A cross-source fusion modeling method and system for 3D point cloud models of forest trees

Abstract

This invention provides a cross-source fusion modeling method and system for a 3D point cloud model of trees. The method includes: acquiring first information based on a camera device mounted on a UAV, the first information including canopy layer image data above a preset height of the trees and understory image data below the preset height of the trees; performing image processing based on the first information, and preprocessing the obtained point cloud data to obtain preprocessed point cloud data; determining canopy layer tree data and understory tree data based on the preprocessed point cloud data to obtain the location data of each tree; performing initial registration of the location data of each tree based on an improved Delaunay triangulation algorithm, and performing a second registration on the initial registration result through an optimization function to obtain a 3D point cloud model of the trees. This invention significantly improves the efficiency and accuracy of 3D tree modeling.

Description

Technical Field

[0001] This invention relates to the field of point cloud data processing technology, and more specifically, to a cross-source fusion modeling method and system for a three-dimensional point cloud model of forest trees. Background Technology

[0002] Against the backdrop of global climate change and ecological degradation, the accurate measurement of forest structural parameters, as an important carbon sink resource, is crucial for ecological protection and sustainable management. Traditional vegetation monitoring methods mainly rely on ground surveys and remote sensing technologies. However, ground surveys are time-consuming, labor-intensive, and costly, making large-scale application difficult. While remote sensing technology can cover large areas, its resolution and accuracy are insufficient to meet the needs of precise measurement. In recent years, the development of 3D modeling technology, especially the application of lidar technology, has provided new avenues for vegetation structure monitoring. However, its high cost and limitations in complex terrain still restrict its widespread application.

[0003] The emergence of drone-based oblique photography and panoramic imaging technologies has provided an efficient and low-cost solution for forest 3D modeling. Drones can quickly acquire high-resolution images of the canopy layer, while panoramic cameras can capture detailed information about the understory. However, effectively fusing data from these two different perspectives and constructing a high-precision forest 3D model through accurate point cloud processing and registration techniques remains a pressing technical challenge. Summary of the Invention

[0004] The purpose of this invention is to provide a cross-source fusion modeling method and system for three-dimensional point cloud models of forest trees, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0005] Firstly, this application provides a cross-source fusion modeling method for a 3D point cloud model of forest trees, including:

[0006] The first information is obtained by using a camera device mounted on a drone. The first information includes image data of the canopy layer above a preset height of the trees and image data of the understory layer below a preset height of the trees.

[0007] Image processing is performed based on the first information, and the processed point cloud data is preprocessed to obtain preprocessed point cloud data.

[0008] Based on the preprocessed point cloud data, the canopy layer tree data and the understory tree data are determined to obtain the location data of each tree.

[0009] The location data of each tree is initially registered based on the improved Delaunay triangulation algorithm, and then the initial registration result is registered a second time through an optimization function to obtain a three-dimensional point cloud model of the trees.

[0010] Secondly, this application also provides a cross-source fusion modeling system for a three-dimensional point cloud model of forest trees, characterized in that it includes:

[0011] The acquisition unit is used to acquire first information based on the camera equipment mounted on the drone. The first information includes canopy layer image data above a preset height of the trees and understory layer image data below a preset height of the trees.

[0012] The processing unit is configured to perform image processing based on the first information and preprocess the processed point cloud data to obtain preprocessed point cloud data.

[0013] The analysis unit is used to determine the canopy layer tree data and the understory tree data based on the preprocessed point cloud data, and to obtain the location data of each tree.

[0014] The registration unit is used to perform initial registration of the location data of each tree based on the improved Delaunay triangulation algorithm, and to perform a second registration on the initial registration result through an optimization function to obtain a three-dimensional point cloud model of the trees.

[0015] The beneficial effects of this invention are as follows:

[0016] This invention acquires image data of the canopy and understory layers using drones and panoramic cameras, respectively. Combined with an improved Delaunay triangulation algorithm and further point-to-surface registration, it achieves high-precision construction of a 3D point cloud model of forest trees. This method effectively solves the problems of low data acquisition efficiency, high cost, and insufficient accuracy in traditional techniques, significantly improving the completeness and accuracy of 3D forest modeling. Furthermore, by optimizing the point cloud preprocessing and registration process, this invention further improves the efficiency and reliability of model construction, providing a more scientific and efficient technical means for forest resource management and carbon sequestration capacity assessment, with broad application prospects and significant ecological importance.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1This is a schematic diagram of the cross-source fusion modeling method for the three-dimensional point cloud model of forest trees described in this embodiment of the invention;

[0020] Figure 2 This is a schematic diagram of the cross-source fusion modeling system for the three-dimensional point cloud model of forest trees described in this embodiment of the invention.

