An under-forest sapling extraction method, device, medium and product based on a UAV laser radar

Abstract

The application discloses an under-forest sapling extraction method and device based on a UAV laser radar, a medium and a product, relates to the technical field of laser radar regional point cloud processing, and comprises the following steps: preprocessing regional point clouds collected by a UAV laser radar to obtain normalized regional point clouds and constructing a canopy height model, determining point sets of each tree and unassigned point sets; determining a plurality of under-forest sapling points in the point sets of each tree and determining under-forest sapling points in the unassigned point sets, so that complete under-forest sapling point clouds are obtained; using a watershed segmentation method, the complete under-forest sapling point clouds are segmented by single trees to obtain parameters of each sapling in a to-be-measured region, and the under-forest sapling extraction is completed. The application can realize accurate separation of upper-layer large trees and lower-layer sapling point clouds in a region where the upper-layer large trees and the lower-layer sapling are overlapped and the height difference is not obvious, and improve the extraction accuracy of the under-forest sapling.

Description

Technical Field

[0001] This application relates to the field of lidar regional point cloud processing technology, and in particular to a method, device, medium and product for extracting understory saplings based on UAV lidar. Background Technology

[0002] The detection of saplings under plantations is crucial for achieving sustainable forest management. It not only directly reflects the success of plantation regeneration and is a core indicator for assessing its ecological stability, but also provides precise data for scientific thinning, tending, and other management activities.

[0003] UAV-based lidar can acquire high-density point clouds over large areas at a relatively low cost, enabling a comprehensive characterization of forest stands with both upper-layer mature trees and lower-layer saplings. Accurately separating the point clouds of upper-layer mature trees and lower-layer saplings is fundamental to achieving precise individual tree segmentation and accurate estimation of individual tree heights within the lower-layer sapling point cloud. However, occlusion by upper-layer mature trees, lidar signal attenuation, and point cloud overlap between upper and lower layers all affect the accuracy of extracting understory saplings.

[0004] Currently, most research focuses on forest structures with single-layer forests or significant differences in tree height between upper and lower layers. For complex forest stand environments where the upper and lower layers are at similar heights and point cloud overlap is severe, efficient and robust methods for point cloud separation and sapling extraction are still lacking. Therefore, there is an urgent need to develop a method for extracting understory saplings from UAV-based lidar regional point clouds, applicable to areas with insignificant height differences, to improve the automation and accuracy of understory vegetation monitoring. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, medium, and product for extracting understory saplings based on UAV lidar, so as to achieve accurate separation of the point cloud of upper-layer trees and lower-layer saplings in areas where upper-layer trees and lower-layer saplings are mixed and the height difference is not obvious, thereby improving the extraction accuracy of understory saplings.

[0006] To achieve the above objectives, this application provides the following solution.

[0007] Firstly, this application provides a method for extracting saplings in forest understory based on UAV lidar, including: Acquire the regional point cloud of the area to be measured by the UAV's lidar; The point cloud of the region is preprocessed to obtain a normalized point cloud of the region; A canopy height model is constructed based on the normalized region point cloud; The canopy height model is segmented into individual trees to obtain the canopy boundaries of each tree; The canopy boundaries of each tree are vertically projected onto the ground plane. The projection range of each tree is used as a constraint condition for point set partitioning. The normalized region point cloud is then partitioned to obtain the point set of each tree and the unassigned point set. Using the region growth method and connectivity analysis, the points in the point set of each tree are divided to obtain multiple understory sapling points in the point set of each tree. The unassigned point set is denoised to obtain the understory sapling points in the unassigned point set; The understory sapling points in the point set of each tree are merged with the understory sapling points in the unassigned point set to obtain a complete understory sapling point cloud; Using the watershed segmentation method, the complete point cloud of understory saplings is segmented into individual trees to obtain the parameters of each sapling in the area to be tested, thus completing the extraction of understory saplings; the parameters include at least one of the following: canopy boundary, tree height, canopy width, and location.

