A fire dynamic monitoring method and system suitable for a UAV
Patent Information
- Application Number
- CN202611529898.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-30
- Publication Date
- 2026-11-03
AI Technical Summary
Existing technologies cannot accurately identify fire areas in multi-fire scenarios, leading to adjacent fire points being misidentified as a single fire point or weak fire points being missed, thus affecting the accuracy of fire monitoring.
By processing infrared images into a grid, analyzing the thermal radiation index and elevation differences of each image grid, dynamically adjusting the temperature threshold, and performing fire detection, we can ensure that weak fire points are not missed and accurately identify fire areas.
It improves the accuracy of fire point identification, enhances the reliability of fire analysis, enables timely detection and location of fire sources, and improves the effectiveness of fire monitoring.
Smart Images
Figure CN123033150A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing technology, and specifically to a method and system for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs). Background Technology
[0002] In recent years, drones have been widely used in the field of fire monitoring. Specifically, drones can quickly reach complex areas such as mountainous areas and dense forests that are difficult for people and vehicles to access, greatly shortening the time for fire detection. In addition, drones equipped with visible light and infrared thermal imaging equipment can penetrate smoke and accurately identify high-temperature points and smoldering fires at night and in bad weather, so as to detect and locate the fire source in a timely manner.
[0003] In application, it was found that in scenarios with multiple fire points, the uniform threshold used by existing technology cannot adapt to the actual distribution of fire points, resulting in inaccurate segmentation of fire point areas. For example, in areas with dense fire point distribution, the temperature superposition between adjacent fire points may cause multiple fire points that were originally adjacent to each other to be identified as one fire point. In addition, in areas with sparse fire point distribution, weak fire points may be missed. Summary of the Invention
[0004] The purpose of this invention is to provide a method and system for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs), which solves the technical problem of poor identification effect of fire point areas in the prior art.
[0005] In a first aspect, one embodiment of the present invention provides a method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs), the method comprising: Based on a preset temperature threshold, fire points are detected in the infrared image of the target fire site to obtain multiple initial fire point regions. The temperature threshold is the minimum temperature value in the fire point detection threshold range, and the temperature value of any image point in the initial fire point region is greater than or equal to the temperature threshold. In the multiple image grids corresponding to the infrared image, the temperature values of the image points corresponding to the initial fire point area in each image grid are accumulated to obtain the thermal radiation index of each image grid. The thermal radiation index is used to indicate the intensity of heat radiation emitted by the corresponding image grid to its neighboring image grids. In the plurality of image grids, the thermal radiation indices of the plurality of neighboring image grids corresponding to each image grid are fused to obtain the neighborhood radiation value of each image grid, wherein the neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation of its neighboring image grids; The temperature threshold is increased based on the neighborhood radiance value of each image grid to obtain the grid threshold corresponding to each image grid, wherein the increase in the temperature threshold is positively correlated with the neighborhood radiance value; Based on the grid threshold corresponding to each image grid, fire point detection is performed on the infrared image of the target fire site to obtain multiple target fire point areas.
[0006] In one embodiment, the step of obtaining multiple image grids corresponding to the infrared image includes: The lower quartile of the area of multiple initial fire point regions is determined as the unit grid area; The infrared image is meshed based on the unit grid area to obtain multiple image grids corresponding to the infrared image.
[0007] In one embodiment, the step of fusing the thermal radiation indices of multiple neighboring image grids corresponding to each image grid to obtain the neighborhood radiation value of each image grid includes: In the plurality of image grids, the elevation difference between each image grid and its corresponding plurality of neighboring image grids is analyzed to determine the radiation weights of the plurality of neighboring image grids corresponding to each image grid; In the plurality of image grids, the thermal radiation index of the plurality of neighboring image grids corresponding to each image grid is weighted and summed based on the radiation weight of the plurality of neighboring image grids corresponding to each image grid to obtain the neighborhood radiation index of each image grid. The neighborhood radiance indices of multiple image grids are normalized to obtain the neighborhood radiance value of each image grid, wherein the neighborhood radiance value and the corresponding neighborhood radiance index are positively correlated.
[0008] In one embodiment, the step of analyzing the elevation differences between each image grid and its corresponding plurality of neighboring image grids to determine the radiative weights of the plurality of neighboring image grids corresponding to each image grid includes: In the plurality of image grids, the absolute difference between the elevation data of each image grid and the elevation data of each of its corresponding neighboring image grids is calculated to obtain multiple elevation differences corresponding to each image grid; Calculate the sum of the reciprocals of multiple elevation differences corresponding to each image grid to obtain the elevation reference index corresponding to each image grid; Calculate the ratio of the reciprocal of the multiple elevation differences corresponding to each image grid to its corresponding elevation reference index to obtain the multiple radiation weights corresponding to each image grid.
