and trunk), providing a clear structural foundation for subsequent load distribution and stress simulation, thus solving the problem of traditional 3D models having complete morphology but unclear mechanical structure.

[0046] By dividing the structure into units according to a set logic (such as the thickness, length, branching position, and material uniformity), the tree skeleton can be decomposed into trunk units, main branch units, lateral branch units, and twig units. This allows subsequent load application and stress calculation to accurately match the mechanical properties of different units (such as trunk units having strong load-bearing capacity and twig units having weak wind resistance), avoiding local stress distortion caused by holistic analysis.

[0047] Structured unit partitioning breaks down complex tree mechanics analysis into collaborative calculations of multiple units. Adaptive mechanical models can be used for different units (e.g., beam unit model for trunk units and rod unit model for branch units), which reduces overall computational complexity and improves the computational accuracy of local key units, thus solving the pain point of complex overall structure and high difficulty in mechanical analysis of large-sized trees.

[0048] By locating stress points based on the set environmental load conditions (such as the prevailing wind direction and rainfall intensity in the park), the mechanical analysis can be more closely aligned with actual application scenarios. For example, in wind resistance analysis, the stress points are concentrated on the branch nodes and crown edges on the windward side, ensuring that the analysis results can directly reflect the stress state of trees in the real environment.

[0049] More preferably, the step of dividing the structural skeleton according to a set partitioning logic to obtain corresponding structural units includes: Identify the main path from the structural skeleton, locate multiple branch points on the main path, and extract candidate branches from each branch point. One or more main branches are selected from the candidate branches based on an importance metric, wherein each selected main branch and its subtree are determined as a corresponding structural unit; wherein the importance metric is to calculate the subtree size of each candidate branch and select one or more candidate branches with the largest subtree size as the main branches.

[0050] The solution of this invention identifies the main trunk path, thus clarifying the core mechanical transmission and load-bearing channels of the tree. This avoids confusion about the primary and secondary relationship between the trunk and branches during the division process, ensuring that subsequent structural units can be developed around the core mechanical framework of the trunk-branch structure. This provides an accurate structural basis for load transfer analysis and load-bearing capacity assessment.

[0051] Branching points are key nodes in the stress distribution of tree structures (such as load diversion and stress concentration locations). By extracting candidate branches at branching points, the splitting positions of structural units are made consistent with the structural boundary points of the tree's natural growth. This not only conforms to the tree's morphology and mechanical characteristics but also avoids the problem of mismatch between unit boundaries and actual stress paths caused by arbitrary splitting, thus improving the rationality of structural units.

[0052] Determining the importance of aligning with tree growth and mechanical properties: Subtree size (which can be understood as the overall scale of a branch and all its derived sub-branches, branches, and leaves, such as volume, weight, and number of nodes) is directly related to two core dimensions: In terms of growth dimension, the larger the subtree, the more likely it is to be the main branch that develops first during the growth of the tree, and it is a core component of the tree's morphology and structure. From a mechanical perspective, the larger the subtree, the greater its windward area and self-weight, and the more significant its impact on the overall tree's stress state (such as load magnitude and transmission path). Therefore, it is a core unit that requires focused attention in mechanical analysis. Thus, using subtree size as a measure of importance ensures that the selection of major branches conforms to natural growth patterns while accurately identifying the key objects for mechanical analysis.

[0053] Focusing on core analysis units: By selecting the candidate branch with the largest subtree size as the main branch, the main branch (core stress unit) and secondary branches (auxiliary stress / redundant unit) are clearly distinguished. This avoids classifying branches with little impact on the overall mechanical state, such as thin branches and weak branches, as structural units on the same level as the main branch. This reduces unnecessary computational redundancy and ensures the analysis accuracy of the core stress unit.

[0054] Each main branch and its subtrees constitute an independent structural unit. This unit naturally forms a complete system of local force-load transfer (such as the main branch unit bearing the self-weight and wind load of its own subtrees and transferring the load to the trunk). This allows subsequent mechanical analysis to model the local characteristics of each unit (such as thickness, length, and material) separately, while clearly simulating the load transfer relationship between units, greatly improving the accuracy of mechanical calculations.

