A method for constructing a digital terrain resource GIS model

Abstract

The application relates to the technical field of GIS modeling, in particular to a digital terrain resource GIS model construction method, which comprises the following steps: determining a spatial complexity based on the confusion degree of all data points in the point cloud of each tree; determining a point cloud confusion degree based on the distance of all data points in any columnar region of the point cloud of each tree to the centroid of the columnar region, the coordinate difference of all data points, and the spatial complexity; determining a tree distribution density based on the distance between each tree in each region and each tree in the neighborhood thereof and the point cloud confusion degree, so as to adjust the grid window size of each region; and constructing a GIS model of a target forest area based on the three-dimensional point cloud after gridding. The application improves the precision and efficiency of digital terrain resource GIS model construction by adaptively adjusting the grid size.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of GIS modeling, and particularly relates to a digital terrain resource GIS model construction method. BACKGROUND

[0002] In the forestry field, by constructing a high-precision three-dimensional GIS model, the forestry department can fuse multi-source data to realize visual analysis and dynamic early warning of forest health and growth trend. This not only provides a scientific tool for the in-depth application of spatio-temporal data, but also provides core support for the decision-making and management of intelligent forestry with its precision and convenience, and is a key direction of the development of forestry informatization.

[0003] Based on GIS modeling is the current mainstream three-dimensional modeling method, but when reconstructing a forest area with complex terrain, the traditional GIS modeling converts the point cloud data of the forest area terrain into regular grids based on a fixed grid size. When modeling a dense forest area with many detailed features based on gridding, if the grid is too large, the shape features of the trees and the surrounding terrain features are easily lost. If a too small grid is used in a relatively flat forest area to be developed, a large number of similar redundant grids are generated, the calculation complexity is increased, and the precision and efficiency of the digital terrain resource GIS model construction are reduced. SUMMARY

[0004] To solve the above technical problems, the application provides a digital terrain resource GIS model construction method to solve the existing problems.

[0005] The digital terrain resource GIS model construction method provided by the application adopts the following technical scheme:

[0006] One embodiment of the application provides a digital terrain resource GIS model construction method, which comprises the following steps:

[0007] Obtaining three-dimensional point cloud of a target forest area;

[0008] Dividing the three-dimensional point cloud into multiple regions by using a region growing algorithm, and extracting point cloud of each tree in each region; dividing the point cloud of each tree into multiple columnar regions, determining spatial complexity of each tree based on a degree of disorder of all data points in the point cloud of each tree, determining point cloud distribution feature value of any columnar region of the point cloud of each tree based on distance from all data points in the columnar region to a centroid of the columnar region, and determining spatial distribution feature value of each tree in combination with coordinate difference of all data points in any columnar region of the point cloud of each tree; and determining point cloud disorder degree of each tree based on the spatial complexity and the spatial distribution feature value.

[0009] determine a tree distribution density in each region based on a distance between each tree in the region and each tree in a neighborhood of the tree and in combination with the point cloud disorder, to adjust a grid window size of each region in the three-dimensional point cloud of the target forest area;

[0010] construct a GIS model of the target forest area based on the rasterized three-dimensional point cloud.

[0011] Preferably, the region growing algorithm is used to divide the three-dimensional point cloud into multiple regions, including:

[0012] The three-dimensional point cloud of the target forest area contains elevation information, and the three-dimensional point cloud of the target forest area is taken as an input of the region growing algorithm, wherein a data point with the largest elevation value in the three-dimensional point cloud is taken as an initial seed point, and all regions of the three-dimensional point cloud are output.

[0013] Preferably, the point cloud of each tree is divided into multiple cylindrical regions, including:

[0014] The centroid in the point cloud of each tree is obtained, the centroid of the tree point cloud is taken as the centroid of an initial cylindrical region, a preset height and a preset radius are respectively taken as the height and the radius of the bottom surface of the initial cylindrical region, and cylindrical regions with the same size as the initial cylindrical region are constructed above and below the initial cylindrical region, and the distance between adjacent cylindrical regions is 0, until the entire tree point cloud is traversed, to obtain the cylindrical regions of each tree point cloud.

