A method and system for extracting the direction of tree trunk based on hierarchical symmetry analysis

Abstract

The application discloses a tree trunk direction extraction method and system based on layered symmetry analysis, which comprises the following steps: preprocessing a tree point cloud; based on a crown elevation model of the point cloud, using an image segmentation algorithm, the preprocessed point cloud is segmented into single-tree point clouds; the single-tree point clouds are divided into multiple non-equidistant height layers in the Z-axis direction; the most symmetric direction of the point cloud in each height layer is found; the angle between the most symmetric direction and the average direction of each height layer is calculated, and the most symmetric direction with an angle greater than a set threshold is removed; and the most symmetric direction of each height layer is weighted and averaged to obtain the trunk direction of the tree. The application uses a method based on layered symmetry analysis to identify the main direction, which is more in line with the natural posture of the tree, thereby improving the fitting accuracy of the trunk direction and having good robustness and generalization ability under complex forest environment and noise interference.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of forestry remote sensing, computer vision and point cloud three-dimensional reconstruction, and particularly relates to a tree trunk direction extraction method and system based on hierarchical symmetry analysis. BACKGROUND

[0002] With the rapid development of LiDAR, unmanned aerial vehicle remote sensing and three-dimensional modeling technology, forestry resource investigation gradually transits from traditional manual operation mode to intelligent analysis mode based on three-dimensional point cloud data. Point cloud technology can capture the spatial form information of each tree in the forest with high precision, providing basic data support for forest structure modeling, single tree parameter estimation, ecosystem simulation and other applications.

[0003] In point cloud-based forest resource management, the trunk direction (i.e. growth direction) of a single tree is a key structural parameter. It not only affects the light acquisition, water transport and wind resistance stability of trees during growth, but also has important application value in subsequent timber volume estimation, falling risk prediction and forestry machinery path planning. However, in natural forests or dense forest environments, due to the complex tree shape, serious branch and leaf interference, uneven point cloud density, and ground aliasing caused by terrain undulations, traditional methods relying on rule-based hierarchical simplification, cylindrical fitting or manual interaction cannot accurately and stably extract the trunk direction of each tree, especially in high-density forest areas or noisy data conditions, with significant errors and low processing efficiency.

[0004] The main deficiencies of the prior art are as follows: 1) Lack of complete preprocessing and segmentation process: Most methods do not perform systematic point cloud denoising and single tree segmentation, resulting in impure fitting objects, which seriously affects the fitting accuracy; 2) Instability of cylindrical model fitting accuracy: traditional least squares fitting method is sensitive to outliers, and is prone to divergence when there are branches, leaves or point cloud interference; 3) Low automation: a large number of methods require manual specification of initial axis or parameters, which cannot meet the needs of large-scale forest automation processing. Therefore, there is an urgent need for a robust, highly automated method that can stably extract the trunk direction under noisy conditions to solve the above problems and promote the development of intelligent forestry investigation technology based on point cloud towards high precision and high efficiency. SUMMARY

[0005] The present application provides a tree trunk direction extraction method based on hierarchical symmetry analysis to overcome the deficiencies of the prior art, which can automatically extract the growth direction of trees in batches under high noise conditions, including the following steps:

[0006] Step 1, preprocessing the tree point cloud collected by unmanned aerial vehicle LiDAR;

[0007] Step 2, based on the point cloud canopy height model, using image segmentation algorithm, the tree point cloud after pre-processing in step 1 is segmented into single tree point cloud;

[0008] Step 3, the single tree point cloud is divided into multiple non-equidistant height layers in the Z-axis direction;

[0009] Step 4, find the most symmetric direction of the point cloud in each height layer;

[0010] Step 5, calculate the angle between the most symmetric direction and the average direction of each height layer, and remove the most symmetric direction with an angle greater than the set threshold;

[0011] Step 6, the most symmetric direction of each height layer is weighted and averaged to obtain the trunk direction of the tree.

[0012] Further, the pre-processing in step 1 includes removing height outliers and isolated points in the point cloud, then performing height normalization, filtering and unifying the coordinate system, and removing the points at the bottom of the point cloud and retaining the point cloud of the main branches at the top.

[0013] Further, the step 2 generates a canopy height model based on point cloud data, uses local maximum value detection to determine the tree top position, and then applies an image segmentation algorithm to the canopy height model image to divide the point cloud into multiple single tree regions.

[0014] Further, in step 3, kernel density estimation or local point number histogram is used to analyze the density distribution of the point cloud in the Z-axis direction, and the density change significantly is taken as the layer boundary point to obtain multiple non-equidistant height layers.

