A method for mapping forestry resources based on remote sensing data

Abstract

The present invention provides a method for mapping forestry resources based on remote sensing data, which relates to the technical field of data processing. The intersection points of a marginal line and a grid are used as known marginal points, and the maximum value of the area of ​​a triangle is used to determine an unknown marginal point between two adjacent known marginal points on the marginal line; based on the marginal points, the surface vertices are determined according to the shape and area of ​​the forest area surface, and the lines between the surface vertices and the marginal points form a grid. Based on the shape and area of ​​the forest area surface, the number of grids is optimized, the surface vertices and the marginal points are increased or decreased, and the spatial positions of the surface vertices and the marginal points are adjusted to achieve optimization of the number and spatial positions of the surface vertices and the marginal points, and then the grid within the marginal line is optimized, and the addition and adjustment of the grid are limited within the marginal line, which is conducive to reducing the amount of calculation and improving the efficiency of adjusting the grid; the accuracy and efficiency of mapping the remote sensing data of the forest area to the grid are improved.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, and in particular to a method for mapping forestry resources based on remote sensing data. Background Art

[0002] In the existing technology, the forestry resources of the forest area are counted through the model. Since the three-dimensional model is an overall model of a certain forest area, the area of ​​the forest area can range from a few hectares to hundreds of thousands of hectares. In the actual use process, it is necessary to monitor a forest area of ​​a specific range, for example, to monitor the forestry resources of a forest area of ​​one hectare or five hectares. The grid of the model cannot coincide with the forestry boundary line actually monitored, and it is difficult for the model to accurately obtain the boundary of the monitored area. Mapping the remote sensing data to the grid of the three-dimensional model will cause data distortion. Since the surface of the forest area is a curved surface, the grid of the model is commonly triangles and quadrilaterals, and the grid of the model is a plane. When there is an order of magnitude difference between the modeling area and the monitoring area of ​​the forest area, the errors in area and shape between the model grid and the surface of the forest area are magnified.

[0003] In view of this, this application is hereby filed. Summary of the invention

[0004] In view of the above problems existing in the prior art, a method for mapping forestry resources based on remote sensing data is provided, the purpose of which is to solve at least one of the above problems.

[0005] The technical solution to achieve the purpose of the present invention is:

[0006] The present invention provides a method for mapping forestry resources based on remote sensing data, which is implemented by a server and a client. The specific steps for implementing the method are as follows:

[0007] In the first step, the server uses supervised learning to build a model based on the remote sensing data of the forest area and forms the grid parameters of the model; based on the grid, the remote sensing data is classified and digitized by unsupervised learning and supervised learning, and mapping parameters are formed; supervised learning is used to map the remote sensing data to the grid one by one in real time according to the coordinates, and the historical mapping parameters are stored; the server updates the grid parameters and mapping parameters to the client in real time; the client collects the boundary line of the monitored forest area by supervised learning, and updates the boundary line to the server in real time;

[0008] In the second step, a continuous and closed edge line is determined by the motion trajectory of the client, and the edge line is on the forest area surface. Then, the server selects edge points on the edge line to express the shape of the edge line, and at least three edge points are used as a plane. The depth value of the forest area surface is calculated with the plane as a reference plane. Points that can express the shape and have depth values ​​are selected as surface vertices in the edge line. New grids are formed based on the surface vertices and edge points. The differences in area and shape before and after meshing of the forest area surface are compared in the edge line, and the number of grids is optimized, and then the positions of the surface vertices and edge points are adjusted.

[0009] The third step is to load the real-time mapping parameters into the grid and collect the real-time statistics.

[0010] Compared with the prior art, the beneficial effects of the present invention include:

[0011] (1) Classify and digitize satellite remote sensing data and load coordinate information to form mapping parameters. The first function is to establish a mapping relationship and improve the accuracy of data mapping. The second function is to facilitate data loading. The third function is to facilitate data classification, entry and statistics.