[0015] Preferably, the spatial complexity of each tree is the information entropy of the coordinates of all data points in the point cloud of each tree.

[0016] Preferably, the point cloud distribution characteristic value of any cylindrical region of each tree point cloud is a result of taking a normalized value of a mean value of distances from all data points in any cylindrical region of each tree point cloud to the centroid of the cylindrical region.

[0017] Preferably, the determination method of the spatial distribution characteristic value of each tree is:

[0018] The range of Z-axis coordinate values of all data points in any cylindrical region of each tree point cloud is calculated, and a ratio of the range to the corresponding point cloud distribution characteristic value is calculated and recorded as a point cloud distribution value of any cylindrical region of each tree point cloud.

[0019] The mean value of the point cloud distribution values of all cylindrical regions of each tree point cloud is taken as the spatial distribution characteristic value of each tree.

[0020] Preferably, the point cloud disorder of each tree is a result of positive fusion of the spatial complexity and the spatial distribution characteristic value of each tree point cloud.

[0021] Preferably, the expression of the tree distribution density in each region is: wherein, represents the tree distribution density within the region g; represents the point cloud disorder of the tree k within the neighborhood of the tth tree within the region g; represents the distance between the tth tree and the tree k within its neighborhood within the region g; represents the number of all trees within the neighborhood of the tth tree; represents the number of all trees within the region g; norm() represents a normalization function.

[0022] Preferably, the adjusting the grid window size of each region within the three-dimensional point cloud of the target forest area comprises:

[0023] the expression of the square grid window length of the region g within the three-dimensional point cloud of the target forest area is: wherein, represents the tree distribution density within the region g; represents the preset initial square grid window length; represents the preset adjustment coefficient.

[0024] Preferably, the constructing the GIS model of the target forest area comprises:

[0025] performing grid processing on the three-dimensional point cloud according to the grid window size to preliminarily generate the rasterized digital elevation model, using the inverse distance weighted interpolation method to refine the elevation value of each grid in the digital elevation model, and fusing with the collected forestry information to obtain the GIS model of the target forest area.

[0026] The present application has at least the following beneficial effects:

[0027] ​The application analyzes the spatial complexity of the overall structure disorder degree of the tree, and constructs point cloud wave disorder degree by combining with the local vertical fluctuation and horizontal aggregation characteristics, comprehensively evaluates the geometric details and information density of the single tree, and provides a basis for subsequent adaptive gridding, which can intelligently use fine grid for complex trees to retain key features, and use coarse grid for simple trees to improve efficiency; further, the application constructs the tree distribution density based on the spatial distance between the point cloud wave disorder degree of the single tree and the trees, and adaptively adjusts the grid window size, automatically uses small grid in the high information area composed of complex and dense trees to finely depict the key details such as crown layer interlacing and understory microtopography, and automatically switches to large grid in the simple and sparse area to simplify the model and improve efficiency, so as to realize the intelligent balance between precision and performance of the forest three-dimensional modeling; in summary, the application can automatically use small grid in the complex area to preserve precision, and switch to large grid in the simple area to improve efficiency, so as to balance the modeling precision and computing performance, and improve the precision and efficiency of the digital terrain resource GIS model construction. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0029] Figure 1 A step flow chart of a digital terrain resource GIS model construction method provided by an embodiment of the present application is shown in the figure.

[0030] Figure 2 A point cloud wave disorder degree extraction process schematic diagram provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0031] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structure, characteristics and effects of the digital terrain resource GIS model construction method according to the present application are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0033] Specifically, the application provides a specific scheme of a digital terrain resource GIS model construction method.

[0034] An embodiment of the application provides a digital terrain resource GIS model construction method, and specifically provides a digital terrain resource GIS model construction method as follows. Figure 1 The method comprises the following steps.

[0035] Step S1: acquiring three-dimensional point clouds of a target forest area.