[0015] Further, in step 4, the point set in each height layer is traversed in the two-dimensional plane XOY with the X-axis as the starting direction , and the point set is projected in the direction . The projection coordinates are calculated as follows:

[0016] (1)

[0017] In the formula, is the projection difference of the point and the point cloud center in the direction in the height layer; is the mean value of the coordinates of all points x in the height layer, is the mean value of the coordinates of all points y in the height layer.

[0018] The expression for evaluating the symmetry measure function is:

[0019] (2)

[0020] wherein, is a symmetry measure function, is the median of all projection difference values, is the variance of projection difference values compared with the median, i.e. the degree of dispersion.

[0021] Take the direction corresponding to the minimum value of as the most symmetric direction of the layer.

[0022] Further, the average direction in step 5 is calculated as follows:

[0023] (3)

[0024] wherein, is the number of divided height layers, is the most symmetric direction of the i height layer.

[0025] Further, the three-dimensional direction vector of the most symmetric direction of each height layer is first calculated in step 6, and the specific calculation method is as follows:

[0026] (4)

[0027] wherein, is the three-dimensional direction vector of the most symmetric direction of the i height layer, is the most symmetric direction of the i height layer, is the height of the i height layer.

[0028] Suppose that after step 5 removes the most symmetric directions with angles exceeding the limit, there are most symmetric directions left, then the trunk direction of the tree is calculated according to the following formula:

[0029] (5)

[0030] wherein, is the trunk direction of the tree, is the three-dimensional direction vector of the most symmetric direction of the i height layer, is the weight of the most symmetric direction of the i height layer, which is the number of point clouds in each height layer; denotes the modulus.

[0031] ​The application further provides a tree trunk direction extraction system based on hierarchical symmetry analysis.

[0032] Moreover, the system comprises a processor and a memory, the memory is used for storing program instructions, and the processor is used for calling the stored instructions in the memory to execute the tree trunk direction extraction method based on hierarchical symmetry analysis.

[0033] Alternatively, the system comprises a readable storage medium, and the readable storage medium stores a computer program, and the computer program is executed to implement the tree trunk direction extraction method based on hierarchical symmetry analysis.

[0034] Compared with the prior art, the application has the following advantages:

[0035] 1) The application uses a method based on intra-layer symmetry analysis to identify the main direction, which is more consistent with the natural posture of trees, thereby improving the accuracy of the trunk direction fitting; 2) The application uses an adaptive hierarchical and direction anomaly rejection mechanism to process point cloud structures, thereby enhancing the robustness and generalization ability in complex forest environments and noise interference; 3) The application realizes efficient batch processing without human intervention by designing a fully automatic data processing flow (from preprocessing to direction weighted fusion); 4) The application uses a general geometric logic construction method framework, which is not dependent on a specific point cloud platform, so that it has good cross-regional and cross-device applicability and promotion potential; 5) The application integrates point cloud preprocessing, single tree segmentation, direction estimation, and structured result output modules, and provides a complete, continuous, and integrable analysis process, which is convenient for subsequent statistics and modeling. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0037] Figure 1 It is a flowchart of the tree trunk direction extraction method based on hierarchical symmetry analysis of the embodiment of the application.

[0038] Figure 2 It is a schematic diagram of the projection of the point set in each height layer of the embodiment of the application in the two-dimensional plane XOY.

[0039] Figure 3 It is a display diagram of the tree trunk direction obtained by the embodiment of the application.

[0040] Figure 4 A diagram illustrating the degree of tree tilt obtained from an embodiment of the present invention is provided.

[0041] Figure 5 This invention provides a diagram illustrating the distribution of tree growth direction along the slope obtained from an embodiment of the invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be further described below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0043] Example 1

[0044] like Figure 1 As shown, this embodiment of the invention provides a method for extracting the trunk direction of a tree based on hierarchical symmetry analysis, including the following steps:

[0045] Step 1: Preprocess the tree point cloud data collected by the UAV LiDAR.

[0046] Preprocessing includes removing height outliers and isolated points from the point cloud, followed by height normalization and filtering, unifying the coordinate system to the WGS 1984 coordinate system, and removing the bottom 20% of points in the point cloud, retaining the point cloud of the main upper branches. In this embodiment, removing height outliers refers to removing… The point, among which, Refers to the first i The height of each point The average height of all points. This refers to the standard deviation of the height of all points. (Used) k Neighborhood density estimation removes isolated points, i.e., removes points by radius. k Points whose number of interior points is less than a set threshold.

[0047] Step 2: Based on the canopy elevation model of the point cloud, an image segmentation algorithm is used to segment the tree point cloud after preprocessing in Step 1 into individual tree point clouds.

[0048] A canopy height model (CHM) is generated based on point cloud data. Local maximum detection is used to determine the treetop location. Then, a watershed segmentation algorithm is applied to the CHM image to divide the point cloud into multiple individual tree regions.

