DEM-based multi-dimensional remote sensing omnidirectional fire monitoring method, device and medium

Abstract

The application discloses a kind of multi-dimensional remote sensing all-around fire monitoring method, equipment and medium based on DEM, by constructing DEM and DSM, obtain canopy height model CHM, according to canopy feature point construction irregular triangle network TIN, reconstruct CHM according to original data scale based on TIN, according to original CHM and reconstruction DMH, judge vegetation change amount, using discriminant analysis method, screening vegetation change amount, obtain significant factor, construct discriminant function model, train the model, according to the discriminant function model after training, judge whether the vegetation change amount in t time period is abnormal, according to the abnormal situation of vegetation change amount in t time period, judge whether there is fire, considering the complexity of topography, determine current fire and fire spread situation in combination with topography and wind potential, improve the monitoring accuracy and prediction accuracy of fire, at the same time, reduce the situation of not monitoring in place caused by dense vegetation coverage, the extraction error rate is high.

Description

Technical Field

[0001] This invention relates to the field of forest fire technology, specifically to a multi-dimensional remote sensing all-round fire monitoring method, equipment, and medium based on DEM. Background Technology

[0002] Given the inherent danger of forest fires, advance forest control is necessary. However, due to the complexity of forest terrain, predictions of fires and their spread are affected by factors such as wind and vegetation. Different vegetation types and levels of vegetation, influenced by terrain, can have varying impacts on the occurrence and spread of fires. Current fire monitoring technologies fail to consider the complexity of forest terrain, vegetation conditions, and wind speed, resulting in inadequate monitoring. Particularly in vegetation data extraction, the extraction of tree height, canopy base height, base diameter, and branch distribution is often ineffective in densely vegetated areas. The high vegetation cover and vigorous growth lead to a high error rate, making it impossible to determine the fire's severity based on vegetation changes in densely vegetated areas, resulting in low monitoring and prediction accuracy. Summary of the Invention

[0003] The technical problem this invention aims to solve is the inability to determine the magnitude of a fire in densely vegetated areas based on changes in vegetation, resulting in low monitoring accuracy. The goal is to provide a multi-dimensional remote sensing all-round fire monitoring method, equipment, and medium based on a DEM (Digital Elevation Model). By constructing a DEM and DSM (Digital Elevation Model), a canopy height model (CHM) is obtained. The CHM is reconstructed using canopy feature points. Based on the original CHM and the reconstructed DMH, vegetation changes are determined. Variables exhibiting vegetation changes are screened, and a discriminant function model is constructed using discriminant analysis to improve the accuracy of identifying anomalies in vegetation changes and thus enhance monitoring precision.

[0004] This invention is achieved through the following technical solution:

[0005] The first aspect of this invention provides a multi-dimensional remote sensing all-round fire monitoring method based on DEM, comprising the following specific steps:

[0006] S1. Acquire image data of the target monitoring area in real time, preprocess the image data, construct a digital elevation model (DEM) and a digital surface model (DSM), and construct a canopy height model (CHM) based on the DEM and DSM.

[0007] S2. Obtain the total number of CHM grids and the number of extracted feature points, determine the canopy feature points, construct an irregular triangular network (TIN) based on the canopy feature points, reconstruct the CHM based on the TIN at the original data scale, and determine the vegetation change within time period t based on the vegetation amount of the grid corresponding to the original CHM and the vegetation amount of the grid corresponding to the reconstructed CHM.

[0008] S3. Using discriminant analysis, the vegetation change was screened to obtain significant factors, a discriminant function model was constructed, and the model was trained.

[0009] S4. Based on the trained discriminant function model, determine whether the vegetation change within time period t is abnormal;

[0010] S5. If the vegetation in the target monitoring area changes abnormally, obtain the current terrain and wind speed based on the DEM model and DSM model, and determine the current fire situation and fire spread by combining the terrain, wind speed and vegetation change.