[0012] (2) The grid is combined with the spatial coordinates and parameterized, which is conducive to efficient and flexible fine-tuning of the model grid;

[0013] (3) The depth value of the forest surface is calculated based on the plane formed by the edge points. Since the relative difference is calculated, the calculation complexity and amount can be reduced. The surface vertices are determined based on the edge points according to the shape and area of ​​the forest surface. The lines between the surface vertices and the edge points form a grid. The number of grids is optimized based on the shape and area of ​​the forest surface. The surface vertices and edge points are increased or decreased, and the spatial positions of the surface vertices and edge points are adjusted to achieve optimization of the number and spatial positions of the surface vertices and edge points. Then, the grid within the edge line is optimized. The addition and adjustment of the grid are limited to the edge line, which is conducive to reducing the amount of calculation and improving the efficiency of adjusting the grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 It is a schematic diagram of the intersection of the edge line and the grid;

[0016] Figure 2It is a schematic diagram of the marginal lines and grids to determine the marginal points;

[0017] Figure 3 It is a schematic diagram of the concave or convex points of the edge line;

[0018] Figure 4 It is a schematic diagram of determining the marginal points by the concave or convex points of the marginal line. DETAILED DESCRIPTION

[0019] To make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described clearly and completely in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them.

[0020] Therefore, the following detailed description of the embodiments of the present invention is not intended to limit the scope of the invention claimed for protection, but merely represents some embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features and technical solutions in the embodiments may be combined with each other.

[0022] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0023] The present invention is further described in detail below with reference to the embodiments.

[0024] like Figures 1 to 4 As shown, the present invention provides a method for mapping forestry resources based on remote sensing data, which is implemented by a server and a client. The specific steps for implementing the method are as follows:

[0025] In the first step, the server uses supervised learning to build a model based on the remote sensing data of the forest area and form the grid parameters of the model; based on the grid, the remote sensing data is classified and digitized by unsupervised learning and supervised learning, and mapping parameters are formed; supervised learning is used to map the remote sensing data to the grid one by one in real time according to the coordinates, and the historical mapping parameters are stored; the server updates the grid parameters and mapping parameters to the client in real time; the client collects the boundary lines of the monitored forest area through supervised learning, and updates the boundary lines to the server in real time; the server and the client update data with each other in real time.

[0026] It should be noted that in the prior art, after the forest surface is modeled, the mesh of the model is solidified and cannot be modified. Generally speaking, the modeling is established according to the entire area of ​​a certain forest area. During the model establishment process, the approximation principle is followed and a plane triangular or quadrilateral mesh is used to express the surface. The specific forest monitoring area is compared with the entire area of ​​a certain forest area, and its area differs by several orders of magnitude. For example, 2 or 3 orders of magnitude, the area of ​​the forest monitoring area is 5 square kilometers, and the modeled forest area is 5,000 square kilometers. Of course, the area of ​​the forest area generally ranges from a few hectares to hundreds of thousands of hectares. There is an order of magnitude difference between the modeled area and the actual monitoring area, which causes the model's mesh to be unable to accurately express the surface shape and area of ​​the forest monitoring area, and thus makes the monitoring data inaccurate. In view of this problem, the present invention adopts a supervised learning management model on the server side and the client side, so that it has the function of adjusting the model grid, and both the server side and the client side have the ability to independently adjust the model grid; in the conventional working mode, the server side assumes the adjustment function of the model grid, because the adjustment of the grid requires strong computing power support, and at the same time, the remote sensing data is received on the server side, and the remote sensing data includes image data; by combining supervised learning and unsupervised learning to process remote sensing data, the server side can effectively extract and process remote sensing data, and quickly identify different types of objects in the forest area, such as forest types, rivers, roads, etc., and the efficient processing of data can reduce manual intervention and improve the speed of data analysis; the server side can not only process current data, but also learn past data patterns to predict future ecological changes. The server side can map remote sensing data to the constructed model grid in real time, thereby providing accurate monitoring of forest resources. Parameter transmission is adopted between the server side and the client side to reduce the amount of data transmission and improve the efficiency of data transmission. Real-time data update between the server and the client ensures that all participants can obtain the latest forest information in real time, and the updated data includes grid parameters, mapping parameters and margin lines.

[0027] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the server side constructs a model based on forest remote sensing data using supervised learning, and optimizes the grid parameters of the model; the specific steps are as follows:

[0028] Step A1: Data collection and preprocessing; obtain remote sensing data of forest areas, including spectral images, climate data, land use data, etc.; process missing values ​​and remove outliers; extract meaningful features, such as NDVI (Normalized Difference Vegetation Index), EVI (Enhanced Vegetation Index), etc.; prepare sample labels for supervised learning and mark them according to actual survey results or existing classification information;

[0029] Step A2: Data division: divide the data set into training set and test set. The common ratio is 70% for training and 30% for testing, or use cross-validation technology.