[0009] In one embodiment, after detecting fire points in the infrared image of the target fire scene based on the grid threshold corresponding to each image grid to obtain multiple target fire point regions, the method further includes: The multiple target fire point regions are tracked based on multiple infrared images to determine multiple displacement rates and multiple area change rates for each target fire point region. Multiple displacement rates of each target fire point region are fused to obtain displacement feature values of each target fire point region, and multiple area change rates of each target fire point region are fused to obtain area feature values of each target fire point region. The displacement feature values are used to indicate the positional movement intensity of the corresponding target fire point region, and the area feature values are used to indicate the area growth intensity of the corresponding target fire point region. Based on the displacement characteristic value and area characteristic value of each target fire point area, a fire point risk assessment is performed on each target fire point area to obtain a risk characteristic value for each target fire point area. The risk characteristic value is used to indicate the degree of danger of the corresponding target fire point area. Fire situation analysis is performed based on the risk characteristic values of multiple target fire areas to obtain fire situation analysis results, wherein the fire situation analysis results are used to indicate the severity of the fire at the target fire site.
[0010] In one embodiment, the displacement feature value is calculated by weighting multiple corresponding displacement rates, and the calculation weight of the displacement rate is negatively correlated with its corresponding time-domain monitoring distance. The area feature value is calculated by weighting multiple corresponding area change rates, and the calculation weight of the area change rate is negatively correlated with its corresponding time-domain monitoring distance. The time-domain monitoring distance is the time difference between the image acquisition time of the corresponding infrared image and the current time.
[0011] In one embodiment, the step of assessing the fire risk of each target fire point region based on its displacement characteristic value and area characteristic value, and obtaining the risk characteristic value of each target fire point region, includes: The displacement characteristic value and area characteristic value of each target fire point region are normalized to obtain the displacement normalization value and area normalization value of each target fire point region. The displacement normalization value and area normalization value of each target fire point area are weighted and calculated to obtain the risk characteristic value of each target fire point area, wherein the calculation weight of the displacement normalization value is greater than the calculation weight of the area normalization value.
[0012] In one embodiment, the step of performing fire analysis based on risk characteristic values of multiple target fire point areas to obtain fire analysis results includes: Among the multiple target fire zones, the target fire zone with a risk characteristic value greater than or equal to a risk threshold is determined as the first fire zone, and the target fire zone with an area greater than or equal to an area threshold is determined as the second fire zone. If the proportion of the first fire point areas is greater than the proportion threshold, and / or the total area of the second fire point areas is greater than the area accumulation threshold, output a fire analysis result indicating a serious fire risk. If the proportion of the first fire point areas is less than or equal to the proportion threshold, and the total area of the second fire point areas is less than or equal to the cumulative area threshold, output a fire analysis result indicating general fire risk.
[0013] In one embodiment, after performing fire analysis based on risk characteristic values of multiple target fire point areas to obtain fire analysis results, the method further includes: The movement trajectories of multiple target fire point areas are acquired and visualized in the 3D model corresponding to the target fire site. The area of the target fire point area is positively correlated with the area of the visualized area of the corresponding movement trajectory.
[0014] Secondly, another embodiment of the present invention also provides a fire dynamic monitoring system suitable for unmanned aerial vehicles (UAVs), the system comprising: The initial detection module is used to detect fire points in the infrared image of the target fire site based on a preset temperature threshold, and obtain multiple initial fire point regions. The temperature threshold is the minimum temperature value in the fire point detection threshold range, and the temperature value of any image point in the initial fire point region is greater than or equal to the temperature threshold. The first processing module is used to accumulate the temperature values of image points corresponding to the initial fire point region in each of the multiple image grids corresponding to the infrared image to obtain the thermal radiation index of each image grid, wherein the thermal radiation index is used to indicate the intensity of heat radiation emitted by the corresponding image grid to its neighboring image grids; The second processing module is used to fuse the thermal radiation indices of multiple neighboring image grids corresponding to each image grid in the plurality of image grids to obtain the neighborhood radiation value of each image grid, wherein the neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation of its neighboring image grids. A threshold adjustment module is used to increase the temperature threshold based on the neighborhood radiation value of each image grid to obtain a grid threshold corresponding to each image grid, wherein the increase in the temperature threshold is positively correlated with the neighborhood radiation value; The re-inspection module is used to detect fire points in the infrared image of the target fire site based on the grid threshold corresponding to each image grid, and obtain multiple target fire point areas.