[0055] More preferably, after establishing the virtual 3D model of the trees in the park, the method further includes: Configure material properties for the branch components in the virtual 3D model according to the tree species, including density, Young's modulus and bending strength; Configure hydrodynamic properties for the components representing the tree canopy in the virtual 3D model, including leaf area density and drag force coefficient, and define the tree canopy as a porous medium in the computational fluid dynamics simulation; Configure corresponding detection logic rules for the virtual 3D model, the detection logic rules including: Monitor health indicators and environmental conditions associated with the tree components in the virtual 3D model; In response to determining that the health indicator is below a first threshold and the environmental conditions exceed a second threshold, the state of the tree component is automatically switched from the first state to the second state, which is at a higher risk. Trigger the system response action associated with the second state; Alternatively, the detection logic rules include: The wind pressure acting on the tree branch components in the virtual three-dimensional model is calculated using fluid dynamics simulation. The calculated wind pressure value is compared with a predefined critical fracture strength value for the tree branch assembly; In response to determining that the wind pressure value exceeds the critical fracture strength value, the fracture of the tree branch component is simulated in the virtual three-dimensional model, and the hydrodynamic properties of the three-dimensional model are updated to reflect the geometric changes after fracture.

[0056] Specifically, material properties such as density, Young's modulus, and bending strength are customized according to tree species, so that the mechanical properties of the virtual model are highly matched with those of real trees (such as the difference in Young's modulus between trees and shrubs, and the difference in bending strength between hardwood and softwood). This solves the problem of mechanical simulation distortion caused by uniform properties and provides accurate physical parameter support for subsequent fracture simulation and strength assessment.

[0057] The correspondence between material properties and structural units (such as trunk and main branches) enables mechanical analysis to accurately reflect the differences in load-bearing capacity of different parts, improving the reliability of structural safety assessment. By binding leaf area density and drag coefficient to canopy components and defining them as porous media, the model closely matches the actual obstruction and penetration characteristics of the canopy on airflow (the porous media model can accurately simulate the flow pattern of airflow through the gaps between branches and leaves), avoiding the wind resistance calculation errors caused by traditional solidified canopies.

[0058] By configuring the computational fluid dynamics (CFD) simulation to accurately output the actual wind pressure distribution of the tree canopy and branches, high-precision fluid dynamics data is provided for subsequent wind pressure threshold judgment and fracture simulation, ensuring that the structural safety analysis is consistent with the real wind environment scenario.

[0059] Simultaneously monitor health indicators (such as the degree of branch decay, leaf chlorophyll content, and root vitality) and environmental conditions (such as extreme temperatures, drought / floods, and pest and disease concentrations) to avoid false alarms caused by a single factor (such as not triggering a high-risk state when only health indicators are slightly low but the environment is suitable), and achieve accurate identification of overlapping risks.

[0060] When health indicators fall below the threshold or environmental conditions exceed limits, the system automatically switches to the second state and triggers response actions (such as pushing maintenance work orders, starting irrigation / pest control equipment, and marking risky trees on the park map). Monitoring and management can be completed without manual intervention, solving the pain points of low efficiency and delayed risk detection in traditional park tree management.

[0061] By obtaining the actual wind pressure value of tree branch components through CFD simulation and comparing it with the predefined critical fracture strength (calculated in combination with tree species material properties and branch size), the system can quantitatively assess whether wind load leads to fracture, avoiding the subjectivity and uncertainty of traditional empirical judgments of wind damage risk. Simulating the tree branch fracture process visually presents the damage morphology of trees under extreme wind conditions, providing a visual basis for wind-resistant planning in parks (such as tree planting spacing and windbreak facility layout). After fracture, the system automatically updates fluid dynamic properties (such as the geometric range of the porous media in the canopy and leaf area density), allowing for further simulation of wind resistance changes and wind pressure redistribution in the remaining branches, achieving continuous simulation of fracture-secondary stress, and providing support for subsequent risk superposition assessments.