[0008] In one embodiment, the preprocessing includes point cloud filtering and normalization.

[0009] In one embodiment, constructing a canopy height model based on the normalized region point cloud includes: Points with a height greater than or equal to 10 meters in the normalized region point cloud are used as points for model construction. Spatial interpolation is performed on all points used in the model construction to obtain the canopy height model.

[0010] In one embodiment, the canopy height model is segmented into individual trees to obtain the canopy boundaries of each tree, including: The canopy height model is divided into tree outlines using the watershed single-tree segmentation algorithm to obtain the canopy boundaries of each tree.

[0011] In one embodiment, a region growth method and connectivity analysis are used to divide the points in the point set of each tree, resulting in multiple understory sapling points in each tree's point set, including: Define the point set of any tree as the current point set; Points whose height values ​​fall within ±0.5 meters of the preset tree height in the current point set are all taken as the original seed points of the current point set; The region growing method is used to iterate through the current point set to find all points to be analyzed in the current point set. The process of finding the target seed point at any given iteration number includes: Determine multiple initial seed points under the current iteration number, and set any one of the initial seed points under the current iteration number as the current initial seed point; when the current iteration number is the initial iteration number, the multiple initial seed points under the current iteration number are each original seed point; when the current iteration number is not the initial iteration number, the multiple initial seed points under the current iteration number are the target seed points under the previous iteration number. All points in the current point set that are within a 1-meter radius of the current initial seed point are identified as the nearest points of the current initial seed point. The nearest points that meet the preset conditions are used as the target seed points under the current iteration number. The current iteration number is updated to the next iteration number, and the search for the target seed points under the next iteration number is carried out until the current point set is traversed. The original seed points of the current point set and the target seed points under all iteration numbers of the current point set constitute all the points to be analyzed in the current point set. The preset conditions are that the height value of the nearest point is less than the height value of the current initial seed point and the angle between the vector formed by the nearest point and the current initial seed point and the vertical direction is between 150° and 210°. Based on the points in the current point set that are within a radius of 0.5 meters of each point to be analyzed, construct the adjacency matrix of the current point set; Based on the adjacency matrix of the current point set, each point to be analyzed in the current point set is divided into different connected components; a connected component includes one or more mutually connected points. Each point to be analyzed in the connected component with the most points corresponding to the current point set is identified as a sapling point in the forest under the current point set.

[0012] In one embodiment, the unassigned point set is denoised to obtain understory sapling points within the unassigned point set, including: Based on the height values ​​of each point in the unassigned point set, determine the maximum height value and the minimum height value; The height difference is obtained by subtracting the maximum and minimum height values; The points in the unassigned point set are divided according to a horizontal resolution of 0.5m × 0.5m and a vertical resolution of height difference to obtain multiple three-dimensional voxels. Each 3D voxel contains understory sapling points, and all understory sapling points in the 3D voxels constitute the understory sapling points in the unassigned point set; the process of determining the understory sapling point in any given 3D voxel includes: If the current 3D voxel meets the first condition, then all points in the current 3D voxel will be identified as understory saplings; the first condition is that the height value of all points in the 3D voxel is less than 3 meters. If the current three-dimensional voxel meets the second condition, then all points in the current three-dimensional voxel are determined to be noise points; the second condition is that more than half of the points in the three-dimensional voxel have a height value higher than the preset tree height or all points in the three-dimensional voxel have a height value higher than 1 meter. If the current 3D voxel does not satisfy either the first or the second condition, then the connectivity analysis method is used to perform connectivity analysis on the points in the current 3D voxel to obtain the understory sapling points in the current 3D voxel.

[0013] In one embodiment, a connectivity analysis method is used to perform connectivity analysis on points in the current three-dimensional voxel to obtain the understory sapling points in the current three-dimensional voxel, including: Construct the adjacency matrix of the current 3D voxel based on the points in the current 3D voxel; Based on the adjacency matrix of the current 3D voxel, each point in the current 3D voxel is divided into different connected components; Each point in the connected component with the most points corresponding to the current 3D voxel is determined as the understory sapling point in the current 3D voxel.