[0015] The present invention has the following beneficial effects: First, fire point detection is performed on the infrared images of the target fire site to initially determine multiple initial fire point areas. A temperature threshold is set as the minimum temperature value within the fire point detection threshold range to minimize the risk of missing weak fire points. Then, the temperature values of the image points corresponding to the initial fire point areas in each image grid are accumulated to determine the intensity of heat radiation emitted by each image grid. Next, the thermal radiation intensity of neighboring image grids is fused to determine the degree of temperature drift caused by the influence of neighboring grids on each image grid. Based on this, the threshold of each image grid is dynamically corrected. Finally, fire point detection is performed based on the corrected grid thresholds of each image grid to obtain more accurate and reliable multiple target fire point areas, thereby improving the reliability of subsequent fire analysis operations based on multiple target fire point areas. Attached Figure Description
[0016] Figure 1 This is a flowchart illustrating a method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a fire dynamic monitoring system suitable for unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention. Detailed Implementation
[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0018] The following description, in conjunction with the accompanying drawings, details a specific scheme for a fire dynamic monitoring method and system applicable to unmanned aerial vehicles (UAVs) provided by the present invention.
[0019] In one embodiment, the present invention provides a method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs), such as... Figure 1 As shown, the method includes: Step S1: Based on a preset temperature threshold, perform fire point detection on the infrared image of the target fire site to obtain multiple initial fire point areas.
[0020] Wherein, the temperature threshold is the minimum temperature value in the fire detection threshold range, and the temperature value of any image point in the initial fire region is greater than or equal to the temperature threshold.
[0021] In this invention, the target fire area should be understood as a specific spatial range in which the fire is detected, such as a hilltop or a forest.
[0022] By regularly patrolling the aforementioned specific spatial range with drones, capturing infrared thermal images, and performing image analysis, the corresponding fire situation can be monitored. Specifically, if the proportion of risk pixels identified in a captured infrared thermal image is greater than a threshold (e.g., 0.015), then it is determined that a fire has been detected in the aforementioned specific spatial range. Risk pixels are specifically pixels in the infrared thermal image whose temperature value is greater than or equal to the temperature threshold.
[0023] The aforementioned percentage threshold is set to avoid false fire warnings caused by environmental noise.
[0024] Infrared images of the target fire site can be understood as images acquired by an infrared thermal imaging module carried by a drone.
[0025] The fire detection threshold range should be understood as the range of temperature detection threshold values for pixels used to identify fire-indicating areas in an infrared image monitoring scenario.
[0026] The minimum temperature value in the fire detection threshold range can be understood as the most sensitive temperature detection threshold that can be set in an infrared image monitoring scenario.
[0027] Analysis revealed that although each pixel in an infrared image directly reflects the temperature value of its corresponding location through image grayscale, for areas with multiple fire points clustered together, the separation gap between adjacent fire points is affected by the thermal radiation of surrounding fire points, causing the temperature value of the corresponding pixel to inevitably drift upwards. This can lead to multiple originally adjacent fire point areas being incorrectly classified as a single fire point area. In addition, for isolated fire points, there may be problems such as missed detection or severe distortion of the area of the detected fire point area due to improper threshold settings.
[0028] Based on this, the present invention performs gridding processing on infrared images and analyzes the influence of thermal radiation from fire points in each image grid. It then configures an adaptive threshold for each image grid and performs fire point detection accordingly. This suppresses the situation where adjacent fire points are divided into a whole fire point. At the same time, by using the minimum temperature value of the fire point detection threshold range as the initial temperature threshold, the risk of weak fire points being missed can be effectively suppressed. Through the above measures, more accurate and reliable multiple target fire point areas can be obtained, thereby improving the reliability of subsequent fire analysis operations based on multiple target fire point areas.
[0029] Specifically, based on a preset temperature threshold, the process of detecting fire points in the infrared image of the target fire site to obtain multiple initial fire point regions is as follows: Among all the pixels in the infrared image, pixels with a temperature value greater than or equal to the temperature threshold are identified as fire zone pixels to obtain multiple fire zone pixels. Connectivity analysis is performed on the multiple fire domain pixels to obtain the aforementioned multiple initial fire point regions.
[0030] Step S2: In the multiple image grids corresponding to the infrared image, the temperature values of the image points corresponding to the initial fire point area in each image grid are accumulated to obtain the thermal radiation index of each image grid.
[0031] The thermal radiation index is used to indicate the intensity of heat radiation emitted by the corresponding image grid to its neighboring image grids.
[0032] In one example, the grid size of the image grid can be manually set based on experience.
[0033] In another example, the grid size of the image grid can be dynamically determined based on the area of multiple initial fire point regions to achieve more accurate grid segmentation processing of infrared images.
[0034] Specifically, the step of obtaining multiple image grids corresponding to the infrared image includes: The lower quartile of the area of multiple initial fire point regions is determined as the unit grid area; The infrared image is meshed based on the unit grid area to obtain multiple image grids corresponding to the infrared image.