[0062] More preferably, UAV images are acquired, and a canopy density index is calculated based on the UAV images, wherein the canopy density index is a dimensionless parameter characterizing the density of tree canopies; The calculated canopy density index is input into a transformation mapping function to obtain the output value of the transformation mapping function, and this output value is assigned as the updated leaf area density value to the corresponding canopy component of the virtual 3D model. The transformation mapping function is constructed as follows: measured data pairs of canopy density index and leaf area density of multiple sample trees are obtained; based on the measured data pairs, the mapping function is trained by a regression algorithm, wherein the input of the regression algorithm includes the canopy density index, and the output is the predicted leaf area density value.

[0063] Specifically, drone imagery can completely cover the canopy of large trees from an aerial perspective, avoiding density assessment biases caused by foliage obstruction during ground-based data collection. It accurately captures the overall density and spatial distribution differences of the canopy (such as the density difference between the top and edges of the canopy), providing high-quality basic data for subsequent attribute updates.

[0064] Eliminating the need for manual climbing or close contact with trees reduces the safety risks and operational difficulties of data collection, making it particularly suitable for scenarios involving batch monitoring of multiple trees within a park, thus improving data acquisition efficiency. As a dimensionless parameter, the canopy density index can mask differences in canopy morphology among different tree species and heights, directly focusing on the core characteristic of canopy density. This makes the canopy densities of different trees comparable, providing the possibility for cross-species adaptation of subsequent mapping functions.

[0065] By training the regression model using measured data of canopy density index and leaf area density from multiple sample trees, the mapping function directly reflects the true statistical relationship between the two (e.g., the higher the canopy density, the greater the leaf area density usually is), avoiding the subjectivity and error of traditional manual setting of leaf area density based on experience, and greatly improving the accuracy of attribute configuration.

[0066] The application of regression algorithms provides quantitative support for the mapping relationship. The prediction accuracy can be optimized by adjusting algorithm parameters (such as the coefficients of linear regression and the goodness of fit of nonlinear regression), ensuring that the output leaf area density value is highly matched with the actual canopy state of the trees.

[0067] If the sample data includes measured data of different tree species, the trained mapping function can automatically adapt to the canopy characteristics of different tree species (such as the difference in the correlation between canopy density and leaf area density between conifers and broad-leaved trees), without having to set mapping rules for each tree species separately, thus improving the versatility and scalability of the technical solution and adapting it to the actual scenario of mixed planting of multiple tree species in the park.

[0068] Leaf area density is a core parameter in canopy porous media models, directly affecting the calculation of airflow resistance through the canopy and the simulation results of wind pressure distribution. By dynamically updating leaf area density through the canopy density index, the canopy fluid properties of the virtual model can reflect the actual growth status of the trees in real time (such as the increase or decrease in leaf area density due to seasonal changes and changes in canopy density after pruning), solving the problem of simulation results being out of sync with real-world scenarios caused by traditional fixed leaf area density. Accurate leaf area density values ​​allow CFD simulations to more accurately simulate the canopy's obstruction and drag effects on wind, thereby improving the accuracy of wind pressure calculations for branch components and providing reliable data support for subsequent structural safety analyses (such as fracture risk assessment).

[0069] By periodically collecting drone imagery to update the canopy density index, and then adjusting the leaf area density in real time through a mapping function, the virtual 3D model can dynamically track tree growth changes (such as the annual increase in canopy density) or the effects of human intervention (such as a decrease in canopy density after pruning), achieving dynamic calibration of tree attributes throughout its entire life cycle. The dynamically updated leaf area density ensures that the virtual model remains consistent with real trees, guaranteeing the continued reliability of results from subsequent functions such as health risk warnings and structural safety simulations, providing stable support for the long-term, refined management of trees in the park.

[0070] More preferably, the system response action includes: Based on the calculated bending stress and the set bending strength, the additional support force required to reduce the stress of the structural unit to a safe level is calculated; the bending stress is obtained using a wind force calculation formula, which is: Where F represents the wind force experienced by the corresponding structural unit. Let be the air density, v be the wind speed at the corresponding point, R be the characteristic radius of the corresponding structural unit, and L be the length of the corresponding structural unit. Based on the additional support force, the installation position and orientation of the support component are simulated in the virtual 3D model; a suggested support scheme including the parameters of the support component is output in the user interface; the support component is an elastic cable; The proposed support scheme was determined in the following manner: In the virtual three-dimensional model, the connection path of the elastic cable between the structural unit and the anchor point is determined; The required tension value of the elastic cable is calculated based on the connection path to ensure that its tension component can offset part of the wind load, thereby making the combined bending stress of the structural unit lower than the set bending strength. Based on the stated tension value and cable specifications, determine whether the load is within a safe operating range. If so, output the corresponding recommended support scheme.

