Video linkage emergency rescue management method based on GIS multi-modal technology
By quantifying fire risk values and optimizing video source scheduling through GIS multimodal technology, the shortcomings in fire risk monitoring boundary delineation were resolved, and the adaptive linkage and spatial coverage optimization of video sources were achieved, eliminating monitoring blind spots and redundancy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHANGCHUN SHOUJIA TECH CO LTD
- Filing Date
- 2026-05-09
- Publication Date
- 2026-06-09
AI Technical Summary
Existing technologies cannot quantify and delineate fire hazard monitoring boundaries, resulting in insufficient adaptive linkage scheduling and spatial coverage optimization of video sources. This leads to problems such as blind spots in open flame areas, ineffective coverage in low-risk areas, and redundant monitoring by multiple devices.
Based on GIS multimodal technology, by quantifying terrain slope, combustible material type and proportion, and combining real-time wind speed and direction, fire risk values are determined, and fire risk gradient contour lines are constructed to optimize the viewpoint allocation and scheduling scheme of video sources, thereby eliminating monitoring blind spots and redundancy.
It achieves full-space quantification of fire hazard status, forms a quantified monitoring boundary that matches the fire spread pattern, eliminates blind spots in open flame areas and ineffective coverage in low-risk areas, optimizes the linkage scheduling of video sources, and avoids redundant monitoring by equipment.
Smart Images
Figure CN122175301A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of rescue management technology, specifically a video-linked emergency rescue management method based on GIS multimodal technology. Background Technology
[0002] In actual forest fire emergency monitoring scenarios, fire spread is influenced by the coupling of multiple physical fields, resulting in a non-uniform and dynamic spatial distribution of fire-prone areas. Existing technologies exhibit the following shortcomings: First, they fail to couple and quantify terrain slope, combustible material type and proportion, and real-time wind speed and direction, making it impossible to match the actual spatial distribution patterns of fire risks. This leads to a lack of objective quantitative basis for delineating control areas, resulting in severely insufficient boundary accuracy. Second, existing solutions lack video source selection and perspective allocation mechanisms adapted to quantified fire-prone areas. They cannot achieve coordinated scheduling and coverage optimization of video equipment based on the spatial characteristics of fire risks. This results in blind spots due to the inability to achieve comprehensive coverage of high-risk open flame areas, and also easily leads to ineffective monitoring coverage in low-risk areas. Furthermore, irregular scheduling of multiple devices causes large-scale duplicate monitoring, resulting in unreasonable allocation of monitoring resources.
[0003] Therefore, there is an urgent need for a GIS-based multimodal video-linked emergency rescue management method that can quantify fire risk values, delineate fire risk control boundaries, and achieve video source adaptability linkage scheduling and spatial coverage optimization to solve the above problems. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a video-linked emergency rescue management method based on GIS multimodal technology. This method solves the problem that existing technologies are unable to quantitatively delineate fire risk monitoring boundaries and control areas based on terrain slope, type and proportion of combustibles, and real-time wind speed and direction. Consequently, they cannot achieve adaptive linkage scheduling and spatial coverage optimization of video sources, resulting in the coexistence of blind spots in open flame areas, ineffective coverage in low-risk areas, and redundant monitoring by multiple devices.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: a video-linked emergency rescue management method based on GIS multimodal technology, comprising the following steps: for any grid cell, determining the baseline fire risk value based on terrain slope, combustible material type and proportion, and determining the fire risk value based on wind speed and direction, combined with fire intensity and the incremental fire risk of spread.
[0006] Based on the terrain slope and wind direction, fire risk gradient contour lines are constructed within the GIS platform, and the inner area is used as a control zone.
[0007] Based on terrain slope and combustible material type, the visibility of schedulable video equipment is analyzed to determine candidate video sources, and the fire risk coverage contribution, spatial straight-line distance and monitoring demand matching degree of the corresponding control zone are calculated.
[0008] Based on the contribution of fire risk coverage, spatial straight-line distance, and the matching degree of monitoring needs, and combined with the spatial distribution of fire risk values, the candidate video sources are assigned viewing angles to determine the video scheduling scheme.
[0009] Based on the baseline fire hazard value, the incremental fire spread hazard, and wind speed and direction, a fire spread path raster set is generated in the GIS platform. Coverage and redundancy are checked in conjunction with the video scheduling scheme, the viewpoint allocation is optimized, and the optimal scheduling scheme is determined.
[0010] Compared with existing technologies, this invention has the following advantages: First, it quantifies the baseline fire hazard value of the grid, then couples wind speed and direction with fire intensity to superimpose the fire hazard increment to obtain the overall fire hazard risk value, achieving full-space quantification of fire hazard status; it constructs fire hazard gradient contour lines based on the coupling of terrain slope and wind direction and delineates control zones, forming a quantified monitoring boundary that matches the fire spread pattern; it combines terrain slope and combustibles to conduct visibility analysis to screen candidate video sources and allocate initial viewing angles, achieving directional adaptation and linkage between video sources and fire hazard areas; finally, it simulates the spread path based on fire hazard parameters and conducts coverage and redundancy checks to optimize the viewing angle and optimize the initial viewing angle allocation scheme. This solves the problems of existing technologies being unable to quantify and delineate fire hazard monitoring boundaries, and unable to achieve video source adaptation linkage scheduling and spatial coverage optimization, eliminating the technical defects of blind spots in open flame areas, ineffective coverage in low-risk areas, and repeated monitoring by multiple devices. Attached Figure Description
[0011] Figure 1 This is a flowchart of the video-linked emergency rescue management method based on GIS multimodal technology according to the present invention;
[0012] Figure 2 This is a flowchart illustrating the construction of fire hazard gradient contour lines in the video-linked emergency rescue management method based on GIS multimodal technology of the present invention;
[0013] Figure 3 This is a flowchart illustrating the process of determining a video dispatching scheme in the video-linked emergency rescue management method based on GIS multimodal technology of the present invention. Detailed Implementation
[0014] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. Please refer to the accompanying drawings. Figure 1 The present invention provides a technical solution: a video-linked emergency rescue management method based on GIS multimodal technology, comprising the following steps: S1, for any grid cell in the target monitoring area, the baseline fire risk value is determined based on the terrain slope, combustible material type and proportion, and the fire risk value is determined by superimposing the spread fire risk increment on the baseline fire risk value based on wind speed and wind direction and fire intensity.