[0035] In the above settings, the lower quartile of the area of multiple initial fire point regions is used as the unit grid area. This can adapt to the area of multiple initial fire point regions while suppressing the influence of extreme values, ensuring the effective implementation of subsequent fire point detection based on each image grid.
[0036] It should be understood that the area of any one of the multiple image grids is less than or equal to the area of the unit grid.
[0037] In the above settings, only the temperature values of the image points corresponding to the initial fire point area in the image grid are used for accumulation in order to avoid numerical interference from image points in non-initial fire point areas, so as to ensure that the calculated thermal radiation index can accurately reflect the intensity of heat dissipation radiation emitted outward from the corresponding image grid.
[0038] Specifically, the higher the thermal radiation index, the stronger the heat radiation emitted by the corresponding image grid to its neighboring image grids, which means that the temperature drift of the neighboring image grids caused by its thermal radiation is more significant.
[0039] It should be added that when the number of multiple initial fire point areas is less than the set threshold (e.g., 6), the preset default area value (one percent of the image area of the infrared image) will be determined as the unit grid area.
[0040] Step S3: In the plurality of image grids, the thermal radiation index of the plurality of neighboring image grids corresponding to each image grid is fused to obtain the neighborhood radiation value of each image grid.
[0041] The neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation from its neighboring image grids.
[0042] In this invention, the number of neighboring image grids corresponding to each image grid is set to 8. The neighboring image grids should be understood as several image grids that are adjacent to the corresponding image grid in the multiple image grids.
[0043] Specifically, the step of fusing the thermal radiation indices of multiple neighboring image grids corresponding to each image grid to obtain the neighborhood radiation value of each image grid includes: In the plurality of image grids, the elevation difference between each image grid and its corresponding plurality of neighboring image grids is analyzed to determine the radiation weights of the plurality of neighboring image grids corresponding to each image grid; In the plurality of image grids, the thermal radiation index of the plurality of neighboring image grids corresponding to each image grid is weighted and summed based on the radiation weight of the plurality of neighboring image grids corresponding to each image grid to obtain the neighborhood radiation index of each image grid. The neighborhood radiance indices of multiple image grids are normalized to obtain the neighborhood radiance value of each image grid, wherein the neighborhood radiance value and the corresponding neighborhood radiance index are positively correlated.
[0044] Since the region corresponding to an infrared image usually has a certain height undulation, and the degree of height undulation will affect the thermal radiation effect of adjacent blocks indicated by adjacent image grids to a certain extent, if the thermal radiation index of multiple neighboring image grids corresponding to each image grid is not processed indiscriminately, the calculated neighboring radiation value will be inaccurate.
[0045] Based on this, the present invention analyzes the elevation difference between an image grid and its corresponding neighboring image grids to adaptively determine the radiation weights of multiple neighboring image grids corresponding to each image grid, thereby reducing the thermal radiation impact of neighboring image grids with significant elevation differences and improving the accuracy of the calculated neighboring radiation values.
[0046] In practical applications, the intrinsic parameters (focal length, principal point coordinates, distortion coefficients) and extrinsic parameters (relative to the installation angle and position of the UAV) of the infrared thermal imaging module can be used, combined with the UAV's own pose data (longitude, latitude, altitude, yaw angle, pitch angle, roll angle), to transform the coordinates of each pixel detected in the infrared image point by point into the digital elevation model corresponding to the target fire site, thereby obtaining the elevation value of each pixel in the infrared image. Then, the average elevation value of all pixels in each image grid is determined as the corresponding elevation data, and subsequently, the aforementioned elevation difference analysis is carried out based on the elevation data of each image grid.
[0047] It should be understood that the greater the elevation difference, the smaller the corresponding radiation weight.
[0048] Specifically, the step of analyzing the elevation difference between each image grid and its corresponding multiple neighboring image grids to determine the radiation weights of the multiple neighboring image grids corresponding to each image grid includes: In the plurality of image grids, the absolute difference between the elevation data of each image grid and the elevation data of each of its corresponding neighboring image grids is calculated to obtain multiple elevation differences corresponding to each image grid; Calculate the sum of the reciprocals of multiple elevation differences corresponding to each image grid to obtain the elevation reference index corresponding to each image grid; Calculate the ratio of the reciprocal of the multiple elevation differences corresponding to each image grid to its corresponding elevation reference index to obtain the multiple radiation weights corresponding to each image grid.
[0049] Each image grid corresponds to multiple elevation differences and multiple neighboring image grids.
[0050] It should be noted that when one or more elevation differences are 0, the subsequent calculation process is skipped, and the average value of the thermal radiation index of one or more neighboring image grids corresponding to zero elevation difference is directly calculated and determined as the corresponding neighborhood radiation index.