[0036] The three-dimensional point clouds of the target forest area are scanned by a UAV carrying a LiDAR device. Birds often exist in the forest area. In order to avoid misjudging the birds as terrain resources of the forest area, it is necessary to remove outliers from the collected point cloud data. Since the point cloud distribution of the birds is sporadic and isolated, and the birds are dynamic targets, they can only appear at a certain moment when the LiDAR device scans, and the point cloud data is extremely small and has no continuous spatial distribution rule, and the spatial distance from the surrounding points is large. Therefore, the outliers can be removed by outlier filtering, and the three-dimensional point clouds collected above are taken as the input of the outlier filtering algorithm. The initial number of neighborhood points is set to 15, the threshold of the standard deviation multiple is set to 1.5, and the three-dimensional point clouds after filtering and denoising are output.

[0037] The outlier filtering and denoising is a known technology, and the specific process of denoising the three-dimensional point clouds will not be repeated.

[0038] It is additionally explained that the XOY plane of the coordinate system in which the three-dimensional point clouds are located is the earth's surface, and the positive direction of the Z axis is the direction perpendicular to the XOY plane and upward.

[0039] Step S2: calculating the point cloud turbulence of each tree itself, combining the distance from the tree to the neighborhood trees, and comprehensively evaluating the tree distribution density of each region to dynamically adjust the grid size of the point cloud modeling.

[0040] Traditional forest terrain modeling is based on a fixed grid digital elevation model. This method ignores the significant difference in tree point cloud distribution. In the forest area, the point cloud in the dense tree area is dense and undulating, while the point cloud in the sparse area or road is relatively flat and sparse. Dividing such unevenly distributed data points into fixed-size grids will result in low modeling efficiency. Therefore, the embodiment adjusts the grid size based on the point cloud distribution density to improve the efficiency of forest terrain resource modeling. The specific process is as follows:

[0041] S2.1: the three-dimensional point cloud is divided into multiple regions by using a region growing algorithm, and the point cloud of each tree in each region is extracted; the point cloud of each tree is divided into multiple cylindrical regions, the spatial complexity of each tree is determined based on the confusion degree of all data points in the point cloud of each tree, the point cloud distribution characteristic value of any cylindrical region of the point cloud of each tree is determined based on the distance from all data points in any cylindrical region of the point cloud of each tree to the centroid of the cylindrical region, and the spatial distribution characteristic value of each tree is determined in combination with the coordinate difference of all data points in any cylindrical region of the point cloud of each tree; and the point cloud turbulence degree of each tree is determined based on the spatial complexity and the spatial distribution characteristic value.

[0042] Since the trees in the forest area are usually arranged in blocks, and the tree density of each block is also different, the three-dimensional point cloud is divided into multiple regions by using a region growing algorithm in this embodiment, specifically:

[0043] The three-dimensional point cloud of the target forest area contains elevation information, and the three-dimensional point cloud of the target forest area is taken as the input of the region growing algorithm, wherein the data point with the maximum elevation value in the three-dimensional point cloud is taken as the initial seed point, the spatial distance threshold is set to 1 m, and the elevation difference threshold is set to 1 m; starting from the seed point, the data points in the eight-neighborhood of the seed point are checked, if the spatial distance and the elevation difference between the current data point in the eight-neighborhood and the seed point are less than the corresponding spatial distance threshold and the elevation difference threshold respectively, then the current data point and the seed point are merged into the same region, and the current data point is taken as a new seed point to repeat the above growing process until there is no data point that meets the growing criteria, and all regions of the three-dimensional point cloud are output.

[0044] Wherein, the region growing algorithm is a known technology, and the specific process of dividing the three-dimensional point cloud of the target forest area by using the region growing algorithm will not be described again.

[0045] Further, the point cloud of each tree in each region is extracted by using a single tree segmentation algorithm in this embodiment, and the single tree segmentation algorithm based on the canopy height model (CHM) is used to extract the tree point cloud in each region in this embodiment. In actual application process, as other implementation manners, the implementer can also select other single tree segmentation algorithm according to specific circumstances, and the selection of the single tree segmentation algorithm is not specially limited in this embodiment.

[0046] Wherein, the single tree segmentation algorithm based on the canopy height model (CHM) is a known technology, and the specific process of extracting the single tree point cloud by using the single tree segmentation algorithm will not be described again.