[0030] Step A3: Select the model; choose an appropriate supervised learning algorithm, such as Random Forest, Support Vector Machine (SVM), Gradient Boosting, Neural Network, etc.

[0031] Step A4, Grid Search: Define the parameter grid, specifically the parameter range for the selected model; for random forests, this includes the number of trees, maximum depth, minimum number of sample splits, etc.; select evaluation criteria, such as accuracy, score, etc.; perform grid search: use cross-validation to evaluate different parameter combinations and find the best parameters;

[0032] Step A5: training the model; using the best parameters to train the model and ensure that the model fits the training set well;

[0033] Step A6, model evaluation: evaluate the model performance on the test set and check the prediction effect; generate confusion matrix, ROC curve, etc. to show the model performance in a more intuitive way; ROC curve (Receiver Operating Characteristic Curve) is a graphical tool used to evaluate the performance of classification models, and is often used for binary classification problems;

[0034] Step A7: Visualize the results; compare the predicted results of the visualization model with the actual results; make heat maps, classification maps, etc. to show the forest classification in different areas;

[0035] Step A8: Adjustment and optimization: further adjust data features or model parameters based on the evaluation results of the model;

[0036] Step A9: deploy the model; deploy the trained model to the server, and the server updates the model parameters to the client.

[0037] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, after the remote sensing data is further classified and digitized, the coordinate information is loaded in the mapping parameters; the specific steps are as follows:

[0038] Step B1: Acquire remote sensing images and form time series images; use high-resolution satellite images to obtain multispectral images of the forest area, and obtain images of different periods to obtain images of dynamic changes in the monitored forest area;

[0039] Step B2: Correct the image and perform regional cropping; perform atmospheric correction, geometric correction, and radiation correction to ensure data consistency and accuracy; crop to include only the forest area of ​​interest to reduce data volume and computational burden;

[0040] Step B3, feature extraction: use information from different bands (such as visible light, near infrared, etc.) to extract features, such as calculating vegetation indices (such as NDVI); generate feature maps based on different bands or analysis results to provide a basis for subsequent classification;

[0041] Step B4, classification; classification includes supervised classification and unsupervised classification; supervised classification mainly classifies data in a conventional way, collects ground measured data for training classification models; trains models and classifies images; verifies classification results, and adjusts model parameters to improve classification accuracy. Unsupervised classification mainly mines potential data and logical relationships in remote sensing data, uses K-means, ISODATA and other algorithms for preliminary classification; applies unsupervised classification algorithms to preliminarily classify images; merges or refines categories based on preliminary results;

[0042] Step B5: digitization: digitize the classification results to generate vector data (such as lines and surfaces) to represent different forest area types; assign attribute information to each vector object, such as forest type, tree age, health status, etc.;

[0043] Step B6, post-processing: post-process the classification results, use filtering, error correction and other means to eliminate isolated pixels and improve the smoothness of classification; merge or split the classification results according to the actual situation of the plot to make them more consistent with the actual situation of the forest area;

[0044] Step B7: Result verification and correction; use confusion matrix, Kappa coefficient and other methods to evaluate the accuracy of classification;

[0045] Step B8, after the remote sensing data is classified and digitized, the coordinate information is loaded into the mapping parameters, which is conducive to quickly and accurately mapping the mapping parameters into the grid.

[0046] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, a coordinate system is used to collect and confirm the boundary line of the monitored forest area; the specific steps are as follows:

[0047] Step C1, select the coordinate system; use geographic coordinate systems (such as WGS84) and projected coordinate systems (such as UTM) to ensure spatial consistency of data;

[0048] Step C2: Data collection: Use satellites, aerial photography or drones to obtain high-resolution image data of the corresponding area; conduct field surveys, measurements and obtain accurate margin line locations. Beidou equipment can be used to collect location information.

[0049] Step C3: Image processing and edge line extraction. Use image processing software, such as ENVI, ArcGIS, etc., to pre-process the image, including denoising and enhancement. Use supervised learning to automatically extract edge lines, and apply edge detection, segmentation and classification techniques to automatically identify edge lines in the image. Advanced algorithms such as deep learning can be used to further improve extraction accuracy.

[0050] Step C4, coordinate marking and verification: record the coordinates of the extracted boundary line and convert them into a specific coordinate format (such as longitude and latitude, metric system, etc.); compare the ground measured data with the remote sensing results to ensure the accuracy of the boundary line;

[0051] Step C5 transmits the boundary line to the server; classifies and digitizes the satellite remote sensing data and loads the coordinate information to form mapping parameters. The first function is to establish a mapping relationship and improve the accuracy of data mapping. The second function is to facilitate data loading. The third function is to facilitate data classification, entry and statistics.