[0014] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for extracting saplings in forests based on UAV lidar.

[0015] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for extracting saplings in forests based on UAV lidar.

[0016] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for extracting saplings in forests based on UAV lidar.

[0017] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application discloses a method, device, medium, and product for extracting saplings in forest understory based on UAV lidar. First, a canopy height model is constructed and individual tree segments are performed to obtain the canopy boundaries of each tree. These boundaries are then used as constraints to divide the normalized region point cloud into point sets, limiting potentially overlapping point clouds to the point sets corresponding to each tree, thus avoiding the blindness of global processing. Based on this, for each point set, a region growing method is used, with points within a preset tree height ±0.5 meters as initial seed points. Multiple iterative searches are performed through height value comparison and directional angle constraints (150° to 210°) to accurately extract the saplings. Points to be analyzed within the sapling cluster are further divided into different connected components using connectivity analysis. The connected component with the most points is then identified as the understory sapling point, thus achieving effective separation of upper-layer trees from lower-layer saplings in complex scenarios with indistinct height differences and severe point cloud overlap. Furthermore, for the unassigned point set, voxelization denoising (including height thresholding and connectivity analysis) is used to further extract the understory sapling points. Finally, the sapling points separated from each tree point set are merged with those extracted from the unassigned point set to form a complete understory sapling point cloud, which is then segmented into individual trees to obtain the parameters of each sapling. This entire process, through the organic combination of region growing and connectivity analysis, overcomes the technical difficulties caused by upper-layer tree occlusion, signal attenuation, and point cloud overlap, significantly improving the extraction accuracy of understory saplings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic diagram of a method for extracting saplings in forest based on UAV lidar according to an embodiment of this application; Figure 2 A graph showing the accuracy index for extracting understory saplings from different sites; Figure 3 An overlay display of the segmentation results of the upper-level large trees and the lower-level saplings; Figure 4 This is a top view of the uppermost tree. Figure 5 A top view of the lower saplings; Figure 6 This is a side view of the upper-level tree. Figure 7 This is a side view of the lower-level saplings; Figure 8This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The purpose of this application is to provide a method, device, medium, and product for extracting understory saplings based on UAV lidar, which aims to achieve accurate separation of the point cloud of upper-layer trees and lower-layer saplings in areas where upper-layer trees and lower-layer saplings are mixed and the height difference is not obvious, thereby improving the extraction accuracy of understory saplings.

[0022] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] In one exemplary embodiment, such as Figure 1 As shown, a method for extracting saplings in forest understory based on UAV lidar is provided, including the following steps.

[0024] Step 1: Obtain the area point cloud of the test area collected by the UAV's lidar.

[0025] Step 2: Preprocess the regional point cloud to obtain a normalized regional point cloud.

[0026] As an optional implementation, step 2 includes preprocessing: point cloud filtering and normalization.

[0027] Step 3: Construct a canopy height model based on the normalized regional point cloud.

[0028] As an optional implementation, step 3 includes the following steps.

[0029] Step 31: Use all points in the normalized regional point cloud with a height greater than or equal to 10 meters as points for model construction.

[0030] Step 32: Perform spatial interpolation on all points used in the model construction to obtain the canopy height model.

[0031] Specifically, when spatially interpolating all points used in the model construction, the resolution is 0.2 meters.

[0032] Step 4: Perform individual tree segmentation on the canopy height model to obtain the canopy boundaries of each tree.

[0033] As an optional implementation, step 4 includes: The canopy height model was divided into tree outlines using the watershed single-tree segmentation algorithm to obtain the canopy boundaries of each tree.

[0034] Step 5: Project the canopy boundaries of each tree vertically onto the ground plane. Use the projection range of each tree as a constraint for point set partitioning. Partition the normalized region point cloud to obtain the point set and unassigned point set of each tree.