[0021] The diagram is labeled as follows: 701, acquisition unit; 702, processing unit; 703, analysis unit; 704, registration unit. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0023] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] Example 1:

[0025] like Figure 1 As shown in the figure, this embodiment provides a cross-source fusion modeling method for a three-dimensional point cloud model of forest trees.

[0026] See Figure 1 The figure shows that the method includes steps S1, S2, S3 and S4.

[0027] Step S1: Obtain first information based on the camera equipment mounted on the drone. The first information includes image data of the canopy layer above the preset height of the trees and image data of the understory layer below the preset height of the trees.

[0028] Understandably, this step involves completing a full data collection within a specific timeframe. The canopy layer image data primarily contains information about the upper structure of the forest, such as tree crowns, branches, leaves, and vegetation cover. The understory image data mainly reflects the structures near the forest floor, including tree trunks, shrubs, and herbaceous plants. To acquire this data, canopy layer images are primarily collected by professional drones equipped with photographic equipment. The drones use continuous shooting mode for aerial photography, setting flight altitude and flight path to ensure an overlap rate of over 70% for each area, thereby improving the accuracy of image matching. Understory image data is acquired using panoramic cameras. The panoramic cameras collect data along an S-shaped trajectory, recording 360-degree panoramic video including GPS information, extracting still images from the video, and simultaneously writing GPS information into the corresponding images. The combination of these data provides a foundation for comprehensive monitoring of the forest ecosystem and provides rich information support for subsequent 3D modeling and analysis.

[0029] Step S2: Perform image processing based on the first information, and preprocess the processed point cloud data to obtain preprocessed point cloud data;

[0030] It is understood that the preprocessed point cloud data includes dense point clouds of the canopy layer and dense point clouds of the understory layer, ensuring the integrity and high resolution of the preprocessed point cloud data. For example, canopy layer image data can generate point clouds reflecting tree height and canopy structure, while understory layer image data can provide three-dimensional information on ground vegetation and topography. This two-dimensional to three-dimensional conversion not only more intuitively displays the spatial structure of the forest, but also provides accurate three-dimensional coordinate information for subsequent point cloud processing and analysis. The generated preprocessed point cloud data retains the original characteristics of the forest environment, providing rich details for further preprocessing and feature extraction, enabling the system to more comprehensively analyze various elements in the forest ecosystem and their interrelationships. It is understood that in this step, a series of optimization processes are performed on the preprocessed point cloud data through a point cloud preprocessing model. Specific steps include sequentially performing cropping, filtering, ground point separation, normal vector calculation, and correction operations on the preprocessed point cloud data. For example, cropping can remove irrelevant areas, focusing on the research target; filtering can remove noise points and outliers, making the point cloud smoother and more accurate; ground point separation can clearly distinguish vegetation from the ground, facilitating subsequent vegetation structure analysis. Normal vector calculation and correction can further optimize the geometric features of the point cloud, ensuring the consistency of point clouds from different data sources in spatial coordinates. These preprocessing steps not only improve the accuracy of the point cloud data but also enhance its operability, making subsequent tree location extraction and 3D modeling more efficient and accurate. In this step, step S2 includes steps S21, S22, S23, and S24.

[0031] Step S21: The canopy layer image data and the understory layer image data are calibrated to generate point cloud data;

[0032] Understandably, in this step, by processing canopy and understory image data to generate preprocessed point cloud data, the system can transform two-dimensional image information into three-dimensional spatial data, thus laying the foundation for three-dimensional forest modeling. Specifically, after completing a full data acquisition within a specific time frame, the canopy images captured by the drone-mounted aerial camera and the understory images captured by the panoramic camera are imported into Ag i soft Metashape software to generate sparse point clouds. Based on this, the preprocessed point cloud data is further generated.

[0033] Step S22: The point cloud data is cropped and denoised to obtain denoised point cloud data;

[0034] Understandably, by cropping the preprocessed point cloud data in this step, irrelevant areas can be effectively removed, focusing the data on the target region and reducing computational load in subsequent processing, thereby improving data processing efficiency. Specifically, the cropping region is first defined in the point cloud processing software. After loading the point cloud data, cropping parameters (such as cropping boundaries and directions) are set, and the cropping operation is executed. This allows for precise extraction of point cloud data within the target region, yielding the cropped result. This process not only helps remove redundant background information but also ensures the relevance and accuracy of subsequent analysis, providing an optimized data foundation for subsequent point cloud filtering, feature extraction, and other operations.

