A three-dimensional leaf dynamic reconstruction method fusing branch structure recognition and random disturbance

Abstract

A three-dimensional tree leaf dynamic reconstruction method fusing branch structure recognition and random disturbance belongs to the technical field of tree three-dimensional reconstruction, forest digitization and visual simulation and computer program in forestry science research, and is based on a real trunk model, automatically determines a leaf growth area through recognition of spatial geometric features (local branch thickness), and simulates spatial distribution of leaves in a natural state in combination with a multi-angle disturbance strategy, realizes regulation and control of different leaf type design and leaf area index (LAI) through adjustment of leaf shape parameters and quantity control, and realizes automatic transition from three-dimensional branch structure to a complete tree model.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of tree three-dimensional reconstruction, forest digitization and visualization simulation and computer program in forestry science research, and relates to a three-dimensional tree leaf dynamic reconstruction method combining branch structure recognition and random disturbance, and relates to tree leaf model reconstruction technology, tree model library construction technology, tree model retrieval and sample plot visualization simulation technology. BACKGROUND

[0002] Efficient and accurate reconstruction of forests is of great significance in forest management simulation, forest ecological process simulation (such as photosynthetic radiation transmission simulation) and forest digital twin system construction. With the development of computer graphics and visualization simulation technology, forest structure analysis, interactive management and growth simulation are gradually realized in the digital operation of virtual forest environment, which not only reduces the cost of real trial and error, but also improves the intuitive understanding of forest structure and ecological mechanism.

[0003] Currently, the three-dimensional reconstruction technology of tree branches has been relatively mature, and has been widely used in many fields such as scientific research, ecological monitoring, film and television animation and games. However, compared with branch modeling, the modeling technology based on real structure characteristics to adaptively construct and restore the natural distribution state of tree leaves still faces many challenges. The existing mainstream tree leaf modeling methods, such as SpeedTree, OnyxTree and other rule growth modeling software, have low modeling efficiency, and the leaf density distribution tends to be uniform, making it difficult to restore the heterogeneity structure of real trees. The modeling method based on L-system is mostly at the theoretical level and is difficult to interface with the branch model obtained by reality scanning and reconstruction. The point cloud attachment or mesh scattering method generally lacks spatial logic and attachment structure basis, and only realizes random attachment at the surface level. SUMMARY

[0004] The present application provides a three-dimensional tree leaf dynamic reconstruction method combining branch structure recognition and random disturbance to solve the technical problems of the prior art.

[0005] A three-dimensional tree leaf dynamic reconstruction method combining branch structure recognition and random disturbance, comprising the following steps: based on a real tree trunk model, automatically determining the leaf growth area by recognizing its spatial geometric features (local branch thickness), and simulating the spatial distribution of leaves in natural state by combining multi-angle disturbance strategy, realizing the control of different leaf type design and leaf area index (LAI) by adjusting leaf shape parameters and quantity control, and realizing the automatic transition from three-dimensional branch structure to complete tree model.

[0006] The advantages of the present application are: the accurate positioning and natural layout of the leaf model construction position in the three-dimensional modeling of trees are realized. The method has the advantages of high automation, strong adaptability, high simulation degree and fast reconstruction speed, and the precision fully meets the application requirements of primary forestry. At the same time, the present application can test the leaf modeling effect in the sample plot level stand modeling, and has good universality and application promotion value.

[0007] The three-dimensional leaf dynamic reconstruction method fusing branch structure recognition and random disturbance provided by the present application can automatically determine the leaf growth area by accurately recognizing the spatial geometric features (such as the local branch thickness) of the tree branches, and simulate the spatial distribution of the leaves in the natural state by combining the multi-angle disturbance strategy. Specifically, the present application improves the authenticity, scientificity and reconstruction efficiency of the three-dimensional leaf reconstruction. In the leaf parameter regulation part, the number of leaf parameters, morphological parameters and leaf inclination and displacement parameters can be adjusted independently to realize high-fidelity reconstruction of different tree species leaves under various leaf area indices (LAI), realize the automatic transition from three-dimensional branch structure to complete tree model, and significantly improve the naturalness and fidelity of tree modeling. In addition, the complete tree model after leaf reconstruction is used to carry out sample plot level stand three-dimensional reconstruction and photosynthetic radiation simulation, which further demonstrates the fidelity, practicality and scalability of the reconstructed leaves. The method not only improves the accurate positioning and natural layout of the leaf model construction position, but also significantly improves the reconstruction speed while ensuring high simulation degree. The technical effect fully meets the accuracy and speed requirements of three-dimensional tree modeling in primary forestry applications, and has good universality and application promotion value. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. As shown in the drawings:

[0009] Figure 1 The fusion structure recognition and random disturbance three-dimensional leaf dynamic reconstruction algorithm construction and test technical roadmap.