[0049] Step 3: Divide the single-tree point cloud into multiple non-equally spaced height layers along the Z-axis.

[0050] The density distribution of the point cloud in the Z-axis is analyzed using kernel density estimation or local point histogram, and the point with significant density change is taken as the boundary point of the layer, ensuring that the layer can accurately depict the local geometric trend.

[0051] Step 4, find the most symmetric direction of the point cloud in each height layer, and take it as the projection direction of the trunk direction in the two-dimensional plane.

[0052] Because the bending degree of trees in different height layers is not necessarily the same, and the tree morphology theory believes that even if the tree trunk is inclined, the left and right sides along the inclined direction should be approximately symmetrically distributed, so the most symmetric direction of each layer of point cloud is found, which is taken as the projection direction of the trunk direction in the two-dimensional plane.

[0053] As shown in Figure 2 , the red dot is the center of the point cloud in the height layer, and the point set in the height layer is projected in the two-dimensional plane XOY with the X-axis as the starting direction, and the point set is projected in the direction , and the projection coordinate is calculated as follows:

[0054] (1)

[0055] In the formula, is the projection difference value of the point and the center of the point cloud in the height layer in the direction; is the mean value of the coordinates of all points in the height layer , and x is the mean value of the coordinates of all points in the height layer . y The expression of the symmetry measure function is:

[0056]

[0057] (2)

[0058] In the formula, is the symmetry measure function; is the median of all projection difference values; is the variance of the projection difference value compared with the median, i.e. the degree of dispersion.

[0059] Take the direction corresponding to the minimum value as the most symmetric direction of the layer.

[0060] Step 5, calculate the angle between the most symmetric direction and the average direction of each height layer, and remove the most symmetric direction with an angle greater than the set threshold. ​​​

[0061] Average direction The calculation method is as follows:

[0062] (3)

[0063] In the formula, is the number of divided height layers, is the most symmetric direction of the i height layer.

[0064] Step 6: Weighted average of the most symmetric direction of each height layer to obtain the trunk direction of the tree.

[0065] The three-dimensional direction vector of the most symmetric direction of the i height layer can be represented as:

[0066] (4)

[0067] In the formula, is the most symmetric direction of the i height layer, is the height of the i height layer.

[0068] Suppose after step 5, there are most symmetric directions left after removing the most symmetric directions with an included angle exceeding the limit, then the trunk direction of the tree is calculated as follows:

[0069] (5)

[0070] In the formula, is the trunk direction of the tree; is the three-dimensional direction vector of the most symmetric direction of the i height layer; is the weight of the most symmetric direction of the i height layer, which is the number of point clouds in each height layer; represents the modulus.

[0071] The geometric center (also known as the centroid or barycenter) of the point cloud is a basic feature of point cloud data, which represents the average position of the point cloud in space. Assuming that the single tree point cloud obtained after step 3 single tree segmentation has points, and the coordinates of each point are , the weighted average of the coordinates of each dimension is obtained to obtain the geometric center point coordinates of the single tree point cloud , that is:

[0072] (6)

[0073] Calculate the equivalent radius of the trunk: ​

[0074] (7)

[0075] (8)

[0076] In the formula, R is the equivalent radius of the tree trunk. This represents the number of points in a single tree point cloud. It is the horizontal projection distance from any point in the single-tree point cloud to the geometric center.

[0077] Output and save the trunk direction vector, geometric center coordinates, and equivalent radius of each tree as a structured CSV file. Calculate the trunk direction and vector of each individual tree. The degree of deviation is used to determine the degree of tree tilt.

[0078] Figure 3 To process tree tilt direction maps obtained from airborne LiDAR data of a natural forest using the method proposed in this invention, the direction of the vector arrows indicates the calculated direction of the trunk of a single tree, and the color and length of the arrows indicate the degree of tilt of the single tree.

[0079] Figure 4 The image shows the result of processing laser point clouds of natural coniferous forests in mountainous terrain, collected by an airborne LiDAR in complex terrain, using the method proposed in this invention. The image contains 24,036 individual trees, with each point representing a tree. The color of the point indicates the direction and vector of the tree's trunk. The degree of deviation, i.e. the degree of tilt.

[0080] Figure 5 for Figure 4 Analysis of the leaning direction of trees in the same area: According to the slope distribution, the leaning direction of trees is divided into eight zones at 45° intervals: North, Northeast, East, Southeast, South, Southwest, West, and Northwest. Different colors represent the trunk direction of each individual tree.