[0015] In actual forest fire monitoring, while the fire hazard level of a specific grid cell is difficult to quantify directly, it can be indirectly represented by certain parameters. The type and proportion of combustible materials determine the inherent combustion potential of a grid cell in the absence of wind and external fire sources. Quantifying the distribution of each type of combustible material locally and globally avoids the smothering of rare but highly dangerous combustible types by simple weighted averaging. Furthermore, terrain slope directly affects the speed at which fire spreads uphill and the difficulty of deploying rescue equipment. A steeper slope results in faster fire spread and greater difficulty in human intervention for the same fire intensity; therefore, slope can be included as a magnifying factor in the baseline fire hazard.
[0016] The process of determining the baseline fire hazard value is as follows: S101. For each type of combustible material, the corresponding total proportion of the whole area is calculated. The total proportion of the whole area is the sum of the proportions of the corresponding combustible material type in all grid cells in the target monitoring area, and the ratio of the total number of grid cells in the target monitoring area.
[0017] S102. For any combustible material type within any grid cell, the ratio of the square of the corresponding proportion to the corresponding total proportion of the entire area is taken as the fire risk contribution.
[0018] S103. Sum the fire risk contribution items corresponding to all combustible types in each grid cell to obtain the relative combustion potential coefficient of the combustible.
[0019] S104. The arithmetic mean of the terrain slopes corresponding to all grid cells within the target monitoring area is used as the slope reference value.
[0020] S105. Calculate the amplification factor of the current grid cell's grid slope relative to the fire hazard based on the grid cell's grid slope and slope reference value.
[0021] The specific calculation formula is as follows: ,in, This is the magnification factor. For grid slope, This is the baseline value for slope. It is a very small positive number (such as 0.001, used to avoid calculation anomalies when the denominator is zero). So when the grid slope is equal to the slope reference value, the amplification factor is 1; when the grid slope is greater than the slope reference value, the amplification factor is greater than 1; when the grid slope is less than the slope reference value, the amplification factor is less than 1 but still greater than 0.
[0022] S106. Multiply the potential coefficient by the amplification coefficient to obtain the background fire hazard value of the corresponding grid cell.
[0023] Combustible material type refers to the types of vegetation or ground cover that may burn in nature, such as coniferous forests, broad-leaved forests, shrubs, herbaceous plants, and litter. Combustible material type percentage represents the percentage (or decimal form) of the area covered by a certain combustible material type within a single grid cell (the smallest spatial analysis unit, such as a 10m x 10m square). Total percentage represents the proportion of the total area covered by a certain combustible material type within the entire target monitoring area to the total area of all grid cells, reflecting the abundance or prevalence of the combustible material type globally.
[0024] Squaring the proportion of combustible material types physically corresponds to a non-linear enhancement of the dominance of each combustible material type within the grid cell. The squaring operation preserves or even amplifies the advantage of combustible materials with higher proportions, while further compressing the influence of combustible materials with lower proportions, thus reflecting the real-world law that the dominant combustible material determines the fire hazard characteristics of the grid.
[0025] The fire risk contribution item represents the degree of contribution of a certain combustible material within a grid cell to the overall background fire risk of the grid cell. The larger the value, the stronger the dominant role of the corresponding combustible material in the fire risk of the grid cell.
[0026] The relative combustion potential coefficient of combustibles represents the inherent combustion potential of combustibles from the perspective of combustibles (without considering external factors such as terrain and wind). The larger the value, the easier it is for the combustible combination in the grid to burn and the more intense the fire may be.
[0027] The grid slope amplification factor relative to fire hazard is used to quantify the amplification effect of the current grid cell's slope on fire hazard relative to the area's average slope. Physically, it represents the factor by which the slope increases the baseline fire hazard value. A factor of 1 indicates that the slope does not additionally increase fire hazard; a factor greater than 1 indicates that the slope exacerbates the fire hazard.
[0028] The baseline fire hazard value is obtained by multiplying the combustible material's combustion potential by the slope amplification factor. It represents the inherent basic fire hazard level of the grid cell itself, which is determined by the terrain and vegetation over a long period of time. The higher the value, the more likely the grid cell is to catch fire or the stronger the fire is when there is no wind or external fire source.
[0029] This embodiment introduces the ratio of the squared percentage to the total percentage across the entire region, enabling high-risk combustible material types to generate a higher fire risk contribution in local grids than common types, thus avoiding the dilution effect of global averaging in traditional weighted averaging methods. Furthermore, by treating terrain slope as an amplification factor relative to the global average baseline rather than an absolute slope value, the fire risk amplification effect of the same slope under different terrain backgrounds is comparable. This avoids marking all areas with slopes greater than a certain threshold as high fire risk; instead, it dynamically adjusts based on the normal terrain of the region, thereby more accurately delineating truly abnormally steep monitoring boundaries.
[0030] Considering that in actual fires, even if an area has a low background fire risk, if it is located downwind of a strong wind and has a nearby fire source, its instantaneous risk may be much higher than that of an area with a high background fire risk but no wind. Current technologies often consider wind speed or wind direction alone, neglecting the fire intensity of the fire source itself and the spatial matching relationship between wind direction and the direction from the fire source to the target. Therefore, it is necessary to treat each fire source as a source of influence, with its intensity determined by its own background fire risk and fire intensity, its propagation direction by the degree of matching between wind direction and the azimuth angle from the fire source to the target, and its propagation capability by wind speed. By accumulating the influence of all fire sources, the dynamic risk increment of the non-fire source grid cell is obtained, which is then added to the background value to form the comprehensive fire risk value.
[0031] Specifically, the process for determining the fire risk value is as follows: S107. If the maximum wind speed in the target monitoring area during a historical period is 0, then the relative wind speed spread coefficient of all grid cells is 0. If the maximum wind speed is not 0, then the ratio of the current wind speed to the maximum wind speed is used as the relative wind speed spread coefficient.
[0032] S108. If there are fire source grid units in the target monitoring area, the ratio of the fire intensity corresponding to the fire source grid unit to the maximum fire intensity of the fire source grid in the target monitoring area shall be used as the relative coefficient of fire intensity.
[0033] S109. For each fire source grid cell, the spatial azimuth angle from the fire source grid cell to the target grid cell is calculated using the GIS platform. Combined with the wind direction, the wind direction spread matching coefficient between the fire source grid cell and the target grid cell is calculated.
[0034] The formula for calculating the wind spread matching coefficient is: ,in, The wind direction spread matching coefficient, It is the spatial azimuth angle. This refers to the wind direction and azimuth. This indicates the angular deviation between the relative orientation of the grid and the direction of fire spread. This indicates that the angle deviation value is converted into a radian value, which satisfies the cosine function calculation specification.
[0035] S110. For each fire source grid cell, multiply the relative wind speed spread coefficient, wind direction spread matching coefficient, and relative fire intensity coefficient corresponding to the fire source grid cell to obtain the fire spread risk increment of the fire source grid cell to the target grid cell.
