Near real-time fire point dynamic monitoring method, device, equipment and medium

Abstract

The application provides a near real-time fire point dynamic monitoring method, device, equipment and medium, relates to the technical field of infrared remote sensing monitoring, and the method comprises the following steps: acquiring thermal infrared remote sensing data and land cover product of a forest and grass monitoring area, and performing data preprocessing; the thermal infrared remote sensing data is segmented to obtain tile data, and cloud extraction is performed based on the tile data; the extracted cloud is used for mask of the thermal infrared remote sensing data, and the medium infrared band and the far infrared band are segmented to obtain an isothermal graph spot layer; the isothermal graph spot layer is masked through water body and building area corresponding to the land cover product; the masked isothermal graph spot is comprehensively identified based on the statistical characteristics of the isothermal graph spot and the comparison characteristics of the isothermal graph spot and adjacent background graph spots, and a fire point detection result in the forest and grass monitoring area is obtained. The application alleviates the technical problems of pixel-by-pixel identification, large amount of calculation, low efficiency and limited precision in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of infrared remote sensing monitoring, and in particular to a near-real-time fire point dynamic monitoring method, device, equipment and medium. BACKGROUND

[0002] Forest and grassland fire is a natural phenomenon that often occurs and is difficult to predict and prevent, plays a very important role in the biochemical cycle of the earth, and once out of control, will cause huge losses to forest and grass resources, ecological environment and life and property. In related technologies, the fire point monitoring algorithm is a pixel-based analysis method. Whether it is cloud and water mask extraction, background window determination, potential fire point identification and secondary review, or false detection fire point elimination, it is all based on certain threshold conditions, and is identified pixel by pixel, which is large in calculation amount, low in efficiency and limited in precision. SUMMARY

[0003] The purpose of the present application is to provide a near-real-time fire point dynamic monitoring method, device, equipment and medium to alleviate the technical problems of pixel-by-pixel identification, large calculation amount, low efficiency and limited precision in the prior art fire point detection.

[0004] In a first aspect, the embodiments of the present application provide a near-real-time fire point dynamic monitoring method, comprising:

[0005] Obtaining thermal infrared remote sensing data and land cover product of a forest and grass monitoring area, and performing data preprocessing on the thermal infrared remote sensing data and the land cover product;

[0006] Segmenting the thermal infrared remote sensing data to obtain tile data, and performing cloud extraction based on the tile data;

[0007] Masking the thermal infrared remote sensing data based on the extracted cloud, and segmenting the mid-infrared band and the far-infrared band of the masked data to obtain a homothermal map spot layer;

[0008] Masking the homothermal map spot layer through water body and building area corresponding to the land cover product;

[0009] Performing fire point comprehensive identification on the masked homothermal map spot based on the statistical characteristics of the homothermal map spot and the comparison characteristics of the homothermal map spot and the adjacent background map spot, to obtain a fire point detection result in the forest and grass monitoring area.

[0010] Optionally, the data preprocessing on the thermal infrared remote sensing data and the land cover product comprises:

[0011] Performing radiation correction and geometric correction on the thermal infrared remote sensing data, so that the thermal infrared remote sensing data and the land cover product are in the same coordinate system;

[0012] Performing data cropping according to the monitoring range of the forest and grass monitoring area;

[0013] extracting band data of the preset band, and performing time period division on the band data to obtain first time period monitoring data and second time period monitoring data.

[0014] Optionally, the thermal infrared remote sensing data is segmented to obtain tile data, and cloud extraction is performed based on the tile data, including:

[0015] The thermal infrared remote sensing data is segmented according to a preset size to obtain tile data;

[0016] Based on the statistical value of each tile in the preset band, large-area low-temperature clouds are extracted, and cloud edges are extracted through iterative growth.