[0052] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, several types of information on forest resources are further effectively monitored and evaluated; the information includes: forest coverage, forest health, forest type and structure, forest biomass and carbon storage, forest change monitoring, fire monitoring and management, soil moisture and water conditions and biodiversity monitoring.

[0053] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the first step, further, the mesh is combined with the spatial coordinates and parameterized; it should be noted that the mesh is a geometric structure used to represent a three-dimensional object. It consists of vertices, edges and faces. Common mesh types include triangular meshes and quadrilateral meshes. Spatial coordinates are numerical values ​​used to describe the position of an object in three-dimensional space. In a three-dimensional coordinate system, coordinates (x, y, z) are usually used to represent the position of a point. Each vertex has a corresponding spatial coordinate. Parameterization is the representation of the geometric shape of an object as a function of one or more parameters. It makes the calculation and deformation of complex shapes simpler. For example, two parameters (u, v) can be used to describe the two-dimensional surface of a mesh, and then these parameters are mapped to points in three-dimensional space.

[0054] Specifically, define the vertices of the mesh and specify the spatial coordinates of each vertex; define the faces of the mesh and determine which vertices constitute a face; determine the parameter range, associate parameterized coordinates (u, v) with each vertex, and define the parameter domain for the entire mesh; create a mapping function and use the function to map the parameters (u, v) to the coordinates (x, y, z) in three-dimensional space. Use parameterized coordinates to generate or modify the mesh through transformation or interpolation; dynamically deform the mesh by modifying the parameterized coordinates; use parameterization to correctly map textures to three-dimensional surfaces; and facilitate efficient and flexible fine-tuning of the model mesh.

[0055] In the second step, a continuous and closed edge line is determined by the motion trajectory of the client, and the edge line is on the surface of the forest area. Then, the server selects edge points on the edge line to express the shape of the edge line, and at least three edge points are used as a plane. The depth value of the forest area surface is calculated with the plane as the reference plane. Points that can express the shape and have depth values ​​are selected within the edge line as surface vertices, and new grids are formed based on the surface vertices and edge points. The differences in area and shape of the forest area surface before and after meshing are compared within the edge line, and the number of grids is optimized, and then the positions of the surface vertices and edge points are adjusted.

[0056] It should be noted that since the edge line is on the forest surface, it is not on a plane, but exists in three-dimensional space. Any three edge points constitute a plane. Of course, there are more than three edge points that constitute a plane. Taking the plane as the reference plane, the distance height between the point on the forest surface and the plane is calculated, and the distance height is the depth value. The depth value of the forest surface is calculated based on the plane formed by the edge points. Since the relative difference is calculated, the calculation complexity and amount can be reduced. Based on the edge points, the surface vertices are determined according to the shape and area of ​​the forest surface. The lines between the surface vertices and the edge points form a grid. Since each grid is composed of triangles or quadrilaterals, triangles and quadrilaterals are planes. Several grids are used to simulate the shape of the forest surface and the areas are close to equal, such as 600 grids or 800 grids. By calculating the curvature of the forest surface, curvature parameters are provided for the grid curvature (not a surface in the strict sense). After the grid is formed, it is compared with the shape of the forest surface, that is, the curvature of the corresponding position is compared. The number of grids is determined based on the shape and area of ​​the forest surface. Optimization is carried out, that is to say, the number of grids should be reduced. The areas of several grids and the area of ​​the corresponding forest surface are calculated by integration. At the same time, the grid can express the shape and area of ​​the forest surface, increase or decrease the surface vertices and edge points, and adjust the spatial positions of the surface vertices and edge points to achieve optimization in quantity and spatial position of the surface vertices and edge points, and then optimize the grids within the marginal line. The addition and adjustment of the grids are mainly limited to the marginal line, which is conducive to reducing the amount of calculation. At the same time, the original grid is basically retained (the original vertices remain basically unchanged) and the efficiency of adjusting the grid is improved.

[0057] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, a combination of supervised learning and manual methods is further used to determine the boundary line; it should be noted that the motion trajectory of the client is not equal to the boundary line, and manual fine-tuning is required to determine the forest area that needs to be monitored.