[0035] Step 6: Using the region growth method and connectivity analysis, the points in the point set of each tree are divided to obtain multiple understory sapling points in the point set of each tree.

[0036] As an optional implementation, step 6 includes the following steps.

[0037] Step 61: Determine the point set of any tree as the current point set.

[0038] Step 62: Select all points in the current point set whose height values ​​fall within ±0.5 meters of the preset tree height as the original seed points of the current point set.

[0039] Specifically, step 62 also includes: determining all points in the current point set whose height value exceeds the preset tree height as upper-level tree points.

[0040] Step 63: Using the region growing method, multiple iterations are performed based on the original seed points of the current point set to determine all points to be analyzed in the current point set; wherein, the process of finding the target seed point at any current iteration number includes the following steps.

[0041] Step 631: Determine multiple initial seed points under the current iteration number, and set any one of the initial seed points under the current iteration number as the current initial seed point; when the current iteration number is the initial iteration number, the multiple initial seed points under the current iteration number are each original seed point; when the current iteration number is not the initial iteration number, the multiple initial seed points under the current iteration number are the target seed points under the previous iteration number.

[0042] Step 632: Determine all points in the current point set that are within a 1-meter radius of the current initial seed point as the nearest points of the current initial seed point.

[0043] Step 633: Take the nearest points that meet the preset conditions as the target seed points under the current iteration number, update the current iteration number to the next iteration number, and search for the target seed points under the next iteration number until the current point set is traversed. The original seed points of the current point set and the target seed points under all iteration numbers of the current point set constitute all the points to be analyzed in the current point set. The preset conditions are that the height value of the nearest point is less than the height value of the current initial seed point and the angle between the vector formed by the nearest point and the current initial seed point and the vertical direction is between 150° and 210°.

[0044] Step 64: Based on the points in the current point set that are within a radius of 0.5 meters of each point to be analyzed, construct the adjacency matrix of the current point set.

[0045] Step 65: Based on the adjacency matrix of the current point set, divide each point to be analyzed in the current point set into different connected components; a connected component includes one or more mutually connected points.

[0046] Specifically, connected components are identified using the Integrated Graph Theory Toolbox in Matlab 2021b.

[0047] Step 66: Determine each point to be analyzed in the connected component with the most points corresponding to the current point set as the understory sapling point in the current point set.

[0048] Specifically, step 66 also includes: determining each point to be analyzed in other connected components corresponding to the current point set as a point in the upper-level tree.

[0049] Step 7: Denoise the unassigned point set to obtain the understory sapling points in the unassigned point set.

[0050] As an optional implementation, step 7 includes the following steps.

[0051] Step 71: Determine the maximum and minimum height values ​​based on the height values ​​of each point in the unassigned point set.

[0052] Step 72: Calculate the difference between the maximum and minimum height values ​​to obtain the height difference.

[0053] Step 73: Divide the unassigned point set into multiple three-dimensional voxels according to a horizontal resolution of 0.5m × 0.5m and a vertical resolution of height difference.

[0054] Step 74: Determine the understory sapling points in each three-dimensional voxel. All understory sapling points in the three-dimensional voxels constitute the understory sapling points in the unassigned point set. The process of determining the understory sapling point in any current three-dimensional voxel includes the following steps.

[0055] Step 741: If the current three-dimensional voxel meets the first condition, then all points in the current three-dimensional voxel are determined as forest sapling points; the first condition is that the height value of all points in the three-dimensional voxel is less than 3 meters.

[0056] Step 742: If the current 3D voxel meets the second condition, then all points in the current 3D voxel are determined to be noise points; the second condition is that more than half of the points in the 3D voxel have a height value higher than the preset tree height or all points in the 3D voxel have a height value higher than 1 meter.