[0036] S111. Sum the fire spread risk increments corresponding to all fire source grid cells to obtain the fire spread risk increments corresponding to non-fire source grid cells.
[0037] S112. If there are no fire source grid units within the target monitoring area, the incremental value of the fire spread risk for all grid units is 0.
[0038] S113. Add the background fire risk value corresponding to each grid cell to the fire spread risk increment to obtain the corresponding fire risk value.
[0039] The relative wind speed spread coefficient is the ratio of the current wind speed to the historical maximum wind speed, ranging from 0 to 1. It represents the relative driving force of the current wind relative to the extreme wind conditions of the grid cell. The larger the value, the stronger the current wind and the faster the fire spreads. Fire intensity represents the heat released by the fire source grid per unit time (unit: kW / m²), physically meaning the intensity of the fire source itself. The greater the intensity, the stronger its downwind fire hazard impact. The relative fire intensity coefficient represents the ratio of the current fire intensity to the maximum value, ranging from 0 to 1, physically meaning the relative intensity of the fire source relative to the strongest fire source. The wind direction spread matching coefficient represents the degree of matching between the direction of the line connecting the fire source to the target and the wind direction: the coefficient is large when the two directions are consistent (the wind is blowing the fire directly towards the target), and small or even zero when they are perpendicular or opposite.
[0040] The fire spread risk increment component represents the additional fire risk value contributed by a single fire source to a non-fire source grid cell. Its physical meaning is the degree of risk increase to the target grid cell caused by the fire source under the wind field conditions at the current location.
[0041] In the calculation of the incremental component of fire spread risk, the relative spread coefficient of wind speed represents the probability factor that the wind has sufficient driving power, the wind direction spread matching coefficient represents the direction matching factor that the wind blows the flame from the fire source to the target direction, and the relative fire intensity coefficient represents the intensity factor that the fire source itself has sufficient energy to spread outward. Only when these three conditions are met simultaneously can the fire source pose a substantial spread threat to the target grid. Any factor being zero will result in an overall risk of zero, and any factor being too low will proportionally reduce the overall risk. Therefore, the mathematical form of multiplying the three factors is a quantitative expression of the logic of multiple conditions being met simultaneously, similar to the classic approach in fire dynamics where the spread rate is determined by the product of multiple independent factors.
[0042] The incremental spread of fire risk is the sum of the incremental components of all fire sources on the same non-fire source grid cell. In physical terms, it is the dynamic risk superimposed on the grid cell by all existing fire sources through the propagation effect of the wind field.
[0043] The fire risk value is the sum of the baseline fire risk value and the incremental fire spread risk. It represents the final real-time fire risk level of the grid cell after comprehensively considering the static terrain and combustible characteristics and the dynamic wind and fire source propagation. It is used for subsequent delineation of control zones and video scheduling.
[0044] This embodiment uses the ratio of the current wind speed to the historical maximum wind speed as the relative wind speed spread coefficient, enabling the fire risk value to adapt to the wind and climate background of different regions. This avoids the excessive expansion of the boundary in areas with frequent strong winds or the excessive narrowing of the boundary in windless areas when using absolute wind speed thresholds. By calculating the matching coefficient between the azimuth angle of each fire source pointing to the target grid and the wind direction, the fire risk value can distinguish the differences between downwind, upwind, and crosswind directions, thus allowing the control boundary to truly extend along the prevailing wind direction and contract against the wind direction. At the same time, the inherent fire risk value of the fire source itself and the relative coefficient of fire intensity are used as multiplier factors and the contributions of all fire sources are accumulated. This allows the risk value of each grid to reflect the differentiated impact of different fire sources due to the different intensity of combustion and the inherent fire risk of their location when multiple fire sources coexist, solving the defect of equalizing the influence of strong and weak fire sources.
[0045] S2. Based on the coupling relationship between terrain slope and wind direction, construct fire risk gradient contour lines corresponding to fire risk values within the GIS platform, and use the area inside the fire risk gradient contour lines as a control zone.
[0046] Because existing forest fire risk monitoring boundary delineation does not consider the topographic driving effect of slope aspect on fire spread and the airflow driving effect of wind direction on fire spread, the delineated fire risk boundaries are disconnected from the actual fire propagation patterns. Furthermore, the fixed shape of the isolines, lacking spatial deformation adaptation, makes it impossible to identify high-risk fire areas. Therefore, this embodiment first couples topographic slope aspect and wind direction to calculate the fire propagation direction, then generates initial isolines based on the median fire risk across the entire area, and finally performs unilateral translational deformation on the isoline segments according to the propagation direction, delineating the inner side of the isolines as control zones. Figure 2 As shown, the specific process is as follows: In the GIS platform, the fire risk value, terrain slope aspect, and wind direction of all grid units in the target monitoring area are bound to the grid spatial coordinates one by one, and all grid units in the target monitoring area adopt a uniform fixed grid side length.
[0047] S201. For each grid cell, taking due north as the reference azimuth, the arithmetic mean of the azimuth angle of the grid cell's terrain slope and the azimuth angle of the wind direction is taken as the direction of fire hazard propagation.
[0048] S202. Using the median value of the fire risk value of the entire target monitoring area as a benchmark, perform continuous surface fitting on the fire risk value of the entire area and extract the initial fire risk gradient contour lines corresponding to the benchmark.
[0049] S203. Match each segment of the initial fire hazard gradient contour line with the fire hazard propagation direction of the corresponding grid cell.
[0050] S204. When the fire hazard propagation direction points outside the initial fire hazard gradient contour line, the corresponding line segment is shifted outward by a fixed distance equal to the grid side length along the fire hazard propagation direction. When the fire hazard propagation direction points inward to the contour line, the corresponding line segment remains unchanged.
[0051] S205. By splicing together all the deformed line segments, a fire hazard gradient contour line is formed.
[0052] The process of fitting a continuous surface to the overall fire risk value is as follows: using the coordinates of the grid cells as the interpolation reference, the inverse distance weighted interpolation (IDW) method is employed for fitting, with fixed interpolation parameters, to generate a continuous surface without discontinuities or distortion. For example, the power exponent in the interpolation parameters is 2, and the search neighborhood consists of 12 adjacent grid cells.
[0053] The fixed grid side length is the smallest unified unit for spatial quantification of the target monitoring area; the smaller the value, the higher the spatial accuracy. The terrain slope azimuth is the north reference angle of the grid slope orientation, and the value represents the spatial direction of the slope inclination. The wind azimuth is the north reference angle of the airflow across the entire area, and the value represents the direction of wind field propagation. The fire risk propagation direction is the arithmetic mean of the slope aspect and wind azimuth; the closer the value is to the natural direction of fire spread, the stronger the directional indication of fire risk diffusion.