[0035] Step S23: Based on the cloth simulation filtering algorithm, the denoised point cloud data is separated to obtain ground points containing only vegetation and / or non-vegetation features;

[0036] Understandably, the cloth simulation filtering algorithm separates ground points from non-ground points in the denoised point cloud data, thereby distinguishing vegetation point clouds from ground and other non-vegetation feature point clouds. This process provides an accurate data foundation for subsequent 3D modeling and analysis. Specifically, the denoised result is first processed for noise reduction and then flipped. Next, the cloth grid is initialized and its size is set, placing the cloth above the highest point of the point cloud in the denoised result. Subsequently, all point cloud data in the denoised result are projected onto the same horizontal plane, and the nearest point and its elevation for each cloth point are recorded. By calculating the displacement of movable grid points under gravity and comparing it with the elevation of the nearest point, points that meet the criteria are set as immovable. This process is iterated until the height change of all points is less than the set cloth algorithm threshold. Finally, by comparing the height difference between the point cloud and the cloth points, ground points are distinguished and extracted, thus obtaining clean vegetation and non-vegetation feature point cloud data.

[0037] Step S24: Calculate the normal vector for each ground point using a neighborhood plane fitting algorithm, and correct the Z-axis of the corresponding ground point based on the calculated normal vector to obtain preprocessed point cloud data.

[0038] Understandably, calculating the normal vector of ground points using a neighborhood plane fitting algorithm can provide accurate directional information for subsequent point cloud correction and analysis, thereby enhancing the geometric features and spatial consistency of the point cloud data. Specifically, firstly, a neighborhood is defined for each ground point in the point cloud, determined based on a fixed radius or a fixed number of neighboring points. Next, all points within the neighborhood of each ground point are collected, and plane fitting is performed using these points, employing the least squares method to find the plane best suited for these points. The normal vector of this plane is the normal vector of the ground point. Correcting the ground points based on the normal vector in this step ensures the spatial consistency of the point cloud data, providing a more accurate foundation for subsequent modeling and analysis. Specifically, firstly, the spatial orientation of each point in the ground is determined using the normal vector. Next, Z-axis correction is applied to each ground point to ensure that the normal vector of all points is consistent with the set reference direction or gravity direction. This correction process involves fine-tuning the position of each point in the point cloud data to eliminate deviations caused by measurement errors or sensor biases, ultimately yielding the preprocessed result.

[0039] Step S3: Based on the preprocessed point cloud data, determine the canopy layer tree data and the understory tree data to obtain the location data of each tree;

[0040] Understandably, this step extracts canopy point clouds and understory point clouds from the preprocessing results. The canopy point cloud contains point cloud information of the tree crowns, branches, and leaves in the upper forest layer, while the understory point cloud contains point cloud information of the tree trunks, shrubs, and other vegetation in the lower forest layer. This provides crucial information for multi-dimensional analysis of forest ecosystems. For example, canopy point cloud data can reflect the crown width, height, and growth status of trees, while understory point cloud data can reveal the location of tree trunks, diameter at breast height (DBH), and the distribution of understory vegetation. This layered extraction method not only helps to accurately locate trees but also provides important data support for forest resource management, biomass estimation, and ecological protection. In this step, step S3 includes steps S31, S32, S33, and S34.

[0041] Step S31: Convert the preprocessed point cloud data into a digital surface model and a digital elevation model;

[0042] Understandably, in this step, the rasterization tool is first selected and the resolution parameters of the raster are set. Then, the rasterization operation is performed to convert the point cloud data into a regular grid system, resulting in multiple grid cells. By extracting the elevation value and surface data of each grid cell and performing interpolation processing, a digital elevation model and a digital surface model can be generated.

[0043] Step S32: Input the digital surface model and digital elevation model into GIS software for subtraction, and extract the crown width from the vegetation height information obtained by subtraction to obtain crown width height information;

[0044] Understandably, in this step, by subtracting the digital elevation model from the digital surface model, information reflecting the canopy height can be obtained. Specifically, for each pixel in both the digital surface model and the digital elevation model, the difference between the elevation values ​​in the digital surface model and the digital elevation model is calculated to obtain the canopy height information. This process not only reveals the vertical structure of vegetation but also provides crucial data support for subsequent vegetation biomass estimation, forest resource management, and ecological protection. The canopy height model is a key component in 3D forest modeling; combined with trunk location information, it provides a foundation for a comprehensive analysis of forest structure.