[0057] Step 743: If the current three-dimensional voxel does not satisfy either the first condition or the second condition, then use the connectivity analysis method to perform connectivity analysis on the points in the current three-dimensional voxel to obtain the understory sapling points in the current three-dimensional voxel.

[0058] As an optional implementation, in step 743, a connectivity analysis method is used to perform connectivity analysis on the points in the current three-dimensional voxel to obtain the understory sapling points in the current three-dimensional voxel, including the following steps.

[0059] Step 7431: Construct the adjacency matrix of the current 3D voxel based on the points in the current 3D voxel.

[0060] Step 7432: Based on the adjacency matrix of the current 3D voxel, divide each point in the current 3D voxel into different connected components.

[0061] Step 7433: Determine each point in the connected component with the most points corresponding to the current 3D voxel as the understory sapling point in the current 3D voxel.

[0062] Step 8: Merge the understory sapling points in the point set of each tree with the understory sapling points in the unassigned point set to obtain a complete understory sapling point cloud.

[0063] Step 9: Using the watershed segmentation method, perform individual tree segmentation on the complete forest understory sapling point cloud to obtain the parameters of each sapling in the test area, thus completing the extraction of forest understory saplings.

[0064] The parameters include at least one of the following: canopy boundary, tree height, canopy width, and location.

[0065] Specifically, point clouds from multiple different sites were used as validation data, and the performance of the method in this application was evaluated using detection accuracy metrics (recall, precision, and F1 score). The accuracy metrics for extracting understory saplings from different sites are as follows: Figure 2 As shown.

[0066] When using the understory sapling extraction method of this application to segment a certain area into upper-layer mature trees and lower-layer saplings, the results are as follows: Figures 3-7 As shown.

[0067] In one exemplary embodiment, a computer device is provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement a method for extracting saplings in forest under cover based on UAV lidar.

[0068] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements a method for extracting saplings in forest understory based on UAV lidar.

[0069] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements a method for extracting saplings in forest understory based on UAV lidar.

[0070] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 8 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media to run. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for extracting saplings in forest understory based on UAV lidar.

[0071] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0072] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0073] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0074] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0075] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0076] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for extracting saplings in forest understory based on UAV lidar, characterized in that, The method for extracting saplings in the forest based on UAV lidar includes: Acquire the regional point cloud of the area to be measured by the UAV's lidar; The point cloud of the region is preprocessed to obtain a normalized point cloud of the region; A canopy height model is constructed based on the normalized region point cloud; The canopy height model is segmented into individual trees to obtain the canopy boundaries of each tree; The canopy boundaries of each tree are vertically projected onto the ground plane. The projection range of each tree is used as a constraint condition for point set partitioning. The normalized region point cloud is then partitioned to obtain the point set of each tree and the unassigned point set. Using the region growth method and connectivity analysis, the points in the point set of each tree are divided to obtain multiple understory sapling points in the point set of each tree. The unassigned point set is denoised to obtain the understory sapling points in the unassigned point set; The understory sapling points in the point set of each tree are merged with the understory sapling points in the unassigned point set to obtain a complete understory sapling point cloud; Using the watershed segmentation method, the complete point cloud of understory saplings is segmented into individual trees to obtain the parameters of each sapling in the area to be tested, thus completing the extraction of understory saplings; the parameters include at least one of canopy boundary, tree height, canopy width, and location; Using the region growth method and connectivity analysis, the points in each tree's point set were divided to obtain multiple understory sapling points in each tree's point set, including: Define the point set of any tree as the current point set; Points whose height values ​​fall within ±0.5 meters of the preset tree height in the current point set are all taken as the original seed points of the current point set; The region growing method is used to iterate through the current point set to find all points to be analyzed in the current point set. The process of finding the target seed point at any given iteration number includes: Determine multiple initial seed points under the current iteration number, and set any one of the initial seed points under the current iteration number as the current initial seed point; when the current iteration number is the initial iteration number, the multiple initial seed points under the current iteration number are each original seed point; when the current iteration number is not the initial iteration number, the multiple initial seed points under the current iteration number are the target seed points under the previous iteration number. All points in the current point set that are within a 1-meter radius of the current initial seed point are identified as the nearest points of the current initial seed point. The nearest points that meet the preset conditions are used as the target seed points under the current iteration number. The current iteration number is updated to the next iteration number, and the search for the target seed points under the next iteration number is carried out until the current point set is traversed. The original seed points of the current point set and the target seed points under all iteration numbers of the current point set constitute all the points to be analyzed in the current point set. The preset conditions are that the height value of the nearest point is less than the height value of the current initial seed point and the angle between the vector formed by the nearest point and the current initial seed point and the vertical direction is between 150° and 210°. Based on the points in the current point set that are within a radius of 0.5 meters of each point to be analyzed, construct the adjacency matrix of the current point set; Based on the adjacency matrix of the current point set, each point to be analyzed in the current point set is divided into different connected components; a connected component includes one or more mutually connected points. Each point to be analyzed in the connected component with the most points corresponding to the current point set is identified as a sapling point in the forest under the current point set.