[0010] Figure 2 The adaptive leaf model reconstruction technical roadmap.

[0011] Fig. 3(a) is a rectangular leaf structure diagram of a leaf model.

[0012] Fig. 3(b) is a rectangular leaf structure diagram of a leaf model, which is the compressed form of Fig. 3(a).

[0013] Figure 3(c) is a diagram of the leaf model of the elliptical leaf structure.

[0014] Figure 3(d) is a diagram of the leaf model of the elliptical leaf structure after compression of Figure 3(c).

[0015] Figure 4 Visualization of some tree species.

[0016] Figure 5 Technical roadmap of digital plot construction.

[0017] Figure 6 Test area, green cover area is 144 small sample plot (20m x 20m) location map.

[0018] Figure 7(a) is a tree point cloud diagram after branch and leaf separation of Qinghai spruce leaf reconstruction details.

[0019] Figure 7(b) is a branch model diagram after reconstruction of Qinghai spruce leaf reconstruction details.

[0020] Figure 7(c) is a complete tree model (already colored) diagram after reconstruction of Qinghai spruce leaf reconstruction details.

[0021] Figure 8 Leaf reconstruction visualization interface diagram.

[0022] Figure 9 Complete Qinghai spruce model diagram (partial example).

[0023] Figure 10 Three-dimensional image (based on UE5) diagram of the constructed large plot (240m x 240m).

[0024] Figure 11 (a) is one of the reconstructed digital small sample plots.

[0025] Figure 11 (b) is the second of the reconstructed digital small sample plots.

[0026] Figure 11 (c) is the third of the reconstructed digital small sample plots.

[0027] Figure 11 (d) is one of the photosynthetic active radiation of the small sample plot.

[0028] Figure 11 (e) is the second of the photosynthetic active radiation of the small sample plot.

[0029] Figure 11 (f) is the third of the photosynthetic active radiation of the small sample plot. DETAILED DESCRIPTION

[0030] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0031] Embodiment 1: as Figure 1 , Figure 2 , Fig. 3(a), Fig. 3(b), Fig. 3(c), Fig. 3(d), Figure 4 , Figure 5 , Figure 6 , Fig. 7(a), Fig. 7(b), Fig. 7(c), Figure 8 , Figure 9 , Figure 10 , Figure 11 (a), Figure 11 (a), Figure 11 (b), Figure 11 (c), Figure 11 (d), Figure 11 (e), and Figure 11 (f) show a three-dimensional leaf dynamic reconstruction method (Adaptive Leaf Synthesis Algorithm, ALSA) that combines branch structure recognition and random disturbance to solve the problem that current tree leaf reconstruction algorithms are difficult to realize natural attachment of tree leaves after processing measured tree trunk data (such as point cloud or three-dimensional grid). This method can effectively improve the authenticity of leaf distribution, improve the algorithm efficiency and automation level in the modeling process, and make up for the shortcomings of existing technologies in leaf shape restoration and structure logic modeling. It provides key technical support for high-precision three-dimensional reconstruction of digital sample plots, forest stands and even regional forests.

[0032] First, a new three-dimensional leaf dynamic reconstruction algorithm (Adaptive Leaf Synthesis Algorithm, ALSA) that combines branch structure recognition and random disturbance is used to reconstruct the leaves of the branch model, and then a digital sample plot is constructed based on the complete reconstructed tree model and the newly constructed attribute similarity matching algorithm, so as to realize the further expansion of the reconstructed complete tree model to the digital sample plot reconstruction and test the leaf modeling effect.