[0045] Step S33: Based on the watershed segmentation algorithm, identify and segment the crown height information to obtain the location of trees in the canopy layer;

[0046] It is understandable that in this step, the canopy height model is processed by the watershed segmentation algorithm, which enables the identification and segmentation of the canopy layer, thereby accurately locating the position of a single tree. This provides important basic data support for subsequent forest structure analysis, tree growth monitoring, and forest resource management. In this step, step S33 includes steps S331, S332, and S333.

[0047] Step S331: By inverting the crown height model, a three-dimensional model containing multiple basins is obtained, and a seed point is preset, wherein a preset point is used as the smallest perforation and the perforation is used as the seed point.

[0048] Understandably, this step, by inverting the crown height model, transforms it into a 3D region model containing multiple "basins." This transformation is a crucial preprocessing step for the watershed segmentation algorithm. In the inverted model, the original crown tops (high points) become low-lying "basins," while the low-lying areas between crowns become "ridges." This transformation allows the watershed algorithm to simulate the convergence of water flow, forming watersheds between the "basins" and thus effectively segmenting the crown. The inverted 3D region model provides a suitable terrain foundation for subsequent segmentation algorithms, enabling them to more accurately identify and locate individual trees. By setting minimum points as seed points in each inverted "basin," the watershed segmentation algorithm is provided with starting positions. Seed points are the algorithm's initial reference points, located at the lowest point of each crown (i.e., the lowest point of the inverted basin). These seed points, as the starting positions for segmentation, guide the algorithm to simulate the convergence of water flow upwards from the lowest point, thereby forming segmentation boundaries between crowns.

[0049] Step S332: Starting from all the seed points, gradually simulate watering the three-dimensional model. Stop watering when the water level rises to the edge of the basin. Take each basin area that is in a submerged state at this time as a segmented area.

[0050] Understandably, in this step, starting from all the set seed points, a "watering simulation" is performed step-by-step on the inverted 3D region model. This process is the core of the watershed segmentation algorithm. By simulating the process of water flowing upwards from the seed points (preset points), the algorithm can dynamically identify the boundaries between tree canopies. Specifically, the water flow gradually rises along the height gradient until it encounters a higher region (i.e., the boundary between tree canopies) and stops. When the simulated water level rises to the edge of each "basin," the watering operation stops. At this point, each "basin region" in a submerged state is defined as an independent segmented region. This process marks the completion of the watershed segmentation algorithm, with each segmented region corresponding to a single tree canopy. This simulation process can effectively distinguish different canopy regions, ensuring that each canopy is accurately segmented. Through step-by-step watering simulation, the algorithm can adaptively handle complex terrain changes, thereby providing accurate segmentation results for subsequent tree location extraction and forest structure analysis.

[0051] Step S333: Assign each segmented region to the location of a tree, and determine the location of each tree in the canopy layer based on the center point of the segmented region.

[0052] Understandably, in this step, by statistically analyzing all segmented regions and determining the center point of each region, the location of each tree within the canopy can be precisely pinpointed. Each segmented region corresponds to an independent canopy, and its boundaries are determined by the watershed segmentation algorithm. By calculating the center point (e.g., geometric center or mass center) of each segmented region, the location information of the trees can be obtained. This process not only achieves precise positioning of individual trees in the forest but also provides crucial foundational data support for subsequent forest resource management, tree growth monitoring, and ecological research.

[0053] Step S34: Fit the understory trees into a circle based on the least squares method, and use the center of the fitted circular image as the location of the understory trees.

[0054] Understandably, this step involves extracting point cloud slices containing tree diameter at breast height (DBH) and fitting circles using the least squares method, using the center coordinates of the fitted circle as the tree location. The canopy layer tree location point set reflects the top structure of the trees, while the understory tree location point set provides information on the trunk and understory vegetation. By separating these two datasets, different layers of forest structure can be processed and analyzed more accurately, providing more comprehensive data support for subsequent 3D modeling and forest resource management. In this step, step S34 includes steps S341 and S342.

[0055] Step S341: According to the preset range threshold, extract point cloud slices containing the diameter at breast height of understory trees from the preprocessed point cloud data.

[0056] Understandably, this step extracts point cloud slices containing the diameter at breast height (DBH) of understory trees from the preprocessed point cloud data using a preset range threshold, significantly reducing the data volume and facilitating subsequent processing. This step is specifically targeted at the DBH of understory trees, ensuring that subsequent analysis focuses on these specific targets and avoids interference from irrelevant data.

[0057] Step S342: Fit the point cloud in the point cloud slice into a circle based on the least squares method, and use the diameter of the circle as the diameter at breast height of the tree, and the coordinates of the center of the circle as the position coordinates of the understory tree to obtain the position of the understory tree.