[0047] Further, since the point cloud distribution of the top of the trees in the dense forest area usually has obvious up-and-down fluctuation distribution rule, that is, the point cloud density in the Z-axis direction of the three-dimensional point cloud is larger, and is relatively sparse at the ground. In order to quantify the density of this vertical distribution, the point cloud of each tree is divided into a plurality of columnar regions in the embodiment, that is, the columnar region is taken as a basic analysis unit, and the specific division process is as follows:

[0048] In the embodiment, the centroid in the point cloud of each tree is obtained, the centroid of the tree point cloud is taken as the centroid of the initial columnar region, the preset height and the preset radius are taken as the height and the bottom radius of the initial columnar region respectively, and the columnar regions with the same size as the initial columnar region are constructed above and below the initial columnar region, the distance between adjacent columnar regions is 0, until the whole tree point cloud is traversed, and the columnar region of each tree point cloud is obtained.

[0049] It should be noted that the values of the preset height and the preset radius are artificially set, and in the embodiment, the value of the preset height is 0.5 m, and the value of the preset radius is 0.5 m. In actual application, as other implementation manners, the implementer can also set it by himself according to the specific situation, and the embodiment does not make special limitation.

[0050] Among them, the method for obtaining the centroid is a known technology, and the specific obtaining process will not be repeated.

[0051] Further, the spatial complexity of each tree is determined based on the confusion degree of all data points in the point cloud of each tree, and specifically:

[0052] In the embodiment, the information entropy of the coordinates of all data points in the point cloud of each tree is taken as the spatial complexity of each tree.

[0053] It should be noted that the specific calculation process of the information entropy of the data point coordinates in the embodiment is as follows: the maximum value and the minimum value are selected from the X, Y and Z coordinates of all data points in the point cloud of each tree, a minimum cuboid bounding box that can cover all data points is determined, the bounding box is divided into B equal cubes, the value of B in the embodiment is 1000, each cube is numbered, the probability of all data points in the point cloud of each tree falling into each cube is counted, and the information entropy is calculated based on the obtained probability and the information entropy formula, so that the information entropy of the coordinates of all data points in the point cloud of each tree, that is, the spatial complexity of each tree, is obtained.

[0054] Among them, the calculation method of the information entropy is a known technology, and the specific calculation process will not be repeated.

[0055] According to the spatial complexity of each tree, it can be understood that the spatial complexity is used to represent the richness of the internal structure of the tree point cloud and the unpredictability of the spatial arrangement. If the information entropy of the coordinates of all data points in the current tree point cloud is larger, that is, the spatial complexity is larger, it means that the distribution of all data points in the current tree point cloud in the space is more chaotic and complex, and there is no obvious regularity, which indicates that the branches and leaves of the tree are very dense, the crown structure is extremely rich, and the form is irregular and the information amount is large. On the contrary, if the information entropy of the coordinates of all data points in the current tree point cloud is smaller, that is, the spatial complexity is smaller, it means that the distribution of all data points in the current tree point cloud in the space is more orderly and simple, and presents obvious regularity, which indicates that the branches and leaves of the tree are relatively sparse.

[0056] Further, based on the distance from all data points in any cylindrical region of each tree point cloud to the centroid of the cylindrical region, the point cloud distribution characteristic value of any cylindrical region of each tree point cloud is determined, specifically:

[0057] In this embodiment, the mean value of the distance from all data points in any cylindrical region of each tree point cloud to the centroid of the cylindrical region is taken as the result of normalization, as the point cloud distribution characteristic value of any cylindrical region of each tree point cloud.

[0058] According to the point cloud distribution characteristic value of any cylindrical region of each tree point cloud, it can be understood that the point cloud distribution characteristic value is used to represent the tightness of the distribution of data points in the tree point cloud. If the distance from all data points in any cylindrical region of the current tree point cloud to the centroid of the cylindrical region is smaller, it means that the distribution of data points in the cylindrical region is more dense, which indicates that the branches and leaves of the tree are very dense, the gap between point clouds is small, and the corresponding point cloud distribution characteristic value is smaller. On the contrary, if the distance from all data points in any cylindrical region of the current tree point cloud to the centroid of the cylindrical region is larger, it means that the distribution of data points in the cylindrical region is more sparse, which indicates that the branches and leaves of the tree are relatively sparse, the gap between point clouds is large, and the corresponding point cloud distribution characteristic value is larger.