[0051] The numerical range of the aforementioned neighborhood radiation values is 0-1. In this invention, the maximum-minimum normalization algorithm is specifically used to normalize the neighborhood radiation indices of multiple image grids.
[0052] Step S4: Increase the temperature threshold based on the neighborhood radiation value of each image grid to obtain the grid threshold corresponding to each image grid.
[0053] The increase in the temperature threshold is positively correlated with the neighborhood radiation value.
[0054] Specifically, the steps for determining the grid threshold corresponding to each image grid are as follows: The threshold adjustment amplitude of each image grid is obtained by multiplying the neighborhood radiation value and the threshold fluctuation value of each image grid, wherein the threshold fluctuation value is the difference between the maximum and minimum values of the fire detection threshold interval. The sum of the threshold adjustment magnitude and the temperature threshold for each image grid is calculated to obtain the grid threshold corresponding to each image grid.
[0055] Step S5: Based on the grid threshold corresponding to each image grid, perform fire point detection on the infrared image of the target fire site to obtain multiple target fire point areas.
[0056] Specifically, among all the pixels in each image grid, the pixels in each image grid whose temperature value is greater than or equal to the corresponding grid threshold are identified as target pixels, so as to obtain multiple target pixels for each image grid; Connectivity analysis is performed on all target pixels to obtain the aforementioned multiple target fire point regions.
[0057] In summary, this invention first performs fire point detection based on infrared images of the target fire site to initially determine multiple initial fire point regions. Setting a temperature threshold as the minimum temperature value within the fire point detection threshold range is intended to minimize the risk of missing weak fire points. Then, the temperature values of image points corresponding to the initial fire point regions in each image grid are accumulated to determine the intensity of heat radiation emitted outward from each image grid. Next, the thermal radiation intensities of neighboring image grids are fused to determine the degree of temperature drift caused by the influence of neighboring grids on each image grid, and the thresholds of each image grid are dynamically corrected accordingly. Finally, fire point detection is performed based on the corrected grid thresholds of each image grid to obtain more accurate and reliable multiple target fire point regions, thereby improving the reliability of subsequent fire analysis operations based on multiple target fire point regions.
[0058] In some implementations, after detecting fire points in the infrared image of the target fire scene based on the grid threshold corresponding to each image grid to obtain multiple target fire point regions, the method further includes: The multiple target fire point regions are tracked based on multiple infrared images to determine multiple displacement rates and multiple area change rates for each target fire point region. Multiple displacement rates of each target fire point region are fused to obtain displacement feature values of each target fire point region, and multiple area change rates of each target fire point region are fused to obtain area feature values of each target fire point region. The displacement feature values are used to indicate the positional movement intensity of the corresponding target fire point region, and the area feature values are used to indicate the area growth intensity of the corresponding target fire point region. Based on the displacement characteristic value and area characteristic value of each target fire point area, a fire point risk assessment is performed on each target fire point area to obtain a risk characteristic value for each target fire point area. The risk characteristic value is used to indicate the degree of danger of the corresponding target fire point area. Fire situation analysis is performed based on the risk characteristic values of multiple target fire areas to obtain fire situation analysis results, wherein the fire situation analysis results are used to indicate the severity of the fire at the target fire site.
[0059] In the above setup, multiple infrared images are periodically acquired, and the position and area changes of multiple target fire points are tracked to determine the degree of danger of each target fire point area. Based on this, the fire status of the target fire site is analyzed to assess the severity of the fire, thereby achieving real-time tracking and hazard warning of the target fire site.
[0060] In this invention, for a target fire site, a drone is used to periodically (e.g., every 30 seconds) acquire corresponding infrared images. The above-mentioned process is used to process each acquired infrared image to obtain multiple target fire point regions corresponding to each infrared image. Then, multi-target tracking (e.g., IOU matching, Hungarian algorithm, etc.) is performed based on these to obtain the image region corresponding to each target fire point region in multiple infrared images. This allows for the acquisition of multiple displacement rates and multiple area change rates for each target fire point region. The displacement rate is the ratio of the spatial distance between the fire point centers of the corresponding target fire point region in two adjacent infrared images to the acquisition period (i.e., 30 seconds). The area change rate is the ratio of the area difference between the regions of the corresponding target fire point region in two adjacent infrared images to the acquisition period.
[0061] The fire center can be understood as the location of the pixel with the highest temperature value in the corresponding infrared image region of the target fire area. In application, if there are multiple pixels with the highest temperature value, the average position of the multiple pixels with the highest temperature is determined as the fire center.