[0054] The median fire risk value is the intermediate critical value of the overall fire risk value. The higher the value, the higher the basic fire risk level delineated by the initial contour lines. The initial fire risk gradient contour lines are the fire risk boundary closure lines corresponding to the median value, which are the basic boundaries of the control zones. The contour line segments are the smallest spatial units after the contour lines are split, and are used to adapt the fire risk propagation direction segment by segment. The translation distance of one grid side length is the fixed quantitative length of the contour line deformation. The fixed value ensures the uniformity and reproducibility of the boundary deformation.
[0055] It should be noted that the area enclosed by the closed loop of the fire hazard gradient contour line is defined as the inner region. The fire hazard risk value of all grid cells in the inner region is greater than the median value of the overall fire hazard risk value. After the fire hazard gradient contour line is shifted in the direction of fire hazard propagation, the inner region expands synchronously. The expanded inner region is used as the fire hazard control zone.
[0056] This embodiment first calculates the fire spread direction using the arithmetic mean of topographic slope and wind direction, which aligns with the natural characteristic that forest fires are driven by both topographic slope and airflow, thus eliminating the one-sidedness in determining the spread direction. Then, it fits contour lines based on the median value of the overall fire risk, ensuring the stability of the basic control boundary. Next, it performs line segment translation only on the fire spread direction pointing outwards from the contour lines, while keeping the line segments inwards unchanged, matching the physical law that fire spreads only to the outer areas and high-risk areas are concentrated inside the contour lines. Furthermore, it uses a uniform fixed grid side length as the translation distance, achieving full quantification of fire boundary deformation, solving the problems of no standard for fire boundary delineation, uncontrollable morphology, and inability to lock high-risk areas in existing technologies.
[0057] S3. Based on terrain slope and combustible material type, perform visibility analysis on all schedulable video devices to determine candidate video sources, and calculate the fire risk coverage contribution, spatial straight-line distance and monitoring requirement matching degree of each candidate video source for the current control zone.
[0058] In the screening and quantitative assessment of forest fire risk video equipment, to eliminate invalid monitoring equipment caused by terrain obstruction, lock onto high-risk combustibles for monitoring, eliminate resource redundancy caused by complete equipment overlap, ensure no blind spots in the control zone, and establish an objective quantitative matching relationship between equipment and fire risk areas, it is necessary to consider the spatial visibility of terrain slope, the fire risk monitoring value of combustible types, the spatial overlap and redundancy of equipment monitoring range, the full coverage of the control zone, and the matching between the spatial distribution of fire risk and the spatial location of equipment. Therefore, the specific process is as follows: S301, for each schedulable video device, if there is visual obstruction caused by terrain slope within the monitoring range, it is eliminated. If there is no obstruction and the monitoring range contains the specified combustible type, it is selected as the initial video device.
[0059] S302. If multiple initially selected video devices have completely overlapping monitoring ranges, the initially selected video device with the largest sum of coverage areas for each type of combustible material within its monitoring range shall be retained to obtain the alternative video devices.
[0060] S303. If all candidate video devices can fully cover the controlled area, then all candidate video devices shall be used as candidate video sources. If they do not fully cover the controlled area, then the initially selected video devices shall be used as candidate video sources.
[0061] S304. For any candidate video source, the maximum monitoring range overlaps with the control zone. Calculate the sum of the fire risk values of the grid cells within the overlapping area, and use the ratio of this sum to the sum of the fire risk values of the entire control zone as the corresponding fire coverage contribution.
[0062] S305. Extract the coordinates of the geometric center point of the control zone, calculate the Euclidean straight-line distance between the spatial coordinates of the candidate video source and the coordinates of the geometric center point, and use it as the corresponding spatial straight-line distance.
[0063] S306. Statistically determine the coverage area ratio of combustible material types within the control zone, and statistically determine the maximum coverage area ratio of a single type of combustible material within the spatial overlap area of candidate video sources. Use the ratio of the maximum coverage area ratio to the coverage area ratio as the corresponding monitoring requirement matching degree.
[0064] Among these, terrain slope and line-of-sight obstruction refer to the spatial state where terrain undulations block the monitoring line of sight of the equipment; if obstruction exists, the equipment has no monitoring effect; designated combustible material type is a high fire risk monitoring target, and equipment containing the designated combustible material type has monitoring value; completely overlapping monitoring ranges indicate that the monitoring areas of the equipment space are 100% overlapping, which will result in resource redundancy; the sum of combustible material coverage areas is the total range of combustible materials within the equipment monitoring area, and the larger the value, the higher the fire risk monitoring value of the equipment; complete coverage of the control zone indicates that the merged range of the equipment has no blind spots enclosing the zone, which is the bottom line of monitoring effectiveness; the larger the fire risk coverage contribution value, the stronger the fire risk coverage capability; the smaller the spatial straight-line distance value, the higher the spatial response efficiency; the larger the monitoring demand matching value, the stronger the combustible material monitoring adaptability.
[0065] It should be noted that terrain slope and line-of-sight obstruction are determined using GIS raster visibility analysis. Obstruction is defined as an elevation difference between the equipment and the target raster. GIS visibility analysis uses a Digital Elevation Model (DEM) raster with a resolution of 10m, and the criterion for determining line-of-sight obstruction is an elevation difference greater than 2m between the equipment and the target raster. In S302, if the sum of the combustible material coverage areas of multiple overlapping devices is equal, the device with the smallest installation number is retained.
[0066] This embodiment first eliminates invalid devices by utilizing terrain visibility, directly avoiding monitoring failures caused by terrain blind spots; then it specifies combustible material screening devices, focusing on high-fire-risk monitoring objects and eliminating worthless devices; next, it uses combustible material coverage area quantification to screen overlapping devices, objectively eliminating resource redundancy and maximizing the retention of monitoring value; finally, it uses full-coverage verification to dynamically switch device sets, streamlining resources while avoiding monitoring blind spots.
[0067] S4. Based on the contribution of fire risk coverage, spatial straight-line distance and monitoring demand matching degree, and combined with the spatial heterogeneity distribution of fire risk values within the control zone, the candidate video sources are initially allocated perspectives to determine the video scheduling scheme of the control zone.
[0068] In the practical application of fire hazard monitoring video perspective allocation, the first goal is to objectively characterize the spatial heterogeneity of uneven fire hazard distribution within the controlled area, avoiding a disconnect between fixed zones and the actual fire hazard distribution; secondly, to quantify the spatial orientation compatibility between candidate video sources and fire hazard sub-areas, while simultaneously achieving optimal one-to-one allocation of monitoring resources; and finally, to fill monitoring gaps in areas without dedicated equipment and generate directly executable equipment scheduling parameters. For example... Figure 3 As shown, the specific process for determining the video scheduling scheme of the control zone is as follows: S401, take the fire risk value of each grid cell in the control zone as a scalar field, calculate the gradient vector and smooth it with the fire risk value as the weight to obtain the main gradient direction field.