[0081] Example 2

[0082] Based on the same inventive concept, the present invention also provides a tree trunk direction extraction system based on hierarchical symmetry analysis, including a processor and a memory. The memory is used to store program instructions, and the processor is used to call the program instructions in the memory to execute the tree trunk direction extraction method based on hierarchical symmetry analysis as described above.

[0083] Example 3

[0084] Based on the same inventive concept, the present invention also provides a tree trunk direction extraction system based on hierarchical symmetry analysis, including a readable storage medium on which a computer program is stored. When the computer program is executed, it implements the tree trunk direction extraction method based on hierarchical symmetry analysis as described above.

[0085] In practice, the method of the present application can be implemented by a person skilled in the art using computer software technology to automatically run the process, and the system device of the method, such as a computer readable storage medium storing the corresponding computer program of the present application and a computer device including the corresponding computer program, should also be within the scope of protection of the present application.

[0086] The specific embodiments described herein are merely illustrative of the spirit of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them without departing from the spirit of the present application or exceeding the scope defined by the appended claims.

Claims

1. A method for extracting the trunk direction of trees based on hierarchical symmetry analysis, characterized in that, Includes the following steps: Step 1: Preprocess the tree point cloud data collected by the UAV LiDAR. Step 2: Based on the canopy elevation model of the point cloud, an image segmentation algorithm is used to segment the tree point cloud after the preprocessing in Step 1 into individual tree point clouds. Step 3: Divide the single-tree point cloud into multiple non-equally spaced height layers along the Z-axis; Step 4: Find the most symmetrical direction of the point cloud in each altitude layer; For the point set in each height layer in the two-dimensional plane XOY Starting from the X-axis, it traverses multiple directions. ,exist The projection of the point set onto a direction is performed, and the projection coordinates are calculated as follows: （1） In the formula, For point With the center of the point cloud within this height layer exist Projection difference in direction, For all points within this height layer x The mean of the coordinates, For all points within this height layer y The mean of the coordinates; The expression for the symmetry metric function is: （2） In the formula, It is a symmetry metric function. The median of all projection differences. This is the variance of the projected difference relative to the median, i.e., the degree of dispersion; Pick The value corresponding to the minimum The direction is the most symmetrical direction of this layer; Step 5: Calculate the angle between the most symmetrical direction and the average direction for each height layer, and remove the most symmetrical direction whose angle is greater than a set threshold. Average direction The calculation method is as follows: （3） In the formula, The number of height layers to be divided, For the first i The most symmetrical direction of each height layer; Step 6: Take a weighted average of the most symmetrical directions at each height level to obtain the trunk direction of the tree; First, calculate the three-dimensional direction vector of the most symmetrical direction for each height layer. The specific calculation method is as follows: （4） In the formula, For the first i The three-dimensional direction vector of the most symmetrical direction of each height layer. For the first i The most symmetrical direction of each height layer For the first i The height of each floor; Assume that after step 5 removes the most symmetrical direction with an excessive included angle, there are still... The formula for calculating the direction of a tree's trunk is as follows: (The formula is not provided in the original text.) （5） In the formula, The direction of the tree trunk. For the first i The weight of the most symmetrical direction of each height layer is determined by the number of point clouds in each height layer. Indicates the modulus.

2. The method for extracting the trunk direction of a tree based on hierarchical symmetry analysis as described in claim 1, characterized in that: The preprocessing in step 1 includes removing height outliers and isolated points from the point cloud, then performing height normalization, filtering, and coordinate system unification, and removing points located at the bottom of the point cloud while retaining the point cloud of the main upper branches.

3. The method for extracting the trunk direction of a tree based on hierarchical symmetry analysis as described in claim 2, characterized in that: In step 2, a canopy elevation model is generated based on point cloud data. Local maximum detection is used to determine the treetop position. Then, an image segmentation algorithm is applied to the canopy elevation model image to divide the point cloud into multiple individual tree regions.

4. The method for extracting the trunk direction of a tree based on hierarchical symmetry analysis as described in claim 1, characterized in that: In step 3, kernel density estimation or local point count histogram analysis is used to analyze the density distribution of the point cloud along the Z-axis. Points with significant density changes are taken as layer boundary points to obtain multiple non-equidistant height layers.

5. A tree trunk direction extraction system based on hierarchical symmetry analysis, characterized in that, It includes a processor and a memory, the memory being used to store program instructions, and the processor being used to call the program instructions in the memory to execute the tree trunk direction extraction method based on hierarchical symmetry analysis as described in any one of claims 1-4.

6. A tree trunk direction extraction system based on hierarchical symmetry analysis, characterized in that, The method includes a readable storage medium on which a computer program is stored, and when the computer program is executed, it implements a tree trunk direction extraction method based on hierarchical symmetry analysis as described in any one of claims 1-4.

Images