[0059] Further, based on the point cloud distribution characteristic value of any cylindrical region of each tree point cloud, and in combination with the coordinate difference of all data points in any cylindrical region of each tree point cloud, the spatial distribution characteristic value of each tree is determined, specifically:

[0060] In this embodiment, the range of the Z-axis coordinate values of all data points in any cylindrical region of each tree point cloud is calculated, and the ratio of the range to the corresponding point cloud distribution characteristic value is calculated, which is recorded as the point cloud distribution value of any cylindrical region of each tree point cloud.

[0061] The mean value of the point cloud distribution values of all cylindrical regions of each tree point cloud is taken as the spatial distribution characteristic value of each tree.

[0062] According to the spatial distribution characteristic value of each tree, it can be understood that the spatial distribution characteristic value is used to represent the coupling characteristics of the vertical fluctuation intensity and the horizontal aggregation tightness of the tree point cloud in the local space. If the range of the Z-axis coordinate values of all data points in any columnar region of the current tree point cloud is larger, it means that the point cloud in the columnar region spans a larger range in the vertical direction, which may represent dense branches and leaves, branched trunks, and very dramatic fluctuations. Therefore, the corresponding spatial distribution characteristic value will also increase, indicating that the current tree has strong vertical structural complexity. At the same time, if the distance from all data points in the current tree to the centroid of the columnar region is smaller, it means that the point cloud distribution of the current tree in the columnar region is denser, and the corresponding spatial distribution characteristic value is larger, indicating that the current tree has strong vertical complex structure, and the corresponding fine grid is needed to improve the modeling accuracy.

[0063] On the contrary, if the range of the Z-axis coordinate values of all data points in any columnar region of the current tree point cloud is smaller, it means that the point cloud in the columnar region is relatively flat in the vertical direction. Therefore, the corresponding spatial distribution characteristic value will also decrease, indicating that the vertical structure of the current tree is relatively simple. At the same time, if the distance from all data points in the current tree to the centroid of the columnar region is larger, it means that the point cloud distribution of the current tree in the columnar region is sparser, and the corresponding spatial distribution characteristic value is smaller, indicating that the vertical structure of the current tree is simple, and a larger grid can be used to improve the modeling efficiency.

[0064] Further, based on the spatial complexity and the spatial distribution characteristic value, the point cloud turbulence degree of each tree is determined, specifically:

[0065] In this embodiment, the result of positively fusing the spatial complexity and the spatial distribution characteristic value of each tree point cloud is taken as the point cloud turbulence degree of each tree.

[0066] Preferably, the point cloud turbulence degree extraction process diagram provided by the embodiment is as shown in Figure 2 .

[0067] It should be understood that positive fusion refers to combining two or more indicators together through addition or multiplication, so as to obtain a comprehensive indicator, so as to more comprehensively and accurately evaluate a phenomenon or problem. This fusion method is not limited to simple arithmetic operations, but can also include more complex statistical models and analysis methods, and the implementer can select them according to the specific circumstances, and the embodiment does not make special limitations.

[0068] Preferably, in this embodiment, the product of the spatial complexity and the spatial distribution characteristic value of each tree point cloud is taken as the point cloud turbulence of each tree. In actual application, as other implementation manners, the implementer can also use other positive fusion methods such as sum according to specific circumstances, and this embodiment does not make special restrictions.

[0069] According to the point cloud turbulence of each tree, it can be understood that the point cloud turbulence is used to quantify the overall set complexity and information density of a single tree, and reflects whether the tree needs to be finely modeled. If the spatial complexity of the current tree is larger, it means that the internal structure of the current tree is complex, which will increase the point cloud turbulence, indicating that the grid of the current tree needs to be refined. At the same time, if the spatial distribution characteristic value of the current tree is larger, it means that the current tree not only has complex shape and dense branches and leaves, but also has uneven surface of the canopy layer, which contains rich geometric details. Correspondingly, the point cloud turbulence is larger, and the grid should be refined when the current tree is rasterized.