[0069] S402. Using the local maximum point of the fire risk value as the seed point, the watershed algorithm is used to divide the control zone into multiple fire risk sub-regions, and the centroid coordinates and skewness coefficient of each fire risk sub-region are calculated.
[0070] S403. For each candidate video source and each fire risk sub-region, calculate the comprehensive pointing deviation.
[0071] S404. Determine the K value based on the arithmetic mean of the spatial straight-line distance between each candidate video source and the spatial straight-line distance between all candidate video sources within the control zone, and select the K fire risk sub-regions with the smallest comprehensive pointing deviation as candidate locking regions for the candidate video sources.
[0072] S405. Establish a bipartite graph with candidate video sources as left nodes and fire risk sub-regions as right nodes. For node pairs belonging to the candidate locked region, the edge weight is calculated by dividing the normalized product of fire risk coverage contribution and monitoring demand matching degree by the square of the normalized spatial straight-line distance. The edge weight represents the comprehensive monitoring adaptation priority of the candidate video source to the corresponding fire risk sub-region. The larger the value, the stronger the equipment's ability to cover high fire risk areas, the higher the combustible material matching degree, and the closer the deployment distance, the higher the priority should be given to assigning monitoring tasks.
[0073] S406. The Hungarian algorithm is used to solve the maximum weight matching of the bipartite graph, so that each candidate video source is independently responsible for a fire risk sub-region, and the unmatched fire risk sub-regions are marked as shared regions.
[0074] S407. For each shared region, extract the point with the largest fire risk value in the grid cell inside the adjacent matched fire risk sub-region, whose main gradient direction points to the boundary, as the guide point. Generate shared viewpoint parameters for the adjacent matched candidate video sources that cover the guide point and do not exceed the boundary of the shared region, including three fixed parameters: horizontal adjustment angle, vertical adjustment angle, and minimum zoom magnification.
[0075] S408. Set the initial viewpoint of each matched candidate video source to the centroid direction of the fire risk sub-region it is responsible for, and then use the arithmetic mean of the shared viewpoint offset vector of adjacent shared regions multiplied by the ratio of the fire risk coverage contribution of the video source to the monitoring demand matching degree as an additional offset to synthesize the final initial horizontal rotation angle, initial vertical rotation angle and initial zoom factor to form a video scheduling scheme.
[0076] The gradient vector calculation adopts the GIS four-neighbor gradient calculation method (an existing general spatial analysis method). Taking a single raster as the center, the fire risk values of the four adjacent raster cells above, below, left, and right are compared to determine the spatial direction in which the fire risk value increases the fastest, and a unique basic direction vector is generated for the raster cell.
[0077] The weighted smoothing process employs a fire hazard value-weighted Gaussian smoothing method, using the fire hazard value of the raster itself as the smoothing weight. The higher the fire hazard value, the greater the influence range of the raster direction vector. Global continuous smoothing is then applied to the direction vectors of adjacent raster units. After smoothing, a GIS raster attribute assignment method is used to lock a unique dominant fire hazard change direction for each raster unit within the control zone, generating a globally standardized principal gradient direction field.
[0078] The watershed algorithm adopts the existing MATLAB / GIS standard watershed segmentation algorithm, with a minimum seed point spacing of 50m and a region merging threshold of 0.1. Taking the seed point as the segmentation starting point, the segmentation boundary is automatically expanded according to the gradient of fire risk value, dividing the control zone into independent fire risk sub-regions that do not overlap, cover the entire area, and have no gaps. The segmentation results are automatically generated.
[0079] The centroid coordinates are calculated using the geometric centroid calculation method for GIS polygon features. For each sub-region, the planar coordinates of all rasters within the region are statistically analyzed, and a unique geometric center point is generated by averaging the coordinates across the entire region. The skewness coefficient is calculated using the statistical standard Pearson skewness method. The fire risk values of all rasters within the sub-region are statistically analyzed, and the degree of asymmetry in the numerical distribution is used as the skewness coefficient.
[0080] The K value is calculated using a ratio rounding method, which divides the spatial straight-line distance of a single video source by the average distance across the entire domain, and then rounds up. The larger the value, the wider the range of possible matches.
[0081] The shared viewpoint offset vector represents the spatial direction and amplitude of viewpoint adjustment; a larger vector magnitude results in a larger adjustment amplitude. These correspond to the horizontal adjustment angle, vertical adjustment angle, and minimum zoom magnification, respectively. The horizontal and vertical adjustment angles are converted into their corresponding offset directions and amplitudes, while the minimum zoom magnification is directly used as the zoom offset amplitude. The final viewpoint parameters are obtained by multiplying the amplitude corresponding to each shared viewpoint offset vector by the ratio of fire risk coverage contribution to monitoring requirement matching, adding the corresponding amplitude from the initial viewpoint, and then combining this with the direction.
[0082] The ratio of fire risk coverage contribution to monitoring demand matching degree represents the adaptability of the video source's own monitoring capabilities; the stronger the coverage capability, the higher the combustible material matching degree, the larger the zoom ratio, and the greater the viewing angle offset, the higher the priority is to ensure blind spot coverage in high-value areas; conversely, the smaller the offset, the less likely it is to deviate excessively from the monitoring area.
[0083] The fire risk value scalar field is a spatial set of global grid fire risk values, with the value indicating the intensity of local fire risk. The gradient vector and the principal gradient direction field represent the spatial directions where fire risk changes the fastest, and the more concentrated the directions, the stronger the directional tendency of fire spread. The local maximum point is the highest point of local fire risk and is the seed point for dividing fire risk sub-regions. The fire risk sub-regions are homogeneous fire risk spatial units, and their number determines the granularity of monitoring partitions. The comprehensive pointing deviation is the spatial pointing error between the equipment and the sub-region, and the smaller the value, the higher the matching degree.
[0084] In this embodiment, sub-regions are first divided based on the fire hazard gradient field and watershed algorithm to match the heterogeneous distribution characteristics of fire hazard space, achieving adaptive non-overlapping division of the monitoring area and avoiding the problem of fixed grid division being disconnected from fire hazard distribution. Then, candidate locking areas are screened by comprehensively considering pointing deviation, filtering out low-matching combinations. Next, the bipartite graph Hungarian algorithm is used to achieve one-to-one maximum weight matching, eliminating the resource waste of repeated viewpoint allocation. Finally, shared viewpoint parameters are generated using the highest fire hazard point as the guiding point, achieving blind-spot-free filling in areas without dedicated equipment. This addresses the practical problems of subjective viewpoint allocation, lack of optimal matching solutions, lack of filling in blank areas, and inability to directly schedule parameters in traditional methods.