[0011] This invention constructs a Canopy Height Model (CHM) by building a Canopy Scale Model (DEM) and a Canopy Scale Model (DSM). The CHM is then reconstructed using canopy feature points. Based on the original CHM and the reconstructed DMH, vegetation change is determined. Variables related to vegetation change are screened, and a discriminant function model is constructed using discriminant analysis. This improves the accuracy of identifying anomalies in vegetation change, enhances monitoring precision, and reduces the risk of inadequate monitoring and high error rates caused by dense vegetation cover.

[0012] Furthermore, S1 specifically includes:

[0013] Acquire the original image data of the target monitoring area and determine the optimal pose of the original image;

[0014] Construct a 3D point cloud based on the optimal pose of the original image;

[0015] DEM and DSM models are generated using 3D point clouds.

[0016] Furthermore, S2 specifically includes:

[0017] Obtain the total number of CHM grids and the number of extracted feature points to determine canopy feature points;

[0018] An irregular triangular network (TIN) is constructed based on canopy feature points, and the CHM is reconstructed based on the TIN at the scale of the original data.

[0019] Based on the vegetation amount of the original CHM grid and the vegetation amount of the reconstructed CHM grid, the vegetation change within time period t is determined.

[0020] Furthermore, the steps for obtaining the vegetation coverage of the CHM-corresponding grid specifically include:

[0021] Convert CHM to a heatmap to obtain RGB image data;

[0022] Multiple directions are used to perform gray-level co-occurrence matrix operations on RGB image data, and the spectral parameters of each band are calculated based on the gray-level co-occurrence matrix.

[0023] Perform spectral characteristic parameter calculations to obtain spectral parameters;

[0024] The vegetation type and quantity are obtained, and the vegetation amount in the target monitoring area is assessed by combining spectral characteristics.

[0025] Furthermore, the calculation of spectral characteristic parameters to obtain spectral parameters specifically includes:

[0026] The continuous projection method is used to eliminate redundant information in the spectral matrix;

[0027] Select an initial iteration vector in the spectral matrix, and automatically generate a set of the remaining vectors other than the initial iteration vector; iteratively calculate the projection of the initial vector onto each vector in the set, and introduce the maximum wavelength of the projected vector into the feature combination;

[0028] Unnecessary information in the spectral data is removed based on feature combinations, and then the effective features of the data are extracted.

[0029] Furthermore, S3 specifically includes:

[0030] To obtain the parameters of vegetation change variables, one variable is introduced at a time, and the variable with the strongest discriminative power among the introduced variables is selected as the first significant factor.

[0031] Repeat the above steps to select the second and third significant factors in turn;

[0032] Test whether the discriminant ability of significant factors has improved. If not, discard them from the discriminant; if so, retain them.

[0033] The screening process continues until all factors in the discriminant have significant discriminative power and no other variables are introduced, at which point the screening ends and a discriminant function model is constructed.

[0034] The training set and prediction set are divided, the newly introduced vegetation change variable parameters are discriminated and analyzed, the newly introduced parameters are classified, the discriminant function model is updated, and the trained discriminant function model is obtained.

[0035] Furthermore, S4 specifically includes:

[0036] Based on the DEM, the terrain of the monitoring target and the terrain of the vegetation growth area are obtained;

[0037] Based on historical climate data and combined with the topography of the vegetation growth area, historical data on vegetation change were obtained to determine the change threshold.

[0038] Furthermore, the acquisition of the current terrain based on the DEM model and DSM model specifically includes:

[0039] Identify areas of abnormal vegetation change, obtain the boundaries of these areas, and obtain the grid center points of the DEM within these areas.

[0040] Random points within the grid are used as the dataset of unknown elevation points to construct multiple terrain training sample subsets, and each terrain training sample subset forms a decision tree.

[0041] The terrain surface is globally fitted using a randomly selected subset of samples, and an elevation prediction value is output for random points.

[0042] A random forest model is constructed based on decision trees. The mean of several elevation predictions for random points by all decision trees in the random forest is calculated to obtain the final elevation value of the random point.

[0043] The mean elevation of random points within the grid is calculated by iterating through the grid. The difference between the elevation values ​​of the grid points and the mean is used as the grid importance measurement information SI.

[0044] Set a threshold to filter out grid points where |SI|≥ the threshold, and determine the terrain feature points.