[0070] On the contrary, if the spatial complexity of the current tree is smaller, it means that the internal structure of the current tree is simple, which will reduce the point cloud turbulence, indicating that the grid of the current tree can be coarsened. At the same time, if the spatial distribution characteristic value of the current tree is smaller, it means that the current tree not only has single shape and sparse branches and leaves, but also has relatively smooth surface of the canopy layer, which contains less geometric details. Correspondingly, the point cloud turbulence is smaller, and a larger grid can be used to improve the modeling efficiency when the current tree is rasterized.

[0071] So far, by analyzing the spatial complexity of the overall structural disorder degree of the tree, and combining the local vertical relief and horizontal aggregation characteristics, the point cloud turbulence is constructed, the geometric details and information density of a single tree are comprehensively evaluated, and the basis for subsequent adaptive rasterization is provided. The complex trees can be intelligently refined with fine grids to retain key features, and the simple trees can be used with rough grids to improve efficiency.

[0072] S2.2 Based on the distance between each tree in each region and each tree in its neighborhood, and combining the point cloud turbulence, the tree distribution density in each region is determined to adjust the grid window size of each region in the three-dimensional point cloud of the target forest area.

[0073] In the gridding processing of the terrain point cloud of the forest area, the determination of the grid size needs to comprehensively evaluate two core dimensions: the complexity of the single-tree structure and the spatial distribution density of the forest, specifically, the more complex the point cloud morphology of a single tree is, such as the more branches and leaves and the irregular crown layer, the richer the collective characteristics of the single tree are; meanwhile, the closer the spatial distance between the single tree and the neighboring trees is, the denser the distribution of the forest in the region is, when both conditions are met, i.e., the single tree with high complexity is in the region with high density, the terrain and vegetation coupling characteristics of the region are also more significant, therefore, a smaller grid unit must be used to finely depict the details such as the accurate contour of the tree, the gap between the tree crowns and the micro-topography under the forest, so as to ensure the fidelity of the modeling.

[0074] Therefore, based on the above analysis, in the embodiment, the distribution density of the trees in each region is determined based on the distance between each tree in the region and each tree in the neighborhood of the tree, and in combination with the point cloud disorder degree of the tree, and the grid window size of each region in the three-dimensional point cloud of the target forest area is adjusted, specifically:

[0075] As an implementation manner, in the embodiment, the distribution density of the trees in the region g is The expression of the distribution density of the trees in the region g is: ; in the expression, the point cloud disorder degree of the tree k in the neighborhood of the tree t in the region g is represented by ; the distance between the tree t in the region g and the tree k in the neighborhood of the tree t is represented by ; the number of all the trees in the neighborhood of the tree t is represented by ; the number of all the trees in the region g is represented by ; and the normalization function is represented by norm().

[0076] According to the distribution density of the trees in each region, it can be understood that the distribution density of the trees reflects the comprehensive spatial congestion degree and the scene complexity of each region; the greater the point cloud disorder degree of the tree k in the neighborhood of the tree t in the region g is, the more complex the geometric morphology of the tree t in the region g is, which directly increases the distribution density of the trees; meanwhile, the smaller the distance between the tree t in the region g and the tree k in the neighborhood of the tree t is, the closer the spatial distance between the trees is, the more crowded the arrangement is, and the greater the distribution density of the trees is; the greater the distribution density of the trees is, the more information density the region g is composed of, which means that the current region contains a large number of details that need to be finely captured, such as the complex crown layer intersection, the gap between the tree crowns and the micro-topography under the forest;

[0077] Conversely, if the point cloud disorder of the tree k in the neighborhood of the tth tree in the region g is smaller, it means that the geometric shape of the tree t in the region g is simpler, which will directly reduce the tree distribution density; at the same time, if the distance between the tth tree in the region g and the tree k in its neighborhood is larger, it means that the spatial distance between the trees is farther, the arrangement is sparser, and the corresponding tree distribution density is smaller; if the tree distribution density is smaller, it means that the region g is a low information density region composed of simple and dispersedly arranged trees, which means that the current region contains less detailed features, the scene is relatively simple, and it is suitable for simplification processing to improve the modeling efficiency.