[0085] Furthermore, for each candidate video source and each fire risk sub-region, the process of calculating the comprehensive pointing deviation is as follows: calculate the projection azimuth angle of the line connecting the spatial coordinates of the candidate video source and the centroid of the fire risk sub-region on the horizontal plane, as well as the vertical pitch angle of the line relative to the horizontal plane.
[0086] From the principal gradient direction field, extract the unit vector of the principal gradient direction of the grid cell containing the centroid of the fire risk sub-region and convert it into an orientation angle. Calculate the absolute value of the difference between the projected azimuth angle and the orientation angle, and denot it as the azimuth deviation.
[0087] Calculate the arithmetic mean of the terrain slope of all grid cells within the fire risk sub-region, and calculate the absolute value of the difference between the vertical pitch angle and the arithmetic mean, denoted as pitch deviation.
[0088] The skewness coefficient is used as a nonlinear amplification factor to nonlinearly amplify both the azimuth and pitch deviations: if the skewness coefficient is greater than 0, the azimuth deviation is multiplied by (1 + skewness coefficient), and the pitch deviation is multiplied by (1 + skewness coefficient). If the skewness coefficient is less than or equal to 0, both the azimuth and pitch deviations are multiplied by (1 / (1 + |skewness coefficient|)). The square root of the sum of the squares of the amplified azimuth and pitch deviations is used to obtain the overall pointing deviation.
[0089] It should be noted that the projected azimuth angle is the horizontal spatial direction of the line connecting the equipment and the centroid of the sub-region, and its value represents the horizontal pointing position of the equipment; the vertical pitch angle is the vertical tilt angle of the line relative to the horizontal plane, and its value represents the vertical observation angle of the equipment; the principal gradient direction unit vector is the standard direction of the fastest change in fire hazard, and the converted azimuth angle is the dominant direction of fire spread; the azimuth deviation is the angular error between the horizontal pointing of the equipment and the direction of fire hazard, and the smaller the value, the higher the horizontal pointing match; the average terrain slope of the sub-region is the overall horizontal tilt of the terrain in the sub-region, and the larger the value, the more significant the terrain undulation; the pitch deviation is the angular error between the vertical angle of the equipment and the terrain slope, and the smaller the value, the better the vertical observation adaptability; the skewness coefficient is the degree of asymmetry in the distribution of fire hazard, and the sign and magnitude of the value determine the extent to which the deviation is amplified or reduced; the comprehensive pointing deviation is the overall pointing error between the equipment and the sub-region, and the smaller the value, the higher the match between the two, and the more suitable it is for allocating monitoring perspectives.
[0090] S5. Based on the baseline fire risk value, the incremental fire spread risk, and wind speed and direction, simulate and generate a grid set of fire spread paths within a preset time period in the GIS platform. Combine this with the video scheduling scheme to verify coverage and redundancy, optimize viewpoint allocation, and determine the optimal scheduling scheme.
[0091] Considering that relying solely on fire hazard values cannot reflect the accelerating effect of wind speed on fire, and that random diffusion cannot reflect the true characteristics of fire propagation along high potential energy directions, global combustion simulation will generate a large number of isolated and invalid grids, leading to data redundancy. Therefore, it is necessary to determine the starting grid and driving potential energy, and determine the propagation direction based on the potential energy difference of the neighborhood to ensure that the simulation process conforms to the physical mechanism of fire spread. Therefore, the process of generating the fire spread path grid set within the future preset time period is as follows: S501, mark all grid units with open flames in the target monitoring area as fire source grid units. If there are no open flames, select the grid unit with the highest fire hazard risk value in the control zone as the virtual fire source grid unit.
[0092] S502. Multiply the fire hazard value of each grid cell by the wind speed to obtain the windward enhancement factor, and then multiply the fire hazard value by the windward enhancement factor to obtain the spread driving potential energy of the grid cell. The first multiplication reflects the basic amplification effect of wind on fire hazard; the second multiplication allows the spread potential energy of grid cells with higher fire hazard to be multiplied.
[0093] S503. Based on the difference between the direction angle of the line connecting each non-fire source grid unit and the fire source grid unit and the wind direction, retain only 2-4 adjacent grid directions with the smallest angle difference and the downwind spread, and then select the adjacent grid direction with the largest spread driving potential energy value as the fire propagation direction of each grid unit.
[0094] S504. Discretize the future preset duration into unit time steps. Within each unit time step, iteratively update the burning grid set according to the fire propagation direction and the ratio of the propagation driving potential energy of adjacent grid units until the cumulative time reaches the future preset duration.
[0095] S505. Extract the set of burning grids at the end of the iteration and remove the fire source grid cells. Then, trace back to the fire source grid cell in the opposite direction of the fire propagation direction for each grid cell. Retain the grid cells that can be successfully traced to form the fire hazard spread path grid set.
[0096] The iterative update process of the burning grid set is as follows: At the beginning of each unit time step, all currently burning grid cells are locked; then, for each burning grid cell, a unique adjacent grid cell to be burned is found along its preset fire propagation direction; next, the ratio of the propagation driving potential energy of the burning grid cell to that adjacent grid cell is calculated to determine whether the ignition condition is met; when the potential energy ratio reaches a preset ignition threshold, the adjacent grid cell is included in the burning grid set and marked as a new burning grid cell; if the threshold is not reached, it is not ignited and is not included; finally, after traversing all burning grid cells to complete the ignition determination, the burning grid set is updated, the current time step ends, and the next time step begins. Starting from the updated burning grid set, all the above steps are repeated until the preset total time is reached. For example, the ignition threshold is set to 1.2.
[0097] The windward enhancement factor is an amplification parameter of wind speed coupled with fire hazard. The larger the value, the stronger the boosting effect of wind speed on fire. The spread driving potential energy is the dynamic value of fire spread in grid cells. The larger the value, the easier the grid is to be ignited and the faster the spread speed. The connecting direction angle is the spatial azimuth angle between non-fire source grids and fire source grids, used to determine the relative direction of fire propagation. The fire propagation direction is the directional spread path of fire in space, which determines the order of grid ignition.
[0098] The preset duration is the total time range for fire risk prediction. The longer the duration, the wider the coverage of the spread simulation. The unit time step is the smallest time unit for iterative calculation. The smaller the step, the higher the simulation accuracy. The combustion grid set is the collection of grids ignited at each moment, representing the spatial range of fire spread. The fire spread path grid set is the effective spread grid that can be traced back to the fire source. It is a collection of fire propagation links.