[0058] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, further, supervised learning is used to determine continuous and closed edge lines, and the specific steps are as follows:

[0059] Step D1, collecting the motion trajectory data of the client; the motion trajectory data includes a series of coordinate points in the form of (x1, y1), (x2, y2), …, (xn, yn), where n represents a natural number;

[0060] Step D2: preprocess the data; use filtering algorithms (such as moving average or median filtering) to reduce noise in the motion trajectory; normalize the coordinates to ensure that all coordinates are within the same scale range for subsequent processing;

[0061] Step D3, closing the trajectory; adding the starting point of the motion trajectory as the last point to the trajectory;

[0062] Step D4, creating edge lines; using the collected motion trajectory points to create continuous edge lines, using interpolation techniques (such as spline interpolation and Bezier curves) to create smooth closed edge lines. The spline curves, such as the Catmull-Rom spline curve, can effectively smooth the trajectory. The moving average can insert additional points between the trajectory points to make the edge lines smoother.

[0063] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, further, the intersection point of the edge line and the grid is used as a known edge point, or / and, the convex point and / or concave point on the edge line is used as a known edge point, and the maximum value of the triangle area is used to determine the unknown edge point between two adjacent known edge points on the edge line. It should be noted that the unknown edge points are determined from the known edge points, and the determined unknown edge points are converted into known edge points, so as to iteratively calculate and improve the accuracy of the edge points in expressing the edge line; combined with Figure 1 and Figure 2 , part of the marginal line L intersects with a triangular mesh W, and the intersection point is the first marginal point L1 and the second marginal point L2. If there is a line segment (straight) between the first marginal point L1 and the second marginal point L2, then the first marginal point L1 and the second marginal point L2 can accurately express this marginal line. If there is a curve between the first marginal point L1 and the second marginal point L2, then the first marginal point L1 and the second marginal point L2 cannot accurately express this marginal line. The first marginal point L1 and the second marginal point L2 are the two vertices of the triangle. A point is taken on the curve between the first marginal point L1 and the second marginal point L2 to determine the third marginal point L3, and the area of ​​the triangle △L1L2L3 is the largest, and △ represents a triangle. Combined with Figure 3 and Figure 4 , determine the convex points or concave points on part of the marginal line L (the convex points or concave points can be converted according to the observed direction) as the fourth marginal point L4 and the fifth marginal point L5, the fourth marginal point L4 and the fifth marginal point L5 are used as the two vertices of the triangle, and take a point on the curve between the fourth marginal point L4 and the fifth marginal point L5 to determine the sixth marginal point L6, and the area of ​​the triangle △L4L5L6 is the largest, and △ represents a triangle.

[0064] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, supervised learning is further used to increase or decrease vertices and edge points, and the spatial positions of vertices and edge points are adjusted. It should be noted that in order to reduce the number of grids while still expressing the shape and area of ​​the surface of the forest area, the surface vertices and edge points are increased or decreased, and their spatial positions are adjusted. Reduce the number of grids by removing unnecessary vertices and edge points; use a grid simplification algorithm (e.g., the Douglas-Peucker algorithm) to perform this operation. Add vertices and edge points in areas that require more accuracy (such as steep slopes or complex terrain); help improve the local resolution of the grid. Move surface vertices to improve the shape and quality of the grid, which includes moving vertices to a smoother position or removing vertices that cause the grid to distort. Move edge points to adjust the range and resolution of the grid, which includes expanding the grid to cover more areas or reducing the grid to focus on specific areas.

[0065] In order to better achieve the purpose of the present invention, in some embodiments of the present invention, in the second step, further, the addition and adjustment of the mesh is limited to the boundary line; further reduce the amount of calculation and improve efficiency. Only move vertices and edge points within the boundary line to ensure that the movement of all vertices and edge points is limited to the boundary line; only add or remove vertices and edge points within the boundary line; only add or remove vertices and edge points within the boundary line to expand or reduce the mesh. Optimize surface vertices and edge points to reduce the number of meshes while still accurately representing the shape and area of ​​the forest surface. The optimized mesh will have the characteristics of minimizing the number of meshes, accurately capturing the surface shape, reasonable computational cost and being limited to the boundary line. Reduce the amount of calculation and improve the efficiency of mesh adjustment while still providing an accurate expression of the forest surface.