[0062] To amplify the impact of recently acquired infrared image data and ensure that the calculated risk characteristic value accurately reflects the danger level of the target fire point area, the displacement characteristic value can be calculated by weighting multiple corresponding displacement rates. The calculation weight of the displacement rate is negatively correlated with its corresponding time-domain monitoring distance. The area characteristic value is calculated by weighting multiple corresponding area change rates. The calculation weight of the area change rate is negatively correlated with its corresponding time-domain monitoring distance, where the time-domain monitoring distance is the time difference between the image acquisition time of the corresponding infrared image and the current time.
[0063] For example, the sum of the reciprocals of multiple time-domain monitoring distances can be calculated to obtain a distance reference sum. Then, the ratio of the reciprocal of each time-domain monitoring distance to the distance reference sum can be calculated to obtain the time-domain value index corresponding to each time-domain monitoring distance, which is to obtain the calculation weight corresponding to each displacement rate and each area change rate.
[0064] The steps for conducting a fire risk assessment for each target fire point region based on its displacement and area characteristic values, and obtaining the risk characteristic value for each target fire point region, include: The displacement characteristic value and area characteristic value of each target fire point region are normalized to obtain the displacement normalization value and area normalization value of each target fire point region. The displacement normalization value and area normalization value of each target fire point area are weighted and calculated to obtain the risk characteristic value of each target fire point area, wherein the calculation weight of the displacement normalization value is greater than the calculation weight of the area normalization value.
[0065] In this invention, the displacement normalization value is the ratio of the corresponding displacement feature value to the preset displacement feature threshold, and the area normalization value is the ratio of the corresponding area feature value to the preset area feature threshold. The displacement feature threshold is the largest historical displacement feature value that has been monitored in the past, and similarly, the area feature threshold is the largest historical area feature value that has been monitored in the past.
[0066] In addition, to ensure the smooth implementation of the scheme under extreme working conditions, when the ratio of the corresponding displacement characteristic value to the preset displacement characteristic threshold is greater than 1, the corresponding displacement normalization value is forcibly set to 1, and when the ratio of the corresponding area characteristic value to the preset area characteristic threshold is greater than 1, the corresponding area normalization value is forcibly set to 1.
[0067] In the above settings, the calculation weight of the displacement normalization value is set to be greater than that of the area normalization value in order to adapt to the situation that fire points with high displacement are more dangerous than fire points with large areas (fire point development is more uncontrollable), which can make the determined risk characteristic value more accurate and reliable.
[0068] For example, the present invention sets the calculation weights of displacement normalization value and area normalization value to 0.6 and 0.4, respectively.
[0069] In this invention, the step of performing fire situation analysis based on the risk characteristic values of multiple target fire point areas to obtain fire situation analysis results includes: Among the multiple target fire zones, the target fire zone with a risk characteristic value greater than or equal to a risk threshold is determined as the first fire zone, and the target fire zone with an area greater than or equal to an area threshold is determined as the second fire zone. If the proportion of the first fire point areas is greater than the proportion threshold, and / or the total area of the second fire point areas is greater than the area accumulation threshold, output a fire analysis result indicating a serious fire risk. If the proportion of the first fire point areas is less than or equal to the proportion threshold, and the total area of the second fire point areas is less than or equal to the cumulative area threshold, output a fire analysis result indicating general fire risk.
[0070] In the above settings, the first fire point area is used to indicate high-risk fire points that are developing, and the second fire point area is used to indicate high-risk fire points that have developed to a certain extent. Through a multi-condition comprehensive judgment method, in the case of multiple rapidly developing high-risk fire points and / or high-risk fire points with a large burning area, the fire situation analysis results indicating serious fire risk are output in a timely manner to ensure the timeliness and accuracy of fire warning information output.
[0071] In application, fire analysis results indicating severe fire risk can be understood as a red fire alert, while fire analysis results indicating general fire risk can be understood as a yellow fire alert.
[0072] In some implementations, after performing fire analysis based on risk characteristic values of multiple target fire point areas to obtain the fire analysis results, the method further includes: The movement trajectories of multiple target fire point areas are acquired and visualized in the 3D model corresponding to the target fire site. The area of the target fire point area is positively correlated with the area of the visualized area of the corresponding movement trajectory.
[0073] The movement trajectory of the target fire point area is obtained by sequentially connecting multiple fire point centers in the order of acquisition time. The visualization area corresponding to the movement trajectory can be understood as a polyline segment in the three-dimensional model. The width of the polyline segment is determined by the average of the area of multiple regions of the target fire point area (corresponding one-to-one with multiple infrared images). The larger the average value, the larger the width of the corresponding polyline segment.
[0074] By visually displaying the movement trajectory of the target fire area and dynamically adapting the area of the visualized movement trajectory to the area, the system accurately indicates the real-time development status of each fire point in the target fire scene to relevant rescue personnel, thereby helping them to make better rescue decisions.