[0099] This embodiment calculates the potential energy driving the spread by coupling fire risk value and wind speed, quantifying the physical boosting effect of wind speed on fire spread. The simulation results closely match the natural laws of fire spread in the field. At the same time, it combines azimuth difference and neighborhood potential energy to determine the propagation direction, abandoning the random spread model and ensuring that the fire spread direction has a clear spatial and dynamic basis. Finally, by removing fire source grids and filtering effective grids through reverse tracing, only the spread path connected to the fire source is retained. The grid set has no redundant data, representing the real fire spread link, which solves the problems of traditional spread simulation having no directional rules, incomplete working condition coverage, and no filtering of redundant paths.
[0100] Due to limitations imposed by factors such as fire emergency video dispatching, including the coverage of the spread path, limited equipment monitoring resources, overlapping and redundant viewpoints, and the timeliness of emergency response, inherent problems exist on-site, such as the initial viewpoint not being able to adapt to the dynamic spread path, the simultaneous existence of monitoring gaps and redundancy, and the lack of a unified response sequence for multiple device viewpoints. Considering the spatial connectivity of the fire spread grid, the spatial exclusivity of the equipment field of view, the step-by-step controllability of viewpoint adjustments, and the priority response characteristics of emergency monitoring, the optimal dispatching scheme can be pre-defined by quantifying gaps and redundancy through grid coverage counts, assigning redundancy responsibility to primary and secondary monitoring sources, implementing refined viewpoint correction through iterative steps, and unifying equipment response direction based on the fire arrival time. This pre-determines the dispatching requirements of no gaps, low redundancy, and high timeliness under dynamic fire conditions. The specific process for determining the optimal dispatching scheme is as follows: S506, calculate the monitoring field of view of each candidate video source according to the video dispatching scheme, count the number of times each grid unit in the fire spread path grid set is covered, and identify the set of coverage gaps (all grids with a coverage count of 0; the larger the range, the higher the risk of monitoring failure).
[0101] S507. Cluster grid cells with a coverage count of 2 or more into redundant regions, and label the primary monitoring source and secondary monitoring source for each redundant region.
[0102] S508. Calculate the viewing angle deviation between each coverage hole and the nearest candidate video source, and the percentage of overlapping coverage area of the secondary monitoring source relative to the primary monitoring source in each redundant region.
[0103] S509. Adjust the viewing angle of the nearest video source covering the hole and the secondary monitoring source in the redundant area in unit step size, iterating until the change rate of the total hole area and the total redundant area is less than a set value. For example, the set value is set to 5%, that is, the iteration stops when the change rate of the total hole area and the total redundant area is less than 5% for two consecutive iterations.
[0104] S510: Forcefully cover the remaining holes and deflect the center line of the viewpoint of all video sources toward the grid direction with the minimum fire arrival time to ensure consistent time response.
[0105] S511. Output the final adjusted horizontal rotation angle, vertical rotation angle, and zoom factor of each candidate video source as the optimal scheduling scheme.
[0106] Among them, the monitoring field of view is the effective spatial observation range of the candidate video source, which determines the grid boundary that the equipment can monitor. It adopts the GIS standard video field of view spatial modeling method, taking the installation coordinates of the video source as the origin, and combining its horizontal rotation angle, vertical rotation angle, and zoom magnification to automatically generate it.
[0107] Redundant areas are the spatial ranges covered by multiple overlapping devices. The higher the percentage of overlapping area, the more serious the resource waste. The main monitoring source is the main device responsible for the redundant area, and the secondary monitoring source is the secondary device with redundant overlap.
[0108] The viewing angle deviation is the difference in spatial angle between the device's viewing angle and the center of the cavity; the smaller the value, the smaller the adjustment range. The unit step size is the fixed angle value of a single viewing angle adjustment; the smaller the step size, the higher the adjustment accuracy. The rate of change is the iterative fluctuation range of the cavity and redundant area; the smaller the value, the more stable the scheduling.
[0109] The fire arrival time is the shortest time it takes for fire to spread to the grid. The smaller the value, the higher the priority of emergency monitoring. The calculation method is as follows: starting from the fire source grid, along the direction of fire propagation, one grid unit spreads within a unit time step; the fire arrival time is equal to the number of grid units in the spread path from the fire source grid multiplied by the unit time step, which is set to 1 minute.
[0110] The process of adjusting the viewing angle and iterating until the change rate of the total area of holes and the total area of redundancy is less than a set value is as follows: First, for each nearest video source covering a hole, the horizontal and vertical rotation angles of the video source are gradually adjusted towards the center point of the hole in fixed unit steps, one unit step at a time. Then, for each secondary monitoring source in the redundant area, the horizontal and vertical rotation angles of the video source are gradually adjusted away from the coverage area of the main monitoring source in fixed unit steps to reduce the overlap area. After completing one full-device viewing angle adjustment, the total area of holes and the total area of redundant areas covering the entire area are recalculated. Then, the change ratio of the total area of holes and the change ratio of the total area of redundancy between the current iteration and the previous iteration are calculated; when both change ratios are less than a preset fixed threshold, the iteration stops; if the threshold is not reached, the above adjustment steps are repeated to continue the next iteration.
[0111] This embodiment divides redundant areas by primary and secondary monitoring sources, clarifies the responsible equipment for perspective adjustment, and avoids secondary overlap caused by disordered adjustments by multiple devices; it uses unit step size to iteratively adjust the perspective to achieve parameter correction; it uses the area change rate as the iteration termination condition to automatically converge to the optimal scheduling state; it forcibly fills the remaining holes and uniformly deflects the perspective towards the earliest arriving fire hazard grid, eliminates monitoring blind spots, and ensures that all devices synchronously focus on the highest emergency priority area, solving the practical problems of poor static adaptability of initial scheduling, inability to quantify and correct hole redundancy, and inconsistent device response timing.
[0112] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.
[0113] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0114] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.
[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
[0116] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A video-linked emergency rescue management method based on GIS multimodal technology, characterized in that: Includes the following steps: For any grid cell, the baseline fire risk value is determined based on the terrain slope, type and proportion of combustibles, and the fire risk value is determined based on wind speed and direction, combined with the fire intensity and the fire spread risk increment. Based on the terrain slope and wind direction, fire risk gradient contour lines are constructed within the GIS platform, and the inner area is used as a control zone. Based on terrain slope and combustible material type, the visibility of dispatchable video equipment is analyzed to determine candidate video sources, and the fire risk coverage contribution, spatial straight distance and monitoring demand matching degree of the corresponding control zone are calculated. Based on the contribution of fire risk coverage, spatial straight-line distance and the matching degree of monitoring needs, and combined with the spatial distribution of fire risk values, the candidate video sources are allocated from different perspectives to determine the video scheduling scheme. Based on the baseline fire hazard value, the incremental fire spread hazard, and wind speed and direction, a fire spread path raster set is generated in the GIS platform. Coverage and redundancy are checked in conjunction with the video scheduling scheme, the viewpoint allocation is optimized, and the optimal scheduling scheme is determined.
2. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of determining the background fire hazard value is as follows: For any type of combustible material within any grid cell, the ratio of the square of the corresponding proportion to the corresponding total proportion of the entire area is used as the fire risk contribution item. The potential coefficient is obtained by summing all fire risk contributions within each grid cell. Determine the slope reference value corresponding to all grid cells, and calculate the magnification factor of the current grid cell based on the grid slope and the slope reference value of the grid cell; Multiplying the potential coefficient by the amplification coefficient yields the background fire hazard value of the corresponding grid cell.
3. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of determining the fire risk value is as follows: The relative wind speed spread coefficient of each grid cell is determined based on wind speed and the maximum wind speed during historical periods. If a fire source grid cell exists, the ratio of the corresponding fire intensity to the maximum fire intensity is used as the relative coefficient of fire intensity. The spatial azimuth angle from the fire source grid cell to the target grid cell is calculated using the GIS platform, and the wind direction spread matching coefficient is calculated in combination with the wind direction. Multiply the relative spread coefficient of wind speed, the wind direction spread matching coefficient, and the relative coefficient of fire intensity of the fire source grid unit to obtain the spread fire risk increment component. Summing up all spread fire risk increment components yields the spread fire risk increment. If there is no fire source grid cell, the fire spread risk increment of all grid cells is 0; The fire risk value is obtained by adding the background fire risk value corresponding to each grid cell to the fire spread risk increment.
4. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of constructing fire hazard gradient contour lines is as follows: The arithmetic mean of the azimuth angle of the terrain slope and the azimuth angle of the wind direction of the grid cell is used as the fire risk propagation direction. The median value of the fire risk value is used as the benchmark to extract the initial fire risk gradient contour line corresponding to the benchmark, and the fire risk gradient contour line is formed by correcting it based on the fire risk propagation direction.
5. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of determining candidate video sources is as follows: Select dispatchable video devices that are not obstructed by the line of sight within the monitoring range and contain the specified type of combustible material as the initial video devices; For multiple initial video devices with completely overlapping monitoring ranges, the one with the largest sum of coverage areas for each type of combustible material within the monitoring range is selected as the alternative video device. If all candidate video devices can fully cover the controlled area, then all candidate video devices will be used as candidate video sources; otherwise, the initially selected video devices will be used as candidate video sources.
6. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of calculating the fire risk coverage contribution, spatial straight-line distance, and monitoring demand matching degree is as follows: For any candidate video source, calculate the overlapping area between the maximum monitoring range and the control zone, and use the ratio of the sum of fire risk values in the overlapping area to the sum of all fire risk values in the control zone as the fire risk coverage contribution. The Euclidean distance between the geometric center point coordinates of the control zone and the spatial coordinates of the candidate video source is taken as the spatial straight-line distance. The percentage of the coverage area of combustible material types within the statistical control zone and the percentage of the maximum coverage area of combustible material types in overlapping areas are used as the ratio of the maximum coverage area percentage to the coverage area percentage as the monitoring demand matching degree.
7. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of determining the video scheduling scheme is as follows: The principal gradient direction field is obtained based on the fire risk value of each grid cell; Using the maximum value of fire risk as the seed point, the watershed algorithm is used to divide the fire risk into multiple fire risk sub-regions; Calculate the overall pointing deviation between each candidate video source and each fire risk sub-region; Candidate locking regions are determined based on spatial straight-line distance; A bipartite graph is constructed with candidate video sources as left nodes and fire risk sub-regions as right nodes. The Hungarian algorithm is used to solve the maximum weight matching of the bipartite graph, so that each candidate video source is responsible for a fire risk sub-region, and unmatched fire risk sub-regions are marked as shared regions. For each shared region, extract the guide point and generate shared viewpoint parameters that cover the guide point and do not exceed the boundary of the shared region for adjacent matched candidate video sources; Based on the initial viewpoint of each matched candidate video source, the final horizontal rotation angle, vertical rotation angle, and zoom magnification are determined by combining the shared viewpoint parameters, fire risk coverage contribution, and monitoring demand matching degree, thus forming a video scheduling scheme.
8. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 7, characterized in that, The process of calculating the overall pointing deviation is as follows: Calculate the projection azimuth angle of the line connecting the spatial coordinates of the candidate video source and the centroid of the fire risk sub-region on the horizontal plane, as well as the vertical pitch angle of the line relative to the horizontal plane. From the principal gradient direction field, extract the orientation angle corresponding to the grid cell where the centroid is located, and use the absolute value of the difference between the projected azimuth angle and the orientation angle as the azimuth deviation. Calculate the arithmetic mean of the terrain slope of all grid cells within the fire risk sub-region, and use the absolute value of the difference between the vertical pitch angle and the arithmetic mean as the pitch deviation. Nonlinear amplification is applied to both the positional and pitch deviations; The square root of the sum of the squares of the magnified azimuth and pitch deviations is used to obtain the overall pointing deviation.
9. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of generating a fire spread path grid set is as follows: The fire risk value of each grid cell is multiplied by the wind speed to obtain the windward enhancement factor, and then the fire risk value is multiplied by the windward enhancement factor to obtain the propagation driving potential energy. The fire propagation direction is assigned to each grid cell based on the propagation driving potential energy of the surrounding neighboring grid cells; Based on the fire propagation direction and the ratio of the propagation driving potential energy of adjacent grid cells, the burning grid set is iteratively updated. At the end of the iteration, the burning grid set is filtered and traced to obtain the fire spread path grid set.
10. The video-linked emergency rescue management method based on GIS multimodal technology according to claim 1, characterized in that, The process of determining the optimal scheduling scheme is as follows: Count the number of times each grid cell in the fire spread path grid is covered, and identify the set of coverage holes; Cluster grid cells with a coverage count of 2 or more into redundant regions, and label the primary and secondary monitoring sources of each redundant region. Calculate the viewing angle deviation between each coverage hole and the nearest candidate video source, and the percentage of overlapping coverage area of the secondary monitoring source relative to the primary monitoring source in each redundant region; The viewing angle of the nearest video source and the secondary monitoring source in the redundant area that cover the hole is adjusted. The iteration continues until the change rate of the total area of the hole and the total area of the redundancy is less than the set value. The remaining hole is then forcibly covered, and the center line of the viewing angle of all video sources is deflected towards the grid direction with the minimum fire arrival time. The final adjusted horizontal rotation angle, vertical rotation angle, and zoom factor of each candidate video source are output as the optimal scheduling scheme.