[0066] The third step is to load the real-time mapping parameters into the grid and collect real-time statistics. It should be noted that the real-time mapping parameters are loaded into the grid, and the real-time mapping parameters to be loaded into the grid are determined, including canopy coverage, tree species composition, tree height or other related data; the real-time data is loaded into the grid layer; the data is associated with the grid cells or vertices using spatial positions; the grid layer is visualized and the real-time mapping parameter data is analyzed. Select the grid cells or vertices to be analyzed based on spatial position, attribute value or other criteria. Perform statistical analysis on the mapping parameter data of the selected grid cells or vertices using a geographic information system (GIS) (such as summing, averaging or calculating percentages). Visualize the statistical analysis results as charts, tables or maps. It helps to understand the distribution and trend of forestry resource data. By loading the real-time mapping parameters into the grid and performing real-time statistical analysis, dynamic and up-to-date information on forestry resource data is obtained. It helps forestry management, resource planning and decision making. For example, real-time canopy coverage data is used to monitor deforestation and identify affected areas; real-time tree species composition data is used to assess the distribution of different tree species and identify dominant tree species; real-time tree height data is combined with established biomass models to estimate the biomass of the forest stand; real-time canopy coverage and tree species composition data are used to predict fire risks and identify high-risk areas. It is applied to forestry resource management to improve efficiency, improve decision-making and promote forest sustainability.

[0067] It should be noted that the technical solution of the present invention is applicable to both forest areas and grasslands.

[0068] It should be noted that for technical features that are not fully explained, conventional technical means are used.

[0069] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described in the present invention. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the above specific implementation methods. Therefore, any modification or equivalent replacement of the present invention; and all technical solutions and improvements thereof that do not depart from the spirit and scope of the invention are included in the scope of the claims of the present invention.

Claims

1. A method for mapping forestry resources based on remote sensing data, which is implemented by a server and a client, and is characterized in that: The specific steps to implement the method are as follows: In the first step, the server uses supervised learning to build a model based on the remote sensing data of the forest area and forms the grid parameters of the model; based on the grid, the remote sensing data is classified and digitized by unsupervised learning and supervised learning, and mapping parameters are formed; Use supervised learning to map remote sensing data to grids one by one in real time according to coordinates, and store historical mapping parameters; the server side updates the grid parameters and mapping parameters to the client side in real time; The client collects and monitors the boundary lines of the forest area through supervised learning, and updates the boundary lines to the server in real time; The server and client update data in real time with each other; The boundary line is transmitted to the server, the satellite remote sensing data is classified and digitized, and the coordinate information is loaded to form mapping parameters; the coordinate information is loaded in the mapping parameters, and the specific steps are as follows: Step B1, acquiring remote sensing images and forming time series images; Step B2: correcting the image and performing region cropping; Step B3, feature extraction, generates feature maps based on different bands or analysis results to provide a basis for subsequent classification; Step B4, classification, which includes supervised classification and unsupervised classification; Step B5, digitization, digitizes the classification results to generate vector data to represent different forest area types; Step B6, post-processing, post-processes the classification results, uses filtering and error correction to eliminate isolated pixels and improve the smoothness of the classification; merges or splits the classification results according to the actual situation of the plot; Step B7, result verification and correction; Step B8, after classifying and digitizing the remote sensing data, load the coordinate information in the mapping parameters; In the second step, a continuous and closed edge line is determined by the motion trajectory of the client, and the edge line is on the forest area surface. Then, the server selects edge points on the edge line to express the shape of the edge line, and at least three edge points are used as a plane. The depth value of the forest area surface is calculated with the plane as a reference plane. Points that can express the shape and have depth values ​​are selected as surface vertices in the edge line. New grids are formed based on the surface vertices and edge points. The differences in area and shape before and after the forest area surface is meshed are compared in the edge line, and then the positions of the surface vertices and edge points are adjusted. The third step is to load the real-time mapping parameters into the grid and collect the real-time statistics.

2. A method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In the first step, the coordinate information is loaded in the mapping parameters.

3. A method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In a first step, the mesh is combined with spatial coordinates and parameterized.

4. The method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In the second step, the margin line is determined by combining supervised learning and manual work.

5. The method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In the second step, the intersection points of the edge lines and the grid are taken as known edge points.

6. The method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In the second step, the convex points and / or concave points on the edge line are taken as known edge points.

7. A method for mapping forestry resources based on remote sensing data according to claim 5 or 6, characterized in that: In the second step, the unknown edge point is determined between two adjacent known edge points on the edge line using the maximum value of the triangle area.

8. The method for mapping forestry resources based on remote sensing data according to claim 1, characterized in that: In the second step, the additions and adjustments to the grid are confined within the margin lines.

Images