[0033] A three-dimensional leaf dynamic reconstruction method that combines branch structure recognition and random disturbance, as Figure 1As shown, the Adaptive Leaf Synthesis Algorithm (ALSA) tree leaf reconstruction algorithm is performed on the tree branch model, a tree model library is established, attribute matching and model screening are performed, (a) the Euclidean distance method defines a similarity function (NEDS), (b) the tree model library is matched with the single tree attribute library, and (c) the optimal matching model is screened, through digital DEM data, tree model and ground material data, Unreal Engine blueprint coding, terrain reconstruction, tree generation, and the establishment of a digital sample plot and leaf reconstruction algorithm test.

[0034] Specifically, first, the original tree branch obj model is imported into the newly constructed leaf reconstruction algorithm (Adaptive Leaf Synthesis Algorithm, ALSA), and adaptive leaf addition is performed on the branch model. The parameters involved in this part need to be determined according to the type of tree constructed and the leaf area index. The complete tree model reconstructed is imported into the tree model library for standby. Then attribute matching and model screening are performed. This step uses the Euclidean distance method to define a similarity function (NEDS) to further match the tree model library with the single tree attribute library. According to the matching result, the optimal matching model is screened, and finally the screened tree model is exported. Combined with the digital DEM data obtained by airborne radar and the ground material data obtained by camera, digital terrain reconstruction is realized by using Unreal Engine blueprint coding, and tree generation is realized by reading the corresponding position of the tree model on the digital terrain, thereby finally realizing the construction of a digital sample plot. Finally, the leaf reconstruction algorithm is tested by observing the reconstruction effect of the digital sample plot and simulating the photosynthetic radiation of the leaves.

[0035] A three-dimensional tree leaf dynamic reconstruction method combining branch structure recognition and random disturbance, the operation steps are as follows:

[0036] 1. Construct a three-dimensional tree leaf dynamic reconstruction algorithm combining branch structure recognition and random disturbance

[0037] Construct a leaf reconstruction algorithm (Adaptive Leaf Synthesis Algorithm, ALSA) with tree branch model as input model, its technical roadmap is as shown Figure 2 The tree branch model is imported into the Adapti ve Leaf Synthesis A lgori thm (ALSA) leaf reconstruction algorithm, the neighborhood search is carried out through the KD tree spatial data structure to carry out model vertex neighborhood search and coarse and fine threshold judgment, and the reconstructed branches are determined. Leaf layout is performed based on parameter-controlled leaf model construction (rectangular, elliptical leaves), random offset + Z-axis rotation + adaptive tilt are adopted, leaf attachment and model coloring and visualization are realized, and tree model library (attributes, models) is obtained through batch processing and model output.

[0038] Specifically, the tree branch model is introduced into the Adaptive Leaf Synthesis Algorithm (ALSA) leaf reconstruction algorithm. The algorithm first uses the KD tree spatial data structure to carry out neighborhood search. Then, the branch of the branch model is judged by thickness threshold, and the neighborhood search of the model vertex is carried out, so that the reconstructed tree branch and the specific reconstruction position are determined. Then, the algorithm carries out leaf model construction based on parameter control (rectangular, elliptical leaves), which mainly defines the morphological structure parameters and quantity parameters of the leaves. Then, the leaf layout is carried out, mainly through random offset, Z-axis rotation and adaptive tilt to make the leaf layout more randomized and naturalized. After the leaf reconstruction position is determined, the leaf attachment is carried out. Finally, the coloring and visualization of the model are realized by defining the leaf color and branch color and designing a visualization interface, in which the model angle can be freely rotated for convenient observation. The whole process can realize batch processing of leaf addition and model output, and finally the model is imported into the tree model library (attributes, models), wherein the attributes include the parameters of the tree model such as diameter at breast height, tree height, crown width, crown area and crown volume.