[0017] Optionally, based on the statistical value of each tile in the preset band, large-area low-temperature clouds are extracted, and cloud edges are extracted through iterative growth, including:

[0018] For tile data of a night period, a first statistical value is determined based on a mean value and a standard deviation of a first preset number of pixels in each tile in the 15th band, and for tile data of a daytime period, a second statistical value is determined based on a mean value and a standard deviation of a second preset number of pixels in each tile in the 1st band on the basis of the night data;

[0019] For tile data of a night period, large-area low-temperature clouds and clouds with spatial differentiation exceeding a threshold value are extracted through the first statistical value, and for tile data of a daytime period, fragmented clouds with spatial differentiation exceeding a preset threshold value are extracted by adding the second statistical value corresponding to the 1st band;

[0020] Taking the extracted large-area low-temperature clouds and non-continuous high-spatial-differentiation clouds as seeds, an iterative growth method is used for multiple cycles of iteration until the edges of the extracted clouds are obtained.

[0021] Optionally, the thermal infrared remote sensing data is masked based on the extracted clouds, and the mid-infrared band and the far-infrared band of the masked data are segmented to obtain a homothermal graph spot layer, including:

[0022] Based on the extracted clouds, the thermal infrared remote sensing data is masked, and the homothermal graph spot segmentation is performed on the masked data based on the brightness values of the remote sensing image pixels of the 7th, 14th and 15th bands to determine the homothermal graph spot layer; the homothermal graph spot layer is in a lower layer of the tile layer, and has a topological association relationship between the upper and lower layers.

[0023] Optionally, the homothermal graph spot layer is masked by the water body class and the building area class corresponding to the land cover product, including:

[0024] The homothermal graph spot layer is water body masked and building area masked by the land cover product;

[0025] The discrimination condition of the water body masking is For water bodies, 1 represents the water body code; the discrimination criteria for building area masks are: 7 represents the building area, and 7 is the building area code. The building area includes fixed heat sources.

[0026] Optionally, based on the statistical characteristics of isothermal patches and the comparative characteristics of isothermal patches with adjacent background patches, fire point identification is performed on the masked isothermal patches to obtain fire point detection results in the forest and grassland monitoring area, including:

[0027] Calculate the difference between the mean values ​​of band 7 and band 14 for each isothermal patch. The weighted average of the differences in the 7th band between the target patch and the background patch with a common boundary. The difference between the target patch and the mean value of the 7th band of its upper layer tile. The calculation formula is as follows:

[0028] (1)

[0029] (2)

[0030] In the formula, For all n pixel values ​​in the j-th band that constitute the isothermal patch or tile The mean; and These are the average values ​​of bands 7 and 14 of the isothermal plot, respectively.

[0031] (3)

[0032] In the formula, The length of the common boundary between the target patch and its adjacent background patches; For the target patch and the first The length of the common boundary of adjacent background patches; The mean value of the 7th band of the target image patch; For the first The average value of the 7th band of adjacent background patches; This represents the number of adjacent background patches;

[0033] (4)

[0034] In the formula, The mean value of the 7th band of the target image patch; The mean value of the 7th band of the tile to which the target image patch belongs;

[0035] Fire spot identification is based on the comprehensive judgment of fire spots, which simultaneously meet the following three difference conditions:

[0036] (5)

[0037] wherein, is the difference between the mean values of the 7th band and the 14th band of the isotherm spot; is the weighted average of the difference between the mean values of the 7th band of the target spot and the background spot sharing a common boundary; is the difference between the mean values of the 7th band of the target spot and the upper tile to which the target spot belongs.

[0038] In a second aspect, the embodiments of the present application provide a near real-time fire point dynamic monitoring device, comprising:

[0039] A data acquisition and preprocessing module is configured to acquire thermal infrared remote sensing data and land cover product of a forest and grass monitoring area, and perform data preprocessing on the thermal infrared remote sensing data and the land cover product;

[0040] A cloud extraction module is configured to segment the thermal infrared remote sensing data to obtain tile data, and perform cloud extraction based on the tile data;

[0041] A cloud mask module is configured to mask the thermal infrared remote sensing data based on the extracted cloud, and segment the mid-infrared band and the far-infrared band of the masked data to obtain an isotherm spot layer;

[0042] A water body and building mask module is configured to mask the isotherm spot layer through a water body class and a building area class corresponding to the land cover product;

[0043] A fire point monitoring module is configured to perform comprehensive identification of the masked isotherm spot based on statistical features of the isotherm spot and comparison features of the isotherm spot and adjacent background spots, to obtain a fire point detection result in the forest and grass monitoring area.