[0071] In this embodiment of the invention, a targeted wind force calculation formula is employed. By using key parameters such as air density, actual wind speed, and the characteristic radius and length of the structural unit, the wind force experienced by the structural unit is quantified, avoiding the subjectivity of traditional empirical estimations of wind loads. Based on this wind force, bending stress is further derived, providing a clear physical quantitative basis for stress assessment and ensuring that the goal of reducing stress to a safe level can be accurately achieved. The characteristic radius and length of the structural unit in the formula directly match the unit attributes of the structural skeleton defined earlier, realizing a logical closed loop of load calculation-structural unit-stress assessment, and improving the targeting and accuracy of additional support force calculation.

[0072] Based on the difference between bending stress and set bending strength, the required additional support force is calculated in reverse, clarifying the core force that the elastic cable needs to provide. This avoids blind design of support schemes (such as excessive support leading to resource waste, or insufficient support still posing safety hazards), and ensures that the support force is just right to match the protection requirements of the risk unit.

[0073] Simulating the installation location (such as the key stress points of structural units) and direction (such as the optimal tensile force direction to offset wind load) of elastic cables in a virtual 3D model can intuitively present the spatial relationship between cables, trees, and anchor points, avoiding the cost and risks of on-site trial installation. At the same time, it ensures that the installation plan conforms to the structural characteristics of trees (such as not damaging the trunk and main branches) and the site environmental conditions (such as the anchor points being accessible and the cable path being unobstructed).

[0074] Virtual simulation supports rapid adjustment and optimization of installation schemes (such as adjusting cable connection points and angles), improving the efficiency of support design, especially suitable for scenarios where selecting support points for large trees is difficult and spatial layout requirements are stringent. The virtual model determines the optimal connection path of the cables between structural units and anchor points, ensuring effective force transmission (e.g., straightening the path to reduce force loss) while also considering the feasibility of on-site construction. Based on the path, the required force value is further calculated to ensure that the force component accurately offsets part of the wind load, making the combined bending stress of the structural unit lower than the set bending strength, achieving the dual goals of meeting protection standards and ensuring reasonable cable stress.

[0075] By combining the calculated tension value with the cable specifications (such as material, diameter, and corresponding rated load), it is verified whether the cable is within the safe working range to avoid secondary safety hazards such as breakage or slack caused by cable overload, and to ensure the long-term reliability of the support scheme.

[0076] The verification process, based on the physical properties of the cables and actual stress requirements, ensures that the design meets both tree protection needs and engineering structural safety standards, thus enhancing the scientific rigor and compliance of the solution. The output includes core information such as support component parameters (e.g., cable material, diameter, and length), installation location, direction, and anchorage requirements, providing clear guidance for on-site construction. This resolves the disconnect between virtual design and practical implementation, reducing construction difficulty and communication costs.

[0077] Specifically, taking *Glehnia littoralis* as an example, we obtain data about its branches through digital modeling, such as their diameter at breast height (DBH) and length. We then obtain the density of *Glehnia littoralis* through literature searches and calculate the volume of its branches, V = πr²L. Finally, we calculate the weight of each branch using G = ρ * V. The wind force received by each branch is calculated using the formula F = 1 / 2π²ρ * V² * R * L. Based on the wind force and weight of each branch calculated in the previous step, if the weight and wind force are in the same direction, the resultant force on the branch is maximized. The branch can withstand the torque generated by this maximum resultant force, thus preventing breakage. The bending strength formula σ = M / W, where M is the bending moment and W is the section modulus, yields the maximum torque and maximum stress received by each branch.