[0045] The method for obtaining the current wind force based on the DEM model and DSM model specifically includes:

[0046] Historical average wind speed is obtained, and the wind speed in the target monitoring area is interpolated using the Kriging interpolation method. A wind speed grid layer is generated based on the altitude.

[0047] Perform raster operations on the grid layer using the exponential law formula;

[0048] Obtain wind speed distribution maps at different grid elevations;

[0049] Based on the terrain features, obtain the slope, aspect, and position to determine the wind force of the target grid cell.

[0050] Furthermore, the determination of the current fire situation and its spread, based on the combination of terrain, wind speed, and vegetation changes, specifically includes:

[0051] Obtain current and historical wind speeds, acquire biological changes over multiple time periods (t), and assess fire intensity.

[0052] Obtain the current wind direction and, in conjunction with terrain features, determine the direction of the fire;

[0053] Obtain the vegetation growth in the predicted fire trajectory area, and combine this with the vegetation flammability to predict the fire intensity in the fire trajectory area.

[0054] A second aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a multi-dimensional remote sensing all-round fire monitoring method based on DEM.

[0055] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a multi-dimensional remote sensing all-round fire monitoring method based on DEM.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0057] By constructing DEM and DSM, a canopy height model CHM is obtained. The CHM is reconstructed using canopy feature points. Based on the original CHM and the reconstructed DMH, vegetation change is determined. Variables of vegetation change are screened, and a discriminant function model is constructed using discriminant analysis to improve the accuracy of identifying anomalies in vegetation change. This improves monitoring accuracy and reduces the situation of inadequate monitoring and high extraction error rate caused by dense vegetation cover. Attached Figure Description

[0058] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0059] Figure 1 This is a flowchart of the fire monitoring method in an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0061] Example 1

[0062] like Figure 1 As shown, the first aspect of this embodiment provides a multi-dimensional remote sensing all-round fire monitoring method based on DEM, including the following specific steps:

[0063] S1. Acquire image data of the target monitoring area in real time, preprocess the image data, construct a digital elevation model (DEM) and a digital surface model (DSM), and construct a canopy height model (CHM) based on the DEM and DSM.

[0064] S2. Obtain the total number of CHM grids and the number of extracted feature points, determine the canopy feature points, construct an irregular triangular network (TIN) based on the canopy feature points, reconstruct the CHM based on the TIN at the original data scale, and determine the vegetation change within time period t based on the vegetation amount of the grid corresponding to the original CHM and the vegetation amount of the grid corresponding to the reconstructed CHM.

[0065] S3. Using discriminant analysis, the vegetation change was screened to obtain significant factors, a discriminant function model was constructed, and the model was trained.

[0066] S4. Based on the trained discriminant function model, determine whether the vegetation change within time period t is abnormal;

[0067] S5. If the vegetation in the target monitoring area changes abnormally, obtain the current terrain and wind speed based on the DEM model and DSM model, and determine the current fire situation and fire spread by combining the terrain, wind speed and vegetation change.

[0068] This invention constructs a Digital Elevation Model (DEM) and a Digital Surface Model (DSM) to obtain a Canopy Height Model (CHM), where CHM = DEM - DSM. The CHM is reconstructed using canopy feature points. Based on the original CHM and the reconstructed DSM, vegetation changes are determined. Variables related to vegetation changes are filtered, and a discriminant function model is constructed using discriminant analysis. This improves the accuracy of identifying anomalies in vegetation changes, enhances monitoring precision, and reduces the risk of inadequate monitoring and high error rates caused by dense vegetation cover.

[0069] In some possible embodiments, S1 specifically includes: acquiring original image data of the target monitoring area, estimating the optimal pose of the original image using SLAM technology; constructing a three-dimensional point cloud based on the optimal pose of the original image; and generating a DEM model and a DSM model using the three-dimensional point cloud.

[0070] In some possible embodiments, S2 specifically includes: obtaining the total number of CHM grids and the number of extracted feature points, and determining the canopy feature points: Wherein, the simplification rate 𝑃 quantitatively describes the retention of canopy feature points, M is the number of extracted feature points, and N is the total number of original CHM rasters.