[0044] In a third aspect, the embodiments of the present application provide an electronic device, comprising a processor and a memory, the memory storing computer executable instructions capable of being executed by the processor, and the processor executes the computer executable instructions to implement the near real-time fire point dynamic monitoring method of any one of the preceding embodiments.

[0045] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, the computer readable storage medium storing computer executable instructions, and when the computer executable instructions are called and executed by a processor, the computer executable instructions cause the processor to implement the near real-time fire point dynamic monitoring method of any one of the preceding embodiments.

[0046] The near real-time fire point dynamic monitoring method, device, equipment and medium provided by the present application solve the problems of large amount of calculation, low efficiency and partial or complete period of missed detection of the existing pixel-based fire point monitoring algorithm, and realize near real-time full-process dynamic monitoring of fire points. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the specific embodiments or the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and all other drawings obtained by those of ordinary skill in the art without creative effort based on these drawings are within the scope of protection of the present application.

[0048] Figure 1 A flow chart of a near real-time fire point dynamic monitoring method provided for an embodiment of the present application;

[0049] Figure 2 A result schematic diagram of adjacent four tile homograph spot segmentation provided for an embodiment of the present application;

[0050] Figure 3 A fire point and forest range superimposition schematic diagram provided for an embodiment of the present application;

[0051] Figure 4 A structural diagram of a near real-time fire point dynamic monitoring device provided for an embodiment of the present application;

[0052] Figure 5 A structural diagram of an electronic device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to make a clear and complete description of the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative effort are within the scope of protection of the present application.

[0055] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0056] The thermal infrared band is essential for fire point monitoring. According to Wien's displacement law, the middle infrared band (MIR) is usually selected for its high sensitivity to high temperature of fire points, and the far infrared band (FIR) is usually selected for its sensitivity to normal temperature background objects. The basic principle is that the middle infrared band of fire points is abnormally increased compared with the surrounding background objects, and the middle infrared band is more significantly increased than the far infrared band. The fire point is identified by the abnormal change of the thermal infrared band. Visible and near infrared bands are often used as auxiliary bands for cloud and water mask extraction in the early stage, and for removing high-brightness false fire points in the later stage, or for further identifying smoke plumes by the human eye to confirm fire points.

[0057] Traditional fire point monitoring identifies abnormal high temperature pixels during satellite overpass by manual or automatic detection algorithm, which is called fire point. Although this application uses the same temperature map spot as the recognition unit, it still uses the traditional "fire point" terminology. The global threshold fire point monitoring algorithm based on temperature anomaly change monitoring began in the mid-1980s, and experienced the exploration and development period of AVHRR data in the last century. The MODIS fire point monitoring algorithm developed in the early 21st century was designed for its global fire monitoring product, and was consistent with the AVHRR algorithm. In the process of years of business operation, it has been continuously optimized and improved, and has become mature, having a profound influence on the fire monitoring algorithm in other parts of the world. When the MODIS fire monitoring algorithm was first proposed in 1998, it used a global fixed threshold method. In 2003, the C4 version introduced a background window and a relative threshold, improving the algorithm's sensitivity to low-temperature small fire points. The C6 version released in 2016 developed from the initial global fixed threshold to block dynamic threshold (each block is 30 rows x 301 columns in size) or adaptive threshold, further improving the algorithm's sensitivity to low-temperature small fire points and regional adaptability.

[0058] Currently, all reported fire point monitoring algorithms are based on pixel analysis methods. Whether it is cloud and water mask extraction, background window determination, potential fire point identification and secondary review, or false fire point removal, it is all based on certain threshold conditions and pixel-by-pixel identification, which is computationally intensive, inefficient, and limited in accuracy. The main problems are as follows:

[0059] 1) Existing pixel-level fire point recognition algorithms are all based on brightness temperature threshold pixel-by-pixel identification. Brightness temperature refers to the temperature of a black body that radiates the same amount of radiation energy as the observed object. Since objects in nature are not perfect black bodies, their emissivity is less than 1, so the brightness temperature is lower than the actual temperature of the object. Inverting DN values to brightness temperature has no practical significance for near-real-time fire point detection and warning, and increases processing steps, reduces processing efficiency, and reduces the distinguishability of temperature change anomalies, resulting in partial or complete missed detection of fire points. For example, the difference in the middle infrared band between fire points and surrounding background is more than 300 in DN value, but the brightness temperature is only a few K.