[0078] By consulting the literature, the bending strength of the *Gnaphalium affine* is found to be 141 MPa, indicating that branch b may not be able to withstand external forces and will break. By reverse-engineering the formula σ=M / W, the maximum torque that branch b can withstand can be calculated. Then, using the torque formula M=r×F, the maximum external force that branch b can receive is obtained. Based on the previous calculation of the limit state, it can be determined that branch b experiences the maximum external force; therefore, the zipper should provide F. 拉 =F 合 -F bmax Therefore, the minimum force provided by the cable can be determined. Given that the cable can provide a maximum tension of 2000N, if a tension of 523N is required in the vertical direction, then the maximum angle α between the cable and the tree trunk is cosα = F. 拉 / F, calculating α≤75°. However, the purpose of the cables is not only to prevent trees from breaking in strong winds, but also to stabilize the tree structure, improving its stability and reducing the risk of branch breakage in extreme weather. An attempt was made to verify the rationality of the cable scheme through calculation. The cable scheme should use 2T cables, but to ensure the structural stability of the ancient tree, 4T cables were used in the scheme.

[0079] The method in this embodiment of the invention, based on image acquisition, eliminates the need for complex laser scanning equipment. Data acquisition can be completed using only conventional image acquisition devices, reducing equipment costs and operational barriers for 3D modeling of large-scale trees. Through multi-scale image acquisition and a series of point cloud and mesh processing procedures, high-fidelity, detailed, and efficient 3D modeling of large-scale trees is achieved, while also possessing both visualization effects and practical application value.

[0080] Example 2 Please see Figure 5 , Figure 5 This is a schematic diagram of the system for three-dimensional modeling of large-sized trees disclosed in an embodiment of the present invention. Figure 5 As shown, the system for 3D modeling large-scale trees may include: Acquisition module 21: used to acquire a multi-scale digital image set of the target tree through an image acquisition device; the multi-scale digital image set is image data captured by the user at different observation distances; Matching module 22: used to process the multi-scale digital image set to identify and match visual feature points between images, and generate an initial three-dimensional point cloud based on the visual feature points; Generation module 23: used to perform density enhancement processing on the initial three-dimensional point cloud to generate a dense three-dimensional point cloud that characterizes the surface morphology of trees; Processing module 24: used to perform meshing processing on the dense three-dimensional point cloud to construct a three-dimensional mesh model of the tree surface; Model building module 25: used to apply color information from the multi-scale digital image set to the three-dimensional mesh model to build a virtual three-dimensional model of the park trees.

[0081] The method in this embodiment of the invention, based on image acquisition, eliminates the need for complex laser scanning equipment. Data acquisition can be completed using only conventional image acquisition devices, reducing equipment costs and operational barriers for 3D modeling of large-scale trees. Through multi-scale image acquisition and a series of point cloud and mesh processing procedures, high-fidelity, detailed, and efficient 3D modeling of large-scale trees is achieved, while also possessing both visualization effects and practical application value.

[0082] Example 3 Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device can be a computer, a server, etc. Of course, in certain cases, it can also be a mobile phone, tablet computer, monitoring terminal, or other smart device, as well as an image acquisition device with processing capabilities. Figure 6 As shown, the electronic device may include: Memory 510 storing executable program code; Processor 520 coupled to memory 510; The processor 520 calls the executable program code stored in the memory 510 to execute some or all of the steps in the method for three-dimensional modeling of large-sized trees in Embodiment 1.

[0083] This invention discloses a computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps in the method for three-dimensional modeling of large-sized trees in Embodiment 1.

[0084] This invention also discloses a computer program product, wherein when the computer program product is run on a computer, the computer performs some or all of the steps in the method for three-dimensional modeling of large-scale trees in Embodiment 1.

[0085] This invention also discloses an application publishing platform, which is used to publish computer program products. When the computer program products are run on a computer, the computer performs some or all of the steps in the method for three-dimensional modeling of large-scale trees in Embodiment 1.

[0086] In various embodiments of the present invention, it should be understood that the sequence number of each process does not necessarily 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 the present invention.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they can be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional units in the various embodiments of the present invention 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.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests to cause a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the methods described in the various embodiments of the present invention.