[0078] Further, in the GIS modeling of the forest area, the grid size should adopt an adaptive strategy to balance accuracy and efficiency. For the area with dense trees, the crown layer structure is complex, the geometric shape is rich and fluctuates significantly, and contains a large amount of detailed information. In order to accurately capture the key features such as tree boundary, crown layer structure and even understory micro-topography, smaller grid units must be used to improve the realism and fidelity of the model. On the contrary, in the area with sparse trees or flat terrain, the feature information is relatively simple, so larger grid windows can be used to reduce data redundancy and improve modeling efficiency on the premise of ensuring basic terrain features.

[0079] Based on the above analysis, the embodiment adjusts the grid window size of each region in the three-dimensional point cloud of the target forest area based on the tree distribution density in each region. Specifically:

[0080] In the embodiment, the length of the square grid window of the region g in the three-dimensional point cloud of the target forest area The expression is: ; In the formula, represents the tree distribution density in the region g; represents the preset initial square grid window length; represents the preset adjustment coefficient.

[0081] It should be noted that the value of the preset initial square grid window length and the value of the preset adjustment coefficient are artificially set. In the embodiment, the value of the preset initial square grid window length is 1 m, and the value of the preset adjustment coefficient is 0.5. In actual application, as other implementation manners, the implementer can also set it according to the specific situation, and the embodiment does not make special limitation.

[0082] According to the length of the square grid window in each region in the three-dimensional point cloud of the target forest area, it can be understood that the greater the tree distribution density in the region g, the more complex the tree distribution in the region g, and the more details need to be retained. Therefore, the length of the grid window should be adjusted to retain the details. The greater the tree distribution density, the smaller the corresponding grid window length. Conversely, the smaller the tree distribution density in the region g, the simpler the tree distribution in the region g, and the fewer details need to be retained. Therefore, the length of the grid window should be adjusted to simplify the model and improve efficiency. The smaller the tree distribution density, the greater the corresponding grid window length.

[0083] So far, the embodiment constructs the tree distribution density based on the point cloud disorder degree of single trees and the spatial distance between trees, and adaptively adjusts the grid window size. In the high-information area composed of complex and dense trees, small grids are automatically used to finely depict key details such as canopy interlacing and understory micro-topography. In the simple and sparse area, large grids are automatically switched to simplify the model and improve efficiency, so as to achieve intelligent balance between precision and performance in forest area three-dimensional modeling.

[0084] Step S3: Based on the three-dimensional point cloud after gridding, the GIS model of the target forest area is constructed.

[0085] According to the adaptive square grid length calculated in step S2, the three-dimensional point cloud of the target forest area is divided into a regular grid to generate a preliminary digital elevation model. The inverse distance weighted method is used for interpolation in each grid of the digital elevation model to accurately calculate the elevation value and retain local topographic features. Further, the standard deviation method is used to identify abnormal elevation points in the interpolation result, and the mean value of the surrounding eight neighborhood grids is used to replace the abnormal elevation points to correct data errors. Then, a moving window mean filter is used to smooth the digital elevation model after the elevation value is refined and corrected to eliminate local noise and optimize the model surface. Finally, the optimized pure digital elevation model is fused with other forestry information such as tree clusters and vegetation types to obtain a GIS model integrating terrain and vegetation information of the target forest area.

[0086] Among them, DEM gridding, inverse distance weighted interpolation, standard deviation method to identify outliers, moving window mean filter and GIS model construction are all known technologies, and their specific principles and processes are not repeated here.

[0087] So far, by evaluating the complexity of the forest area and adaptively adjusting the grid size, the embodiment can automatically use small grids in complex areas to retain precision, and switch to large grids in simple areas to improve efficiency, thereby balancing the modeling precision and computing performance, and improving the precision and efficiency of the digital terrain resource GIS model construction.