[0075] In another embodiment, the present invention also provides a fire dynamic monitoring system suitable for unmanned aerial vehicles (UAVs), such as... Figure 2 As shown, a fire dynamic monitoring system 200 suitable for unmanned aerial vehicles (UAVs) includes: The initial detection module 201 is used to perform fire point detection on the infrared image of the target fire site based on a preset temperature threshold to obtain multiple initial fire point regions. The temperature threshold is the minimum temperature value of the fire point detection threshold range, and the temperature value of any image point in the initial fire point region is greater than or equal to the temperature threshold. The first processing module 202 is used to accumulate the temperature values of the image points corresponding to the initial fire point area in each of the multiple image grids corresponding to the infrared image to obtain the thermal radiation index of each image grid, wherein the thermal radiation index is used to indicate the intensity of the heat radiation emitted by the corresponding image grid to its neighboring image grids. The second processing module 203 is used to fuse the thermal radiation indices of multiple neighboring image grids corresponding to each image grid in the plurality of image grids to obtain the neighborhood radiation value of each image grid, wherein the neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation of its neighboring image grids. The threshold adjustment module 204 is used to increase the temperature threshold based on the neighborhood radiation value of each image grid to obtain the grid threshold corresponding to each image grid, wherein the increase of the temperature threshold is positively correlated with the neighborhood radiation value; The re-inspection module 205 is used to detect fire points in the infrared image of the target fire site based on the grid threshold corresponding to each image grid, and obtain multiple target fire point areas.
[0076] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the fire dynamic monitoring system and the fire dynamic monitoring method for UAVs provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.
[0077] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0078] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs), characterized in that, The method includes: Based on a preset temperature threshold, fire points are detected in the infrared image of the target fire site to obtain multiple initial fire point regions. The temperature threshold is the minimum temperature value in the fire point detection threshold range, and the temperature value of any image point in the initial fire point region is greater than or equal to the temperature threshold. In the multiple image grids corresponding to the infrared image, the temperature values of the image points corresponding to the initial fire point area in each image grid are accumulated to obtain the thermal radiation index of each image grid. The thermal radiation index is used to indicate the intensity of heat radiation emitted by the corresponding image grid to its neighboring image grids. In the plurality of image grids, the thermal radiation indices of the plurality of neighboring image grids corresponding to each image grid are fused to obtain the neighborhood radiation value of each image grid, wherein the neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation of its neighboring image grids; The temperature threshold is increased based on the neighborhood radiance value of each image grid to obtain the grid threshold corresponding to each image grid, wherein the increase in the temperature threshold is positively correlated with the neighborhood radiance value; Based on the grid threshold corresponding to each image grid, fire point detection is performed on the infrared image of the target fire site to obtain multiple target fire point areas.
2. The fire dynamic monitoring method applicable to UAVs according to claim 1, characterized in that, The step of obtaining multiple image grids corresponding to the infrared image includes: The lower quartile of the area of multiple initial fire point regions is determined as the unit grid area; The infrared image is meshed based on the unit grid area to obtain multiple image grids corresponding to the infrared image.
3. The fire dynamic monitoring method applicable to UAVs according to claim 1, characterized in that, The step of fusing the thermal radiation indices of multiple neighboring image grids corresponding to each image grid to obtain the neighborhood radiation value of each image grid includes: In the plurality of image grids, the elevation difference between each image grid and its corresponding plurality of neighboring image grids is analyzed to determine the radiation weights of the plurality of neighboring image grids corresponding to each image grid; In the plurality of image grids, the thermal radiation index of the plurality of neighboring image grids corresponding to each image grid is weighted and summed based on the radiation weight of the plurality of neighboring image grids corresponding to each image grid to obtain the neighborhood radiation index of each image grid. The neighborhood radiance indices of multiple image grids are normalized to obtain the neighborhood radiance value of each image grid, wherein the neighborhood radiance value and the corresponding neighborhood radiance index are positively correlated.
4. The fire dynamic monitoring method applicable to UAVs according to claim 3, characterized in that, The step of analyzing the elevation differences between each image grid and its corresponding multiple neighboring image grids in the plurality of image grids to determine the radiation weights of the multiple neighboring image grids corresponding to each image grid includes: In the plurality of image grids, the absolute difference between the elevation data of each image grid and the elevation data of each of its corresponding neighboring image grids is calculated to obtain multiple elevation differences corresponding to each image grid; Calculate the sum of the reciprocals of multiple elevation differences corresponding to each image grid to obtain the elevation reference index corresponding to each image grid; Calculate the ratio of the reciprocal of the multiple elevation differences corresponding to each image grid to its corresponding elevation reference index to obtain the multiple radiation weights corresponding to each image grid.