[0071]

[0072] Furthermore, two error indices, Mean Absolute Error (MAE) and Mean Square Error (MSE), were used to quantitatively describe the terrain simplification effect. MAE measures the degree of similarity between the reconstructed CHM and the original CHM, while MSE reflects the dispersion of the checkpoint elevation deviation. Their calculation formulas are shown in equations (2-2) and (2-3), respectively. Here, 𝑍𝑜𝑖 is the elevation value of the 𝑖th grid cell in the original CHM, 𝑍𝑒𝑖 is the elevation value of the 𝑖th grid cell in the reconstructed CHM, and 𝑛 is the total number of grid cells.

[0073] A higher simplification rate indicates a higher degree of simplification in the CHM and fewer retained canopy feature points; a lower simplification rate indicates a lower degree of simplification in the CHM and more retained canopy feature points. An irregular triangular network (TIN) is constructed based on the canopy feature points, and the CHM is reconstructed based on the TIN at the original data scale. The vegetation change within time period t is determined by comparing the vegetation cover of the grid corresponding to the original CHM with that of the reconstructed CHM grid.

[0074] In some possible embodiments, the step of obtaining vegetation amount corresponding to the CHM grid specifically includes: converting the CHM into a heat map, the heat map including the pixel values ​​of each of the red, green and blue bands, and obtaining RGB image data based on the CHM heat map, the RGB image data including the pixel values ​​of each of the red, green and blue bands of the RGB image.

[0075] The RGB image data is processed by fusing the pixel values ​​of the CHM heatmap with those of the RGB image to obtain the pixel values ​​of the red, green, and blue bands of the fused image. Multiple gray-level co-occurrence matrix (GLCM) operations are performed on the fused image, and the spectral parameters of each band are calculated based on the GLCM. Spectral feature parameters are calculated, specifically using a continuous projection method to eliminate redundant information in the spectral matrix. An initial iteration vector is selected in the spectral matrix, and the remaining vectors are automatically generated into a set. The projection of the initial vector onto each vector in the set is calculated iteratively, and the maximum wavelength of the projected vector is incorporated into the feature combination. Unnecessary information in the spectral data is removed based on the feature combination, and the effective features are extracted. Vegetation type and quantity are obtained, and the vegetation cover in the target monitoring area is assessed by combining spectral features. By converting the CHM into a heatmap for vegetation monitoring, the monitoring of vegetation growth and fire occurrence can be conducted more clearly and efficiently.

[0076] In some possible embodiments, S3 specifically includes: acquiring the topography of the monitoring target based on DEM, acquiring the topography of the vegetation growth area; acquiring historical data on vegetation change monitoring based on historical climate data and the topography of the vegetation growth area, determining the change threshold, monitoring the abnormality of vegetation change based on the vegetation growth change data of the same historical period, determining the disaster situation based on the abnormal change of vegetation, and improving the detection accuracy.

[0077] In some possible embodiments, the importance of the DEM grid is quantitatively assessed by extracting terrain feature points, calculating the mean elevation of random points within each grid, and subtracting the mean elevation of the center point of the original DEM grid from the mean elevation of the corresponding random points within the grid. This difference describes the degree of terrain undulation within the grid and can be used as an indicator of grid importance. The calculation formula includes: Where D is the elevation value of the initial DEM grid center point, Di is the elevation value of a random point within the DEM grid, and n is the number of random points. The larger the terrain undulation within the grid, the larger the SI; the flatter the terrain, the smaller the SI. Therefore, grid points with a greater impact on the terrain structure can be extracted as terrain feature points by setting a threshold for SI, i.e., grid points with |SI|≥ the threshold are considered terrain feature points.