[0058] Understandably, this step utilizes the center coordinates of a fitted circle to determine the location coordinates of the understory trees, achieving accurate tree positioning. This step automates the extraction of diameter at breast height (DBH) and location information from point cloud data, reducing manual intervention and improving efficiency.

[0059] Step S4: Based on the improved Delaunay triangulation algorithm, the location data of each tree is initially registered, and the initial registration result is registered a second time through an optimization function to obtain a three-dimensional point cloud model of the trees.

[0060] Understandably, in this step, the improved Delaunay triangulation algorithm is used to perform initial registration of the forest location point cloud data, achieving preliminary alignment of the canopy and understory point cloud data. The resulting initial registration provides the foundation for subsequent fine registration, ensuring good alignment of point clouds from different data sources at forest locations, and providing reliable support for further forest 3D modeling and analysis. In this step, step S4 includes...

[0061] Step S41: Based on the location data of each tree, construct a combination of at least one canopy layer tree location point and a combination of at least one understory tree location point, each combination containing three non-collinear tree location points;

[0062] Understandably, in this step, by selecting at least two sets of tree location points (each set containing three non-collinear tree location points) from the tree canopy and understory tree location point sets respectively, and calculating the distance between each pair of tree location points within each set, distance constraints can be provided for the subsequent construction of the Delaunay triangulation network.

[0063] Step S42: Calculate the distance between any two tree locations within each combination, and use the minimum calculated distance across all combinations as the constraint threshold.

[0064] Understandably, this step, by calculating the minimum distance between every pair of tree locations in all combinations and selecting a global minimum as a constraint threshold, effectively filters out non-tree points (such as shrubs or noise points), thereby improving the accuracy and reliability of point cloud registration. This process provides important reference for subsequent triangulation construction and point cloud alignment, ensuring accurate matching of point cloud data at different levels in terms of tree locations.

[0065] Step S43: Combine all possible non-collinear points of the canopy tree location points and the understory tree location points into candidate point groups;

[0066] Understandably, in this step, by exhaustively exploring all possible combinations of three non-collinear points in both the canopy and understory tree location point sets, candidate point sets are generated, providing a comprehensive point set foundation for subsequent Delaunay triangulation construction and point cloud registration. This process ensures that all possible triangle combinations are considered, thus providing sufficient candidates for subsequent distance constraints and similarity assessments. By exhaustively exploring all possible combinations, the geometric relationships between tree location points can be captured more comprehensively, ensuring high-precision point cloud registration and 3D modeling.

[0067] Step S44: Determine whether the distance between any two non-collinear points in each candidate point group is greater than the constraint threshold. If it is greater, then obtain the triangle corresponding to each candidate point group by connecting the points in each candidate point group pairwise.

[0068] Understandably, in this step, by determining whether the distance between any two non-collinear points within each candidate point group is greater than a set constraint threshold, point groups that do not meet the tree spacing characteristics can be effectively filtered out, thereby improving the accuracy and reliability of point cloud registration. Specifically, for each candidate point group, if the distance between any two non-collinear points within it is greater than the constraint threshold, then these point groups are considered to satisfy the geometric characteristics of tree location, and their internal points can be connected pairwise to form corresponding triangles. This process provides precise triangle combinations for subsequent Delaunay triangulation construction and point cloud alignment, ensuring accurate matching of point cloud data at different levels in terms of tree location. Through this distance constraint, non-tree points (such as shrubs or noise points) can be more effectively identified and filtered out, thereby improving the accuracy and efficiency of point cloud registration.

[0069] Step S45: Connect all candidate point groups that are greater than the constraint threshold to form a triangular network, thereby obtaining the Delaunay triangular network of the canopy layer and the Delaunay triangular network of the understory layer.

[0070] Understandably, in this step, for all candidate point groups that satisfy the constraints, the points within each group are connected pairwise to form triangles, thus constructing Delaunay triangulation networks for the canopy and understory layers respectively. This process is based on the principle of Delaunay triangulation, generating triangulation networks with optimal geometric properties by connecting points that satisfy distance constraints. The canopy layer Delaunay triangulation network reflects the spatial distribution of the tree top structure, while the understory Delaunay triangulation network reveals the layout of tree trunks and understory vegetation. By constructing these two levels of triangulation networks respectively, the geometric features of the forest structure can be represented more intuitively, while providing an accurate geometric framework for subsequent point cloud registration and 3D modeling.