[0039] (1) Leaf shape design and parameter control

[0040] The present application provides two morphological design schemes in the construction of leaf models: one is a rectangular structure leaf modeling method, which is constructed by four vertices and spliced by two triangular facets, has the advantages of simple structure and high calculation efficiency, and is suitable for coniferous tree species or fast modeling scenarios with low leaf shape accuracy requirements. The other is an approximate elliptical leaf modeling method, which adopts a fan-shaped triangular facet construction method, and is distributed around the center point by setting multiple boundary vertices to simulate the natural contour of real broad-leaved tree species leaves. The number of segments can be adjusted as needed to control the accuracy, and it is suitable for ecological simulation and visualization reconstruction tasks that require high restoration degree. As shown in FIGS. 3(a), 3(b), 3(c) and 3(d). The leaf shape of FIG. 3(a) is a long strip (rectangle), which is composed of two triangular facets. This leaf shape has a simple structure and can simulate most lanceolate leaves and some broad-leaved tree leaves; FIG. 3(b) is the result of stretching the leaf shape of FIG. 3(a); the leaf shape of FIG. 3(c) is elliptical, which is composed of several fan-shaped facets, and each facet edge has a certain curvature. This configuration can simulate most broad-leaved tree leaves; and FIG. 3(d) is the result of stretching the leaf shape of FIG. 3(c).

[0041] (2) Leaf attachment and layout design

[0042] First, the vertices of the branch model are searched for neighborhood, and the local thickness characteristics of each vertex are estimated. By setting the thickness threshold, the vertex area representing the thin branch is identified and screened out as the potential attachment point of the leaf. On this basis, the automatic generation and random placement of leaves are carried out, which specifically includes: a slight spatial offset at the attachment point position, a random angle rotation around the Z axis or other axes, and adaptive inclination angle control combined with the growth direction. This process effectively simulates the natural growth distribution state of real tree leaves. Finally, the generated leaf model is integrated with the original tree trunk model to build a complete three-dimensional tree structure Figure 4 The reconstruction effect of some tree species is shown), and the attribute information (such as tree height, diameter at breast height, crown width, crown area and crown volume) of each tree is stored in the tree model database together with its three-dimensional model, forming a standardized callable model library.

[0043] (3) Visual window design

[0044] After completing the reconstruction of the tree leaf model, the model is colored, and then a visualization window for viewing the modeling results is written using the visualization library (open3D) provided with Python, making it easy to view the modeling effect.

[0045] 2. Digital plot construction

[0046] In order to further test the reconstruction and display effect of the tree leaf model reconstructed by the tree leaf reconstruction algorithm in the plot-level stand condition, a multi-attribute-based similarity matching algorithm is constructed, which uses the Normalized Euclidean Distance Similarity (NEDS) function to calculate the attribute difference value (diameter at breast height, tree height, crown characteristics) of each candidate model and the target tree, and selects the model with the smallest NEDS for model matching and replacement, so as to ensure that the model is highly similar to the corresponding tree geometry. The matching result is used for the automatic construction of the subsequent digital stand, and the similarity function calculation formula is as follows:

[0047]

[0048] In the formula, H d , W d , D d , A d , V d are the height, crown width, diameter at breast height, crown area and crown volume of the target tree, respectively; H m , W m , D m , A m , V mare the corresponding values of a certain tree model in the model library; w1-w5 are weight coefficients (w1=0.4, w2=0.2, w3=0.1, w4=0.2, w5=0.1).

[0049] The DEM data is imported into the visualization rendering engine to generate the sample plot terrain, and the terrain is material mapped. Then the attribute data of each tree in the sample plot and the matched three-dimensional tree model library data are read by using blueprint programming, and the three-dimensional model of the tree is automatically generated at the corresponding coordinate position to realize the construction and display of the digital sample plot, as shown in the flowchart of Figure 5 The single-tree attribute data and the tree model selected from the tree model library are read by attribute, programmed by UE5 blueprint, and matched by model, position, and attribute to realize the digital sample plot. The airborne radar point cloud data is extracted from the stand DEM, textured with the ground texture, and referenced with the real stand to realize the digital sample plot.

[0050] Specifically, first, the stand DEM is extracted based on the airborne radar point cloud data, the DEM file is converted into a.png file recognizable by UE and imported into UE5, the ground texture material is imported on the terrain to realize the ground texture mapping. At the same time, the single-tree attribute data is read by attribute, and the tree model selected from the tree model library is read by model, position, and attribute matching, so that the same tree three-dimensional model as the corresponding position in the real world can be generated at the corresponding position. Through the above steps, the digital sample plot is constructed and referenced with the real stand.