[0060] 2) Daytime cloud area reflection of solar radiation is the main cause of false detection of fire points. The multi-threshold cloud mask detection method used by NOAA-AVHRR, MODIS, VIIRS, Himawari8 / 9 AHI, FY3-MERSI, etc. is to define the pixels with high reflectance in the red and near-infrared bands and low brightness temperature in the thermal infrared band through reflectivity or brightness temperature and its ratio and difference. It is effective for large low-temperature clouds, but the problem of missing detection of fragmented clouds and cloud edges has not been solved. Although fire points will not be misdetected as clouds, the missed clouds will often be misdetected as fire points.

[0061] 3) The existing fire point monitoring algorithm is based on pixel analysis, which has high complexity due to the inability to use semantic information. For example, the MODIS C6 version of the fire point monitoring algorithm requires physical quantity inversion, cloud and water mask extraction, followed by global or dynamic local threshold identification of potential fire points, secondary review of potential fire points using a background window and relative threshold, and finally using a multi-band threshold method to remove misdetected fire points such as solar flares and high-brightness ground objects. Each identification step is based on pixels, which is computationally intensive, inefficient, and has partial or complete missed detection, making it impossible to achieve full-process dynamic monitoring of fire points, which seriously affects the implementation of subsequent rescue measures.

[0062] Therefore, the embodiments of the present application provide a near-real-time fire point dynamic monitoring method, device, equipment and medium, which solves the problem of large computational complexity, low efficiency, and partial or complete missed detection of the existing pixel-based fire point monitoring algorithm, and achieves near-real-time full-process dynamic monitoring of fire points.

[0063] The embodiments of the present application provide a near-real-time fire point dynamic monitoring method, as shown in Figure 1 The method comprises the following steps:

[0064] Step S110, acquiring thermal infrared remote sensing data and land cover product of the forest and grass monitoring area, and performing data preprocessing on the thermal infrared remote sensing data and the land cover product.

[0065] In an embodiment, the acquired thermal infrared remote sensing data and land cover product of the forest and grass monitoring area can be, for example, Himawari-8 data covering China on a certain day (e.g., 38 periods from 6:50 to 13:00 UTC on October 22, 2022) and ESRI 2022 land cover product.

[0066] Step S120, segmenting the thermal infrared remote sensing data to obtain tile data, and performing cloud extraction based on the tile data.

[0067] Step S130, performing mask on the thermal infrared remote sensing data based on the extracted cloud, and segmenting the mid-infrared band and the far-infrared band of the masked data to obtain isothermal map spot layer.

[0068] Step S140, the isotherm spot layer is masked by the water body class and the building area class corresponding to the land cover product.

[0069] Step S150, the masked isotherm spot is comprehensively identified based on the statistical features of the isotherm spot and the comparison features of the isotherm spot and the adjacent background spot, and a fire point detection result in the forest and grass monitoring area is obtained.

[0070] The near real-time fire point dynamic monitoring method provided by the embodiment of the application is described in detail below.

[0071] In an embodiment, the data preprocessing of the thermal infrared remote sensing data and the land cover product can include the following steps 1-1 to 1-3:

[0072] Step 1-1, the thermal infrared remote sensing data is radiometrically and geometrically corrected to make the thermal infrared remote sensing data and the land cover product have the same coordinate system;

[0073] Step 1-2, the data is cropped according to the monitoring range of the forest and grass monitoring area;

[0074] Step 1-3, the waveband data of the preset waveband is extracted, and the waveband data is divided into time periods to obtain first time period monitoring data and second time period monitoring data.