[0088] It should be noted that the above-mentioned order of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes specific embodiments of the present specification. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0089] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

[0090] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not limit them; modifying the technical solutions described in the above embodiments, or equivalently replacing some technical features, does not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for building a digital terrain resource GIS model, characterized in that, The method comprises the following steps: Obtaining a three-dimensional point cloud of a target forest area; Dividing the three-dimensional point cloud into multiple regions by using a region growing algorithm, and extracting the point cloud of each tree in each region; segmenting the point cloud of each tree into multiple cylindrical regions, determining the spatial complexity of each tree based on the confusion degree of all data points in the point cloud of each tree, determining the point cloud distribution characteristic value of any cylindrical region of the point cloud of each tree based on the distance of all data points in any cylindrical region of the point cloud of each tree to the centroid of the cylindrical region, and combining the coordinate difference of all data points in any cylindrical region of the point cloud of each tree to determine the spatial distribution characteristic value of each tree; and determining the point cloud turbulence of each tree based on the spatial complexity and the spatial distribution characteristic value; Determining the tree distribution density in each region based on the distance between each tree in each region and each tree in its neighborhood, and combining the point cloud turbulence to adjust the grid window size of each region in the three-dimensional point cloud of the target forest area; Constructing a GIS model of the target forest area based on the rasterized three-dimensional point cloud; The point cloud turbulence of each tree is the positive fusion result of the spatial complexity and the spatial distribution characteristic value of the point cloud of each tree. The expression of the tree distribution density in each region is: ; wherein, represents the tree distribution density in region g; represents the point cloud disorder of tree k in the neighborhood of the tth tree in region g; represents the distance between the tth tree in region g and tree k in its neighborhood; represents the number of all trees in the neighborhood of the tth tree; represents the number of all trees in region g; and norm() represents a normalization function.

2. The method for constructing a digital terrain resource GIS model as described in claim 1, characterized in that, The three-dimensional point cloud of the target forest area contains elevation information, and the three-dimensional point cloud of the target forest area is taken as the input of the region growing algorithm, wherein the data point with the maximum elevation value in the three-dimensional point cloud is taken as the initial seed point, and all regions of the three-dimensional point cloud are output. The point cloud of each tree is segmented into multiple cylindrical regions, including:

3. The method of claim 1, wherein the digital terrain resource GIS model is constructed by: Obtaining the centroid in the point cloud of each tree, taking the centroid of the tree point cloud as the centroid of the initial cylindrical region, taking the preset height and the preset radius as the height and the bottom radius of the initial cylindrical region, respectively, constructing cylindrical regions with the same size as the initial cylindrical region above and below the initial cylindrical region, and the distance between adjacent cylindrical regions is 0, until the whole tree point cloud is traversed, to obtain the cylindrical regions of the point cloud of each tree. The spatial complexity of each tree is the information entropy of the coordinates of all data points in the point cloud of each tree.

4. The method for constructing a digital terrain resource GIS model as described in claim 1, characterized in that, The point cloud distribution characteristic value of any cylindrical region of the point cloud of each tree is the result of taking the mean value of the distance of all data points in any cylindrical region of the point cloud of each tree to the centroid of the cylindrical region to the normalized value.

5. The method for constructing a digital terrain resource GIS model as described in claim 1, characterized in that, The determination method of the spatial distribution characteristic value of each tree is:

6. The method of claim 1, wherein the digital terrain resource GIS model is constructed by: Calculating the range of the Z-axis coordinate values of all data points in any cylindrical region of the point cloud of each tree, calculating the ratio of the range to the corresponding point cloud distribution characteristic value, and recording it as the point cloud distribution value of any cylindrical region of the point cloud of each tree; The mean value of the point cloud distribution values of all cylindrical regions of the point cloud of each tree is taken as the spatial distribution characteristic value of each tree. The adjustment of the grid window size of each region in the three-dimensional point cloud of the target forest area comprises:

7. The method for constructing a digital terrain resource GIS model as described in claim 1, characterized in that, The construction of the GIS model of the target forest area comprises: a square grid window length of a region g in a three-dimensional point cloud of a target forest area The expression is: ; in which, represents a tree distribution density in the region g; represents a preset initial square grid window length; represents a preset adjustment coefficient.

8. The method of claim 1, wherein the digital terrain resource GIS model is constructed by: ​ The three-dimensional point cloud is rastered according to a grid window size to preliminarily generate a rastered digital elevation model, the elevation value of each grid in the digital elevation model is refined by using an inverse distance weighted difference value method, and the forestry information collected is fused to obtain a GIS model of the target forest area.

Images