[0051] Example 2: as Figure 1 , Figure 2 , Fig. 3(a), Fig. 3(b), Fig. 3(c), Fig. 3(d), Figure 4 , Figure 5 , Figure 6 , Fig. 7(a), Fig. 7(b), Fig. 7(c), Figure 8 , Figure 9 , Figure 10 , Figure 11 (a), Figure 11 (a), Figure 11 (b), Figure 11 (c), Figure 11 (d), Figure 11 (e), and Figure 11 (f) show a three-dimensional tree leaf dynamic reconstruction method (Adaptive Leaf Synthesis Algorithm, ALSA) fusing branch structure recognition and random disturbance, and a digital Qinghai spruce leaf reconstruction and sample plot reconstruction example in Qinghai Qilian Mountain National Forest Park.

[0052] 1. Branch model data acquisition

[0053] A Picea crassifolia forest plot (240 m x 240 m, containing 144 subplots, 20 m x 20 m) in the SiGou monitoring station of Qilian Mountain, Qinghai Province, China, was selected as the implementation object. Figure 6 In this example, airborne and ground handheld LiDAR data were collected and fused. After resampling, denoising and normalization, high-quality point clouds were obtained. The point cloud data of each tree were extracted by single-tree segmentation, and its diameter at breast height, tree height, crown shape, and spatial position were obtained. Subsequently, non-missing point clouds were selected for branch and leaf separation, and three-dimensional reconstruction of point clouds to branches and trunks was completed.

[0054] 2. Tree leaf reconstruction

[0055] The reconstructed branch and trunk model was imported into the tree leaf three-dimensional reconstruction algorithm (ALSA) proposed in this application for tree leaf modeling. First, the parameter threshold was set to determine the tree leaf reconstruction tree branch, and then the parameters (rectangular leaf, length, width, and number of leaves) were set according to the actual Picea crassifolia leaf shape. Then the leaf model was set to rotate around the Z-axis (angle range 355° to 360°) and then tilt about 60° or so, and superimposed with a small random displacement, so that the generated Picea crassifolia leaf has the characteristics of natural random distribution ( Figure 7c ), based on the visualization function of the algorithm, the quality of the generated model can be freely rotated to view ( Figure 8 ). All the reconstructed tree trunks and leaves were combined to form a complete three-dimensional tree model ( Figure 9 ), and stored in the tree model library, while recording the attribute parameters of each model.

[0056] Digital forest reconstruction and tree leaf reconstruction effect test

[0057] Based on the extracted single-tree attribute data, the most matching three-dimensional tree model was automatically selected from the model library using the NEDS similarity function, realizing the precise replacement of the model. Combined with the DEM data obtained by airborne LiDAR and the shp file of 144 20 m x 20 m subplots, the terrain was cut and reconstructed, and the terrain and texture were imported into UE5. Through the blueprint, the batch loading and light rendering of the tree model were realized, and the digital modeling of large-scale forest was completed ( Figure 10 ). In the range of the subplots, further three-dimensional radiative transfer model was used to render and simulate the absorbed photosynthetically active radiation of the reconstruction results ( Figure 11 (a)、 Figure 11 (b)、 Figure 11 (c)、 Figure 11 (d)、 Figure 11 (e) and Figure 11(f)). Through the above digital plot construction process, the application of the leaf adaptive reconstruction algorithm in the present application at the stand scale is further verified. The tree leaf model generated by the algorithm is natural in spatial distribution and has good visualization effect. At the same time, the three-dimensional radiation transfer simulation results also reflect the potential application value of the algorithm in the direction of ecological process modeling and structure-radiation interaction research, and have broad application prospects in the fields of digital twin forest, ecological simulation and forestry visualization.

[0058] As Figure 11 (a)、 Figure 11 (b)、 Figure 11 (c)、 Figure 11 (d)、 Figure 11 (e) and Figure 11 (f) are shown, the yellow arrow represents the direction of light incidence, and the difference in color of different tree leaves reflects the spatial heterogeneity of light in the plot, which verifies the naturalness and accuracy of tree leaf reconstruction, and embodies the good visualization effect and application potential.

[0059] From the results, the three-dimensional leaf dynamic reconstruction method proposed in the present application has remarkable effect, which not only improves the realism of leaf attachment position and the simulation degree of the whole model, but also realizes the natural difference expression of tree shape in plot modeling. The method supports efficient, fast and accurate automatic modeling and visualization display under the condition of part of coniferous and broad-leaved tree species and structure, which can effectively improve the detail performance and immersion of forest structure simulation. Further digital plot construction effectively verifies the display effect and practical application significance of the leaf reconstruction algorithm in large-scale modeling, and shows good scalability and engineering promotion potential.