2. The method for extracting saplings in forest understory based on UAV lidar according to claim 1, characterized in that, The preprocessing includes point cloud filtering and normalization.

3. The method for extracting saplings in forest understory based on UAV lidar according to claim 1, characterized in that, Constructing a canopy height model based on the normalized region point cloud includes: Points with a height greater than or equal to 10 meters in the normalized region point cloud are used as points for model construction. Spatial interpolation is performed on all points used in the model construction to obtain the canopy height model.

4. The method for extracting saplings in forest understory based on UAV lidar according to claim 1, characterized in that, The canopy height model is segmented into individual trees to obtain the canopy boundaries of each tree, including: The canopy height model is divided into tree outlines using the watershed single-tree segmentation algorithm to obtain the canopy boundaries of each tree.

5. The method for extracting saplings in forest understory based on UAV lidar according to claim 1, characterized in that, The unassigned point set is denoised to obtain understory sapling points in the unassigned point set, including: Based on the height values ​​of each point in the unassigned point set, determine the maximum height value and the minimum height value; The height difference is obtained by subtracting the maximum and minimum height values; The points in the unassigned point set are divided according to a horizontal resolution of 0.5m × 0.5m and a vertical resolution of height difference to obtain multiple three-dimensional voxels. Each 3D voxel contains understory sapling points, and all understory sapling points in the 3D voxels constitute the understory sapling points in the unassigned point set; the process of determining the understory sapling point in any given 3D voxel includes: If the current 3D voxel meets the first condition, then all points in the current 3D voxel will be identified as understory saplings; the first condition is that the height value of all points in the 3D voxel is less than 3 meters. If the current three-dimensional voxel meets the second condition, then all points in the current three-dimensional voxel are determined to be noise points; the second condition is that more than half of the points in the three-dimensional voxel have a height value higher than the preset tree height or all points in the three-dimensional voxel have a height value higher than 3 meters. If the current 3D voxel does not satisfy either the first or the second condition, then the connectivity analysis method is used to perform connectivity analysis on the points in the current 3D voxel to obtain the understory sapling points in the current 3D voxel.

6. The method for extracting saplings in forest understory based on UAV lidar according to claim 5, characterized in that, Connectivity analysis is used to analyze the points in the current 3D voxel to obtain the understory sapling points in the current 3D voxel, including: Construct the adjacency matrix of the current 3D voxel based on the points in the current 3D voxel; Based on the adjacency matrix of the current 3D voxel, each point in the current 3D voxel is divided into different connected components; Each point in the connected component with the most points corresponding to the current 3D voxel is determined as the understory sapling point in the current 3D voxel.

7. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for extracting saplings in forests based on UAV lidar as described in any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for extracting saplings in forests based on UAV lidar as described in any one of claims 1-6.

9. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for extracting saplings in forests based on UAV lidar as described in any one of claims 1-6.

Images