[0078] The current terrain is obtained based on the DEM and DSM models, specifically including: identifying areas of abnormal vegetation change, obtaining area boundaries, and obtaining the grid center points of the DEM within these areas; using random points within the grid as a dataset of unknown elevation points, constructing multiple terrain training sample subsets, with each subset forming a decision tree; globally fitting the terrain surface using the randomly selected sample subsets, outputting an elevation prediction value for each random point; constructing a random forest model based on the decision trees, calculating the mean of several elevation prediction values ​​for random points from all decision trees in the random forest, and obtaining the final elevation value of the random point; iterating through and calculating the mean elevation of random points within the grid, subtracting the grid point's elevation value from this mean, and using the difference as the grid importance metric (SI); setting a threshold, filtering out grid points where |SI|≥ the threshold, and determining terrain feature points.

[0079] In some possible embodiments, the current wind speed is obtained based on the DEM model and DSM model, specifically including: obtaining the historical average wind speed, interpolating the wind speed of the target monitoring area using the Kriging interpolation method, generating a wind speed grid layer based on the altitude; performing grid operations on the grid layer according to the exponential law formula; obtaining the wind speed distribution map at different grid elevations; obtaining the slope, aspect and position based on the terrain characteristics, and determining the wind speed of the target grid cell.

[0080] In some possible embodiments, determining the current fire intensity and fire spread by combining topography, wind speed, and vegetation changes specifically includes: obtaining the current wind speed and historical wind speed; obtaining biological changes over multiple time periods t to determine the fire intensity; obtaining the current wind direction and, in combination with topographic features, determining the fire direction; obtaining the vegetation growth in the predicted fire direction area and, in combination with the flammability of the vegetation, predicting the fire intensity in the fire direction area.

[0081] The process of obtaining the predicted fire trajectory includes: acquiring an image of the current fire area, dividing the image into grids, obtaining the number of flame pixels in each grid, filtering the pixel values ​​to prevent data distortion due to noise interference, eigenvalue-encoding the divided grid flame image to form a feature matrix, and using the feature matrix and the number of pixels in each grid to predict the size and trajectory of the flame.

[0082] The fire intensity prediction method for a region based on vegetation flammability includes: obtaining the fire direction, obtaining the terrain in the direction of the fire, and combining the vegetation growth data monitored by the terrain, including vegetation growth status, vegetation type, and vegetation water storage in the current season, to comprehensively determine the vegetation flammability and predict the fire size.

[0083] The second aspect of this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a multi-dimensional remote sensing all-round fire monitoring method based on DEM.

[0084] The third aspect of this embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a multi-dimensional remote sensing all-round fire monitoring method based on DEM.