[0060] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for dynamic reconstruction of three-dimensional leaves integrating branch structure recognition and random perturbation, characterized in that, The process includes the following steps: Based on a real tree trunk and branch model, the system automatically determines the leaf growth area by identifying its spatial geometric features, and simulates the spatial distribution of leaves under natural conditions by combining a multi-angle perturbation strategy. By adjusting leaf morphology parameters and quantity control, the system achieves the regulation of different leaf shape designs and leaf area index (LAI), thus realizing an automated transition from a three-dimensional trunk and branch structure to a complete tree model. First, the original tree branch model is imported into the newly constructed leaf reconstruction algorithm. Adaptive leaf addition is then performed on the branch model; the parameters for this step are determined based on the tree species and leaf area index. The reconstructed tree model is then imported into a tree model library for later use. Next, attribute matching and model selection are performed. This step uses the Euclidean distance method to define a similarity function NEDS to further match the tree model library with the individual tree attribute library. Based on the matching results, the optimal matching model is selected. Finally, the selected tree model is exported. Combining digital DEM data acquired by airborne radar and ground material data acquired by the camera, Unreal Engine Blueprint encoding is used to realize digital terrain reconstruction and to generate trees by reading the tree model's position corresponding to the digital terrain, thus ultimately constructing a digital plot. Finally, the leaf reconstruction algorithm is validated by observing the digital plot reconstruction effect and conducting photosynthetic radiation simulations on the leaves. The tree branch and trunk model is imported into the leaf reconstruction algorithm. First, a neighborhood search is performed using a KD tree spatial data structure. Then, the thickness threshold of the branches in the branch and trunk model is judged, and a neighborhood search of the model vertices is performed to clarify the reconstructed branches and specific reconstruction locations. Next, rectangular and elliptical leaves are constructed based on parameter control. This part mainly defines the morphological and structural parameters and quantity parameters of the leaves. Then, the leaf layout is carried out, mainly through random offset, Z-axis rotation, and adaptive tilt to make the leaf layout more random and natural. After clarifying the leaf reconstruction location, the leaves are attached. Finally, the model is colored and visualized by defining the leaf and branch colors and designing a visualization interface. The model angle can be freely rotated in the visualization interface to realize batch processing of adding leaves and model output. Finally, the model is imported into the tree model library's attributes and models. The attributes include parameters such as the tree model's diameter at breast height (DBH), tree height, crown width, crown area, and crown volume.

2. The method for dynamic reconstruction of three-dimensional leaves integrating branch structure recognition and random perturbation according to claim 1, characterized in that, Tree models selected from the tree model library and the single tree attribute library are read through attribute reading and programmed using UE5 blueprints. Then, digital plots are created by matching models, locations, and attributes. Airborne radar point cloud data is extracted from forest stand DEMs, textured with ground, and referenced to real forest stands to create digital plots.

3. The method for dynamic reconstruction of three-dimensional leaves integrating branch structure recognition and random perturbation according to claim 2, characterized in that, First, forest stand DEMs are extracted based on airborne radar point cloud data. The DEM files are then converted into .png format files that can be recognized by UE and imported into UE5. Ground texture materials are imported onto the terrain to achieve ground texture mapping. The attribute data of individual trees is read using UE5 blueprint programming functions. Tree models selected from the tree model library are read, and model, location, and attribute matching is performed to generate a 3D tree model that is identical to the corresponding location in the real world at the corresponding location.

4. The method for dynamic reconstruction of three-dimensional leaves integrating branch structure recognition and random perturbation according to claim 1, characterized in that, The matching results are used for the subsequent automatic construction of digitized forest stands, and the similarity function is calculated as follows: ; In the formula: H d , W d , D d , A d , V d These are the target tree's height, crown width, diameter at breast height (DBH), crown area, and crown volume; H m , W m , D m , A m , V m These are the corresponding values ​​for a specific tree model in the model library; w 1 ~ w 5 These are the weighting coefficients. w 1 =0.4, w 2 =0.2, w 3 =0.1, w 4 =0.2, w 5 =0.1.

Images