5. The fire dynamic monitoring method applicable to UAVs according to claim 1, characterized in that, Based on the grid threshold corresponding to each image grid, fire point detection is performed on the infrared image of the target fire scene to obtain multiple target fire point regions. The method further includes: The multiple target fire point regions are tracked based on multiple infrared images to determine multiple displacement rates and multiple area change rates for each target fire point region. Multiple displacement rates of each target fire point region are fused to obtain displacement feature values of each target fire point region, and multiple area change rates of each target fire point region are fused to obtain area feature values of each target fire point region. The displacement feature values are used to indicate the positional movement intensity of the corresponding target fire point region, and the area feature values are used to indicate the area growth intensity of the corresponding target fire point region. Based on the displacement characteristic value and area characteristic value of each target fire point area, a fire point risk assessment is performed on each target fire point area to obtain a risk characteristic value for each target fire point area. The risk characteristic value is used to indicate the degree of danger of the corresponding target fire point area. Fire situation analysis is performed based on the risk characteristic values of multiple target fire areas to obtain fire situation analysis results, wherein the fire situation analysis results are used to indicate the severity of the fire at the target fire site.
6. The method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The displacement feature value is calculated by weighting multiple corresponding displacement rates. The calculation weight of the displacement rate is negatively correlated with its corresponding time-domain monitoring distance. The area feature value is calculated by weighting multiple corresponding area change rates. The calculation weight of the area change rate is negatively correlated with its corresponding time-domain monitoring distance. The time-domain monitoring distance is the time difference between the image acquisition time of the corresponding infrared image and the current time.
7. The method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The steps for conducting a fire risk assessment for each target fire point region based on its displacement and area characteristic values, and obtaining the risk characteristic value for each target fire point region, include: The displacement characteristic value and area characteristic value of each target fire point region are normalized to obtain the displacement normalization value and area normalization value of each target fire point region. The displacement normalization value and area normalization value of each target fire point area are weighted and calculated to obtain the risk characteristic value of each target fire point area, wherein the calculation weight of the displacement normalization value is greater than the calculation weight of the area normalization value.
8. The fire dynamic monitoring method applicable to UAVs according to claim 5, characterized in that, The steps for conducting fire situation analysis based on the risk characteristic values of multiple target fire points to obtain fire situation analysis results include: Among the multiple target fire zones, the target fire zone with a risk characteristic value greater than or equal to a risk threshold is determined as the first fire zone, and the target fire zone with an area greater than or equal to an area threshold is determined as the second fire zone. If the proportion of the first fire point areas is greater than the proportion threshold, and / or the total area of the second fire point areas is greater than the area accumulation threshold, output a fire analysis result indicating a serious fire risk. If the proportion of the first fire point areas is less than or equal to the proportion threshold, and the total area of the second fire point areas is less than or equal to the cumulative area threshold, output a fire analysis result indicating general fire risk.
9. The method for dynamic fire monitoring applicable to unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, After conducting fire situation analysis based on the risk characteristic values of multiple target fire point areas and obtaining the fire situation analysis results, the method further includes: The movement trajectories of multiple target fire point areas are acquired and visualized in the 3D model corresponding to the target fire site. The area of the target fire point area is positively correlated with the area of the visualized area of the corresponding movement trajectory.
10. A dynamic fire monitoring system suitable for unmanned aerial vehicles (UAVs), characterized in that, The system includes: The initial detection module is used to detect fire points in the infrared image of the target fire site based on a preset temperature threshold, and obtain multiple initial fire point regions. The temperature threshold is the minimum temperature value in the fire point detection threshold range, and the temperature value of any image point in the initial fire point region is greater than or equal to the temperature threshold. The first processing module is used to accumulate the temperature values of image points corresponding to the initial fire point region in each of the multiple image grids corresponding to the infrared image to obtain the thermal radiation index of each image grid, wherein the thermal radiation index is used to indicate the intensity of heat radiation emitted by the corresponding image grid to its neighboring image grids; The second processing module is used to fuse the thermal radiation indices of multiple neighboring image grids corresponding to each image grid in the plurality of image grids to obtain the neighborhood radiation value of each image grid, wherein the neighborhood radiation value is used to indicate the degree to which the temperature of the corresponding image grid increases due to the thermal radiation of its neighboring image grids. A threshold adjustment module is used to increase the temperature threshold based on the neighborhood radiation value of each image grid to obtain a grid threshold corresponding to each image grid, wherein the increase in the temperature threshold is positively correlated with the neighborhood radiation value; The re-inspection module is used to detect fire points in the infrared image of the target fire site based on the grid threshold corresponding to each image grid, and obtain multiple target fire point areas.