[0071] Step S46: By comparing the similarity of the side lengths and angles of the triangles in the canopy layer Delaunay triangulation network and the understory Delaunay triangulation network, matching triangles are identified. Based on all matching triangles, the position data of each tree is aligned, and non-tree points are filtered out by side length and preset angle constraints to obtain the initial registration result.

[0072] Understandably, in this step, matching triangles are identified by comparing the similarity of side lengths and angles of triangles in the canopy and understory Delaunay triangulation networks. This process, based on the similarity of geometric features, enables preliminary alignment of canopy and understory point cloud data at tree locations. Specifically, firstly, the side lengths and angles of all possible combinations of three non-collinear points in the canopy and understory point sets are calculated; then, constraint thresholds for side lengths and angles are set, retaining only point groups that satisfy these constraints; based on these point groups, a Delaunay triangulation network is constructed, and matching triangles are identified by comparing the similarity of their side lengths and angles. Based on these matching triangles, the canopy and understory point cloud data are aligned at tree locations, thus achieving preliminary fusion of point cloud data from different levels. Simultaneously, by setting constraints on side lengths and angles, non-tree points (such as shrubs or noise points) that do not conform to the geometric characteristics of trees are filtered out, further improving the purity and registration accuracy of the point cloud data. The final initial registration result is a set of point cloud data of the canopy and understory that have been aligned at the location of the trees.

[0073] It is understandable that step S47 is included after step S46 in this process.

[0074] Step S47: Perform further registration on the initial registration result to obtain a three-dimensional point cloud model of the forest.

[0075] Understandably, in this step, the initial registration result is further optimized based on point-to-surface registration to obtain a high-precision 3D point cloud model of the forest. Specifically, the further registration involves projecting the initially registered point cloud data onto the initial registration result (as a reference surface) and ensuring precise spatial alignment of the point cloud data through an optimization function that minimizes the projection error. This process includes: calculating the vertical distance from each point in the initially registered point cloud data to the initial registration result (reference surface) and constructing a minimum error equation to minimize the sum of squares of these distances. By iteratively adjusting the rotation and translation parameters of the initially registered point cloud data, the alignment accuracy of the point cloud is gradually optimized until the error reaches a set threshold or converges. After optimization, the initially registered point cloud data is precisely aligned with the initial registration result in spatial position, ultimately yielding a 3D point cloud model of the forest. This process effectively eliminates minor deviations that may exist in the initial registration stage, further improving the accuracy and consistency of the point cloud data. The minimum error equation is specifically expressed as:

[0076]

[0077] Where dmin represents the minimum projection error obtained through the optimization process, that is, the minimum projection error achieved by adjusting the point cloud data during the fine registration of the point cloud; argmin M , representing the search for the transformation parameter M that minimizes the error function; M includes a rotation matrix R and a translation vector t, used to describe the rigid body transformation from the initially registered point cloud data to the initially registered result; m is the total number of points in the point cloud; i is an index variable used for iteration and enumeration; a point qi in the initially registered point cloud data is set and mapped to the initially registered result; the d i This represents the point closest to point qi in the initial registration result; the normal vector of the tangent plane containing the point in the initial registration result is n. i The (Mq) i -d i )n i This represents the error from point to surface, specifically the point cloud data after initial registration and transformation, specifically the point cloud Mq after initial registration. i The distance to the tangent plane of the initial registration result, where the physical meaning of the tangent plane represents the point cloud Mq after initial registration. i Projected onto normal vector n i The distance in the direction of the minimum error equation; the physical meaning of the minimum error equation is that it represents the distance from the transformed points of the initially registered point cloud data to the tangent plane of the initial registration result, that is, the distance of the initially registered point cloud data projected onto the normal vector direction. By minimizing this distance, the accurate representation of the point cloud data in terms of the location, shape, and structure of trees can be ensured, thereby providing high-quality data support for forest 3D modeling, vegetation analysis, and resource management.

[0078] Example 2:

[0079] like Figure 2 As shown, this embodiment provides a cross-source fusion modeling system for a 3D point cloud model of forest trees. See [link to documentation]. Figure 2 The system includes an acquisition unit 701, a processing unit 702, an evaluation unit 703, and a construction unit 704.

[0080] The acquisition unit 701 is used to acquire first information based on the camera equipment mounted on the drone. The first information includes canopy layer image data above a preset height of the trees and understory layer image data below a preset height of the trees.

[0081] Processing unit 702 is used to perform image processing based on the first information, and to preprocess the processed point cloud data to obtain preprocessed point cloud data.