[0090] In the embodiments provided by this invention, it should be understood that B corresponding to A means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0091] Those skilled in the art will understand that some or all of the steps in the various methods of the embodiments described can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0092] The foregoing has provided a detailed description of the method, system, electronic device, and storage medium for three-dimensional modeling of large-sized trees disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for three-dimensional modeling of large-sized trees, characterized in that, include: A multi-scale digital image set of the target tree is acquired through an image acquisition device; the multi-scale digital image set consists of image data taken by the user at different observation distances. The multi-scale digital image set is processed to identify and match visual feature points between images, and an initial three-dimensional point cloud is generated based on the visual feature points. The initial three-dimensional point cloud is subjected to density enhancement processing to generate a dense three-dimensional point cloud that characterizes the surface morphology of trees. The dense three-dimensional point cloud is meshed to construct a three-dimensional mesh model of the tree surface; Color information from the multi-scale digital image set is applied to the three-dimensional mesh model to create a virtual three-dimensional model of the trees in the park.

2. The method for three-dimensional modeling of large-sized trees as described in claim 1, characterized in that, The acquisition of a multi-scale digital image set of the target tree through an image acquisition device includes: At the first observation distance, a global outline image of the target tree is acquired using an image acquisition device; At the second observation distance, an image of the main branch structure of the target tree is acquired by an image acquisition device, wherein the second observation distance is less than the first observation distance; At a third observation distance, a surface texture image of the target tree is acquired using an image acquisition device, wherein the third observation distance is less than the second observation distance; The global contour image, the main branch structure image, and the surface texture image constitute a multi-scale data image set.

3. The method for three-dimensional modeling of large-sized trees as described in claim 1, characterized in that, After performing meshing processing on the dense 3D point cloud to construct a 3D mesh model of the tree surface, the method further includes: The three-dimensional mesh model is repaired, including filling data holes and smoothing surface noise. Assign wind resistance importance weights to the mesh elements of the three-dimensional mesh model; A mesh simplification algorithm is performed on the three-dimensional mesh model based on the aforementioned wind resistance importance weights; Assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the Gaussian curvature or average curvature of the mesh vertices on the surface of the three-dimensional mesh model, and assigning the wind resistance importance weights according to the calculated curvature values, wherein regions with absolute curvature values ​​higher than a first threshold are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: identifying ridges or edges representing branching in the three-dimensional mesh model, and assigning higher wind resistance importance weights to mesh elements near the identified ridges or edges. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: calculating the projected area of ​​each triangular facet of the three-dimensional mesh model on a plane perpendicular to the wind direction for at least one predetermined wind direction; assigning wind resistance importance weights to the corresponding mesh elements based on the projected area, wherein facets that contribute larger projected areas are assigned higher wind resistance importance weights. Alternatively, assigning wind resistance importance weights to the mesh elements of the three-dimensional mesh model includes: performing a fluid dynamics simulation on the three-dimensional mesh model to obtain wind pressure distribution data on the model surface, and assigning wind resistance importance weights to the mesh elements based on the wind pressure distribution data, wherein regions with wind pressure absolute values ​​or wind pressure gradients higher than a second threshold are assigned higher wind resistance importance weights.

4. The method for three-dimensional modeling of large-sized trees as described in claim 1, characterized in that, After establishing the virtual 3D model of the trees in the park, the following is also included: A virtual 3D model of the target tree is obtained, and the virtual 3D model is processed to obtain a structural skeleton, the structural skeleton including multiple nodes and branch components connecting the nodes; The structural skeleton is divided according to the set division logic to obtain the corresponding structural units; The stress points are determined on the structural units of the structural frame according to the set environmental load conditions.

5. The method for three-dimensional modeling of large-sized trees as described in claim 4, characterized in that, The step of dividing the structural skeleton according to a set partitioning logic to obtain corresponding structural units includes: Identify the main path from the structural skeleton, locate multiple branch points on the main path, and extract candidate branches from each branch point. One or more main branches are selected from the candidate branches based on an importance metric, wherein each selected main branch and its subtree are determined as a corresponding structural unit; wherein the importance metric is to calculate the subtree size of each candidate branch and select one or more candidate branches with the largest subtree size as the main branches.