[0085] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A DEM-based multi-dimensional remote sensing all-around fire monitoring method, characterized in that, The method comprises the following specific steps: S1, real-time acquisition of target monitoring area image data, pre-processing of the image data, construction of a digital elevation model DEM and a digital surface model DSM, and construction of a canopy height model CHM based on the DEM and the DSM; S2, acquisition of the total number of CHM grid and the number of extracted feature points, determination of canopy feature points, construction of a triangulated irregular network TIN based on the canopy feature points, reconstruction of the CHM based on the TIN according to the original data scale, and determination of the vegetation change amount in the t period according to the vegetation amount of the original CHM corresponding grid and the vegetation amount of the reconstructed CHM corresponding grid; S3, using discriminant analysis to screen the vegetation change amount, obtaining significant factors, and constructing a discriminant function model and training the model; S4, judging whether the vegetation change amount in the t period is abnormal according to the trained discriminant function model; S5, if the vegetation amount change of the target monitoring area is abnormal, obtaining the current terrain and wind potential based on the DEM model and the DSM model, combining the terrain, wind potential and vegetation change amount to determine the current fire potential and fire spread situation; Wherein, the current terrain is obtained based on the DEM model and the DSM model, specifically including: Obtain the vegetation change amount abnormal area, obtain the region boundary, and obtain the DEM grid center point in the region; Using random points in the grid as elevation unknown point data set, a plurality of terrain training sample subsets are constructed, and each terrain training sample subset forms a decision tree; A random sample subset is used to globally fit the terrain surface, and an elevation prediction value is output for the random point; A random forest model is constructed according to the decision tree, the mean value of the several elevation prediction values of all decision trees in the random forest for the random point is calculated, and the final elevation value of the random point is obtained; The mean value of the elevations of the random points in the grid is calculated, the elevation value of the grid point is subtracted from the mean value, and the difference is taken as the grid importance measurement information SI; A threshold is set, and the grid points with |SI| greater than or equal to the threshold are screened out to determine the terrain feature points; The current wind potential is obtained based on the DEM model and the DSM model, specifically including: Obtain the historical average wind speed, interpolate the wind speed of the target monitoring area by Kriging interpolation method, and generate a wind speed grid layer according to the elevation; Performing grid operation on the grid layer according to the exponential rate formula; Obtaining the wind speed distribution diagram at different grid elevations; According to the terrain characteristics, the slope, slope direction and slope position are obtained, and the wind potential of the target grid unit is determined. 2.The DEM-based multi-dimensional remote sensing all-around fire monitoring method according to claim 1, characterized in that, The S1 specifically includes: Obtaining the original image data of the target monitoring area, and determining the optimal pose of the original image; Based on the optimal pose of the original image, a three-dimensional point cloud is constructed; The DEM model and the DSM model are generated by using the three-dimensional point cloud. 3.The DEM-based multi-dimensional remote sensing all-around fire monitoring method according to claim 1, characterized in that, The CHM corresponding grid vegetation amount acquisition step specifically includes: Convert the CHM into a heat map to obtain RGB image data; Perform gray level co-occurrence matrix operation on the RGB image data in multiple directions, and calculate the spectral parameters of each band according to the gray level co-occurrence matrix; Perform spectral feature parameter calculation to obtain spectral parameters; Obtain the vegetation type and quantity combined with the spectral feature to evaluate the vegetation amount of the target monitoring area. 4.The DEM-based multi-dimensional remote sensing all-around fire monitoring method according to claim 3, characterized in that, The spectral feature parameter calculation to obtain spectral parameters specifically includes: The continuous projection method is used to eliminate the redundant information in the spectral matrix. An initial iteration vector is selected in the spectral matrix, and a set of remaining vectors other than the initial iteration vector is automatically generated; the projection of the initial vector on each vector in the set is calculated in a loop, and the maximum wavelength of the projection vector is introduced into the characteristic combination; According to the characteristic combination, unnecessary information in the spectral data is removed, and the effective features of the data are extracted. 5.The DEM-based multi-dimensional remote sensing omnidirectional fire monitoring method according to claim 1, wherein, The S3 specifically comprises: Obtaining a vegetation change variable parameter, introducing one variable each time, and taking the variable with the most discriminant ability in the introduced variable as a first significant factor, Repeating the above steps to sequentially select a second significant factor and a third significant factor; Testing whether the discriminant ability of the significant factor is improved, if not, discarding it from the discriminant formula, if yes, retaining it; Until all factors in the discriminant formula have significant discriminant ability and no other variable is introduced, the screening is ended, and a discriminant function model is constructed; Dividing the training set and the prediction set, discriminating and analyzing the newly introduced vegetation change variable parameter, classifying the newly introduced parameter, and updating the discriminant function model. 6.The DEM-based multi-dimensional remote sensing omnidirectional fire monitoring method according to claim 1, wherein, The S4 specifically comprises: Based on DEM, obtaining the terrain of the monitoring target and the terrain of the vegetation growth area; Based on historical climate data, combining the terrain of the vegetation growth area to obtain the monitoring historical data of the vegetation change, and determining the change threshold. 7.The DEM-based multi-dimensional remote sensing omnidirectional fire monitoring method according to claim 1, wherein, Based on the combination of topography, wind potential and vegetation change, the determination of the current fire potential and the fire spread situation specifically comprises: Obtaining the current wind speed and the historical wind speed, obtaining the biological change amount in multiple t periods, and judging the fire potential; Obtaining the current wind direction, combining the terrain characteristics, and judging the fire direction; Obtaining the vegetation growth amount of the predicted fire direction area, combining the flammability of the vegetation, and predicting the fire potential of the fire direction area.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the DEM-based multi-dimensional remote sensing all-around fire monitoring method of any one of claims 1 to 7.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to realize the DEM-based multi-dimensional remote sensing all-around fire monitoring method of any one of claims 1 to 7.

Images