[0082] Analysis unit 703 is used to determine the canopy layer tree data and the understory tree data based on the preprocessed point cloud data, and obtain the location data of each tree;

[0083] The registration unit 704 is used to perform initial registration of the location data of each tree based on the improved Delaunay triangulation algorithm, and to perform a second registration of the initial registration result through an optimization function to obtain a three-dimensional point cloud model of the trees.

[0084] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0085] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A cross-source fusion modeling method for a 3D point cloud model of forest trees, characterized in that, include: The first information is obtained by using a camera device mounted on a drone. The first information includes image data of the canopy layer above a preset height of the trees and image data of the understory layer below a preset height of the trees. Image processing is performed based on the first information, and the processed point cloud data is preprocessed to obtain preprocessed point cloud data. Based on the preprocessed point cloud data, the canopy layer tree data and the understory tree data are determined to obtain the location data of each tree. The location data of each tree is initially registered based on the improved Delaunay triangulation algorithm, and the initial registration result is then registered a second time using an optimization function to obtain a three-dimensional point cloud model of the trees. The process involves initial registration of the location data for each tree based on an improved Delaunay triangulation algorithm, followed by a second registration of the initial registration results using an optimization function. This includes: Based on the location data of each tree, construct at least one combination of tree location points in the canopy layer and at least one combination of tree location points in the understory layer. Each combination contains three non-collinear tree location points. Calculate the distance between each pair of tree locations within each combination, and use the minimum calculated distance across all combinations as the constraint threshold; Combine all possible non-collinear points of canopy trees and understory trees into candidate point groups; Determine whether the distance between any two non-collinear points in each candidate point group is greater than the constraint threshold. If it is greater, then obtain the triangle corresponding to each candidate point group by connecting the points in each candidate point group pairwise. Connect all candidate point groups that are greater than the constraint threshold to form a triangular network, resulting in the Delaunay triangular network for the canopy layer and the Delaunay triangular network for the understory layer. By comparing the similarity of the side lengths and angles of the triangles in the canopy triangulation and the understory triangulation, matching triangles are identified. Based on all matching triangles, the position data of each tree is aligned, and non-tree points are filtered out by side length and preset angle constraints to obtain the initial registration result. The initial registration results are further registered to obtain a three-dimensional point cloud model of the forest. The determination of canopy forest data and understory forest data based on the preprocessed point cloud data includes: The preprocessed point cloud data is converted into a digital surface model and a digital elevation model; The digital surface model and digital elevation model are input into GIS software for subtraction, and the vegetation height information obtained by subtraction is used to extract the crown width to obtain the crown width height information; The crown height information is identified and segmented based on the watershed segmentation algorithm to obtain the location of trees in the canopy layer; The understory trees are fitted into a circle using the least squares method, and the center of the fitted circular image is used as the location of the understory trees.

2. The cross-source fusion modeling method for a three-dimensional point cloud model of forest trees according to claim 1, characterized in that, Image processing is performed based on the first information, and the resulting point cloud data is preprocessed, including: The canopy layer image data and the understory layer image data are calibrated to generate point cloud data; The point cloud data is cropped and denoised to obtain denoised point cloud data. The denoised point cloud data is separated based on the cloth simulation filtering algorithm to obtain ground points containing only vegetation and / or non-vegetation features. The normal vector of each ground point is calculated by a neighborhood plane fitting algorithm, and the Z-axis of the corresponding ground point is corrected based on the calculated normal vector to obtain preprocessed point cloud data.

3. The cross-source fusion modeling method for a three-dimensional point cloud model of forest trees according to claim 1, characterized in that, The crown height information is identified and segmented based on the watershed segmentation algorithm, including: By inverting the crown height model, a three-dimensional model containing multiple basins is obtained, and a seed point is preset, wherein a preset point is used as the smallest perforation and this perforation is used as the seed point; Starting from all the seed points, the three-dimensional model is gradually simulated with water. When the water level rises to the edge of the basin, water injection is stopped, and each basin area that is now submerged is taken as a segmented area. Each segmented region corresponds to the location of a tree, and the location of each tree in the canopy layer is determined based on the center point of the segmented region.

4. The cross-source fusion modeling method for a three-dimensional point cloud model of forest trees according to claim 1, characterized in that, The understory trees are fitted into a circle using the least squares method, and the center of the fitted circular image is used as the location of the understory trees, including: Based on the preset range threshold, point cloud slices containing the diameter at breast height of understory trees are extracted from the preprocessed point cloud data. The point cloud slices are fitted into circles using the least squares method, and the diameter of the circle is used as the diameter at breast height of the trees. The coordinates of the center of the circle are used as the position coordinates of the understory trees to obtain the position of the understory trees.