6. The method for three-dimensional modeling of large-sized trees as described in claim 4, characterized in that, After establishing the virtual 3D model of the trees in the park, the following is also included: Configure material properties for the branch components in the virtual 3D model according to the tree species, including density, Young's modulus and bending strength; Configure hydrodynamic properties for the components representing the tree canopy in the virtual 3D model, including leaf area density and drag force coefficient, and define the tree canopy as a porous medium in the computational fluid dynamics simulation; Configure corresponding detection logic rules for the virtual 3D model, the detection logic rules including: Monitor health indicators and environmental conditions associated with the tree components in the virtual 3D model; In response to determining that the health indicator is below a first threshold and the environmental conditions exceed a second threshold, the state of the tree component is automatically switched from the first state to the second state, which is at a higher risk. Trigger the system response action associated with the second state; Alternatively, the detection logic rules include: The wind pressure acting on the tree branch components in the virtual three-dimensional model is calculated using fluid dynamics simulation. The calculated wind pressure value is compared with a predefined critical fracture strength value for the tree branch assembly; In response to determining that the wind pressure value exceeds the critical fracture strength value, the fracture of the tree branch component is simulated in the virtual three-dimensional model, and the hydrodynamic properties of the three-dimensional model are updated to reflect the geometric changes after fracture.

7. The method for three-dimensional modeling of large-sized trees as described in claim 6, characterized in that, The three-dimensional modeling method also includes: Acquire drone images and calculate the canopy density index based on the drone images, wherein the canopy density index is a dimensionless parameter characterizing the density of tree canopies; The calculated canopy density index is input into a transformation mapping function to obtain the output value of the transformation mapping function, and this output value is assigned as the updated leaf area density value to the corresponding canopy component of the virtual 3D model. The transformation mapping function is constructed as follows: measured data pairs of canopy density index and leaf area density of multiple sample trees are obtained; based on the measured data pairs, the mapping function is trained by a regression algorithm, wherein the input of the regression algorithm includes the canopy density index, and the output is the predicted leaf area density value.

8. The method for three-dimensional modeling of large-sized trees as described in claim 6, characterized in that, The system response actions include: Based on the calculated bending stress and the set bending strength, the additional support force required to reduce the stress of the structural unit to a safe level is calculated; the bending stress is obtained using a wind force calculation formula, which is: Where F represents the wind force experienced by the corresponding structural unit. Let be the air density, v be the wind speed at the corresponding point, R be the characteristic radius of the corresponding structural unit, and L be the length of the corresponding structural unit. Based on the additional support force, the installation position and orientation of the support component are simulated in the virtual 3D model; a suggested support scheme including the parameters of the support component is output in the user interface; the support component is an elastic cable; The proposed support scheme was determined in the following manner: In the virtual three-dimensional model, the connection path of the elastic cable between the structural unit and the anchor point is determined; The required tension value of the elastic cable is calculated based on the connection path to ensure that its tension component can offset part of the wind load, thereby making the combined bending stress of the structural unit lower than the set bending strength. Based on the stated tension value and cable specifications, determine whether the load is within a safe operating range. If so, output the corresponding recommended support scheme.

9. A system for three-dimensional modeling of large-sized trees, characterized in that, include: Acquisition module: used to acquire a multi-scale digital image set of the target tree through an image acquisition device; the multi-scale digital image set is image data taken by the user at different observation distances; Matching module: used to process the multi-scale digital image set to identify and match visual feature points between images, and generate an initial three-dimensional point cloud based on the visual feature points; Generation module: used to perform density enhancement processing on the initial three-dimensional point cloud to generate a dense three-dimensional point cloud that characterizes the surface morphology of trees; Processing module: used to perform meshing processing on the dense 3D point cloud to construct a 3D mesh model of the tree surface; Model building module: used to apply color information from the multi-scale digital image set to the three-dimensional mesh model to build a virtual three-dimensional model of the park trees.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program causes a computer to perform the method for three-dimensional modeling of large-sized trees as described in any one of claims 1 to 8.

Images