5. A cross-source fusion modeling system for a 3D point cloud model of forest trees, characterized in that, include: The acquisition unit is used to acquire first information based on the camera equipment mounted on the drone. The first information includes canopy layer image data above a preset height of the trees and understory layer image data below a preset height of the trees. The processing unit is configured to perform image processing based on the first information and preprocess the processed point cloud data to obtain preprocessed point cloud data. The analysis unit is used to determine the canopy layer tree data and the understory tree data based on the preprocessed point cloud data, and to obtain the location data of each tree. The registration unit is used to perform initial registration of the location data of each tree based on the improved Delaunay triangulation algorithm, and to perform a second registration on the initial registration result through an optimization function to obtain a three-dimensional point cloud model of the trees. The process involves initial registration of the location data for each tree based on an improved Delaunay triangulation algorithm, followed by a second registration of the initial registration results using an optimization function. This includes: Based on the location data of each tree, construct at least one combination of tree location points in the canopy layer and at least one combination of tree location points in the understory layer. Each combination contains three non-collinear tree location points. Calculate the distance between each pair of tree locations within each combination, and use the minimum calculated distance across all combinations as the constraint threshold; Combine all possible non-collinear points of canopy trees and understory trees into candidate point groups; Determine whether the distance between any two non-collinear points in each candidate point group is greater than the constraint threshold. If it is greater, then obtain the triangle corresponding to each candidate point group by connecting the points in each candidate point group pairwise. Connect all candidate point groups that are greater than the constraint threshold to form a triangular network, resulting in the Delaunay triangular network for the canopy layer and the Delaunay triangular network for the understory layer. By comparing the similarity of the side lengths and angles of the triangles in the canopy triangulation and the understory triangulation, matching triangles are identified. Based on all matching triangles, the position data of each tree is aligned, and non-tree points are filtered out by side length and preset angle constraints to obtain the initial registration result. The initial registration results are further registered to obtain a three-dimensional point cloud model of the forest. The analysis unit includes: The first analysis subunit is used to convert the preprocessed point cloud data into a digital surface model and a digital elevation model. The second analysis subunit is used to input the digital surface model and the digital elevation model into GIS software for subtraction, and extract the crown width from the vegetation height information obtained by subtraction to obtain crown width height information; The third analysis subunit is used to identify and segment the crown height information based on the watershed segmentation algorithm to obtain the location of trees in the canopy layer; The fourth analysis subunit is used to fit the understory trees into a circle based on the least squares method, and to use the center of the fitted circular image as the location of the understory trees.

6. The cross-source fusion modeling system for three-dimensional point cloud models of forest trees according to claim 5, characterized in that, The processing unit includes: The first processing subunit is used to calibrate the canopy layer image data and the understory layer image data to generate point cloud data; The second processing subunit is used to crop and denoise the point cloud data to obtain denoised point cloud data. The third processing subunit is used to separate the denoised point cloud data based on the cloth simulation filtering algorithm to obtain ground points containing only vegetation and / or non-vegetation features. The fourth processing subunit is used to calculate the normal vector of each ground point using a neighborhood plane fitting algorithm, and correct the Z-axis of the corresponding ground point based on the calculated normal vector to obtain preprocessed point cloud data.

7. The cross-source fusion modeling system for three-dimensional point cloud models of forest trees according to claim 5, characterized in that, The third analysis subunit includes: The fifth analysis subunit is used to obtain a three-dimensional model containing multiple basins by inverting the crown height model, and to preset a seed point, wherein the preset point is used as the smallest perforation and the perforation is used as the seed point; The sixth analysis subunit is used to perform watering simulation on the three-dimensional model step by step, starting from all the seed points. When the water level rises to the edge line of the basin, the watering is stopped, and each basin area that is in a state of flooding at this time is taken as a segmented area. The seventh analysis subunit is used to assign each segmented region to the location of a tree and determine the location of each tree in the canopy layer based on the center point of the segmented region.

8. The cross-source fusion modeling system for a three-dimensional point cloud model of forest trees according to claim 5, characterized in that, The fourth analysis subunit includes: The eighth analysis subunit is used to extract point cloud slices containing the diameter at breast height of understory trees from the preprocessed point cloud data according to a preset range threshold. The ninth analysis subunit is used to fit the point cloud in the point cloud slice into a circle based on the least squares method, and use the diameter of the circle as the diameter at breast height of the tree, and the coordinates of the center of the circle as the position coordinates of the understory trees to obtain the position of the understory trees.

Images