[0075] In an example, the remote sensing data is first radiometrically and geometrically corrected to ensure that the remote sensing data and the land cover product have the same coordinate system. Then, the data is cropped according to the monitoring area range. In this embodiment, about 150,000 km 2 of the monitoring area as the core, north latitude 37°-42°, east longitude 115°-118°, which is equivalent to a medium-sized provincial area, and the corresponding thermal infrared data size is 150 columns x 250 rows. Three thermal infrared wavebands, waveband 7, 14 and 15, are extracted and denoted as 、 、 , the visible light and near infrared wavebands are increased in the daytime, which are waveband 1, 2, 3 and 4, denoted as 、 、 、 . The 2022 land cover product is down-sampled to 1 km, denoted as , which are combined into daytime and nighttime monitoring data and . Among them, , .

[0076] Further, the thermal infrared remote sensing data is segmented to obtain tile data, and cloud extraction is performed based on the tile data, which can include the following steps 2-1 and 2-2:

[0077] Step 2-1, the thermal infrared remote sensing data is segmented according to a preset size to obtain tile data. In an embodiment, based on the first law of geography, spatial similarity decays with distance, appropriate parameters are selected to divide the data into tiles. The appropriate parameters are obtained through multiple trials according to the spatial resolution of the data used and the scale of the research target. The spatial resolution of the thermal infrared data is 2 km, and since the research target is large-scale cloud and grass, different odd parameters of 3, 5, 7, and 9 are tried, and finally it is found that the parameter 5, i.e. 10 km, has good effect and can be used as the preferred embodiment, without sacrificing too much clear sky, and can solve the problem of fragmented cloud and fuzzy transition of cloud edge. In an example, the data can be segmented into tiles of 10 km x 10 km size, denoted as Level 0.

[0078] Step 2-2, based on the statistical value of each tile in a preset band, large low-temperature clouds are extracted, and the cloud edge is extracted through iterative growth. Optionally, the following steps 2-2-1 to 2-2-3 can be further included:

[0079] Step 2-2-1, for tile data in the night period, a first statistical value is determined based on the mean and standard deviation of a first preset number of pixels in each tile in the 15th band, and for tile data in the daytime period, a second statistical value is determined based on the mean and standard deviation of a second preset number of pixels in each tile in the 1st band on the basis of the night data.

[0080] In an embodiment, the night statistics of the 15th band The mean of 25 pixels in each tile And the standard deviation . On this basis, the daytime statistics of the 1st band The mean of 100 pixels in each tile And the standard deviation .

[0081] Step 2-2-2, for tile data in the night period, large low-temperature clouds and clouds with spatial heterogeneity exceeding a threshold value are extracted through the first statistical value, and for tile data in the daytime period, fragmented clouds with spatial heterogeneity exceeding a preset threshold value are extracted by increasing the second statistical value corresponding to the 1st band.

[0082] The mean And the standard deviation of the 15th band are used to extract large low-temperature clouds and clouds with high spatial heterogeneity, and the standard deviation of the 1st band is increased Extract the broken cloud with high spatial heterogeneity. Then, take the extracted cloud as a seed and grow outward using the iterative growth method until the edge of the cloud. Existing cloud extraction is based on pixels, so it can only be based on the reflection value or radiation value of a single pixel in multiple bands, and the ratio or difference between bands, etc. The algorithm is complex, and can only effectively extract large low-temperature clouds, and has no effect on non-low-temperature broken clouds and the edges of clouds. The embodiments of the present application can define non-low-temperature broken clouds with high spatial heterogeneity by statistical characteristics by dividing into tiles, and extract the edges of the clouds by the iterative growth method, solving the problem of long-term broken cloud and cloud edge missing detection.

[0083] 1) Low-temperature cloud extraction, taking the mean value of the 15th band of the tile as the discriminant condition to extract large low-temperature clouds. The mean value calculation formula of the 15th band and the 1st band of the tile is as follows:

[0084]

[0085]

[0086] In the formula, and are the average values of all n pixel values of the 15th band and the 1st band of the tile, respectively.

[0087] 2) High spatial heterogeneity cloud extraction, taking the standard deviation of the 15th band as the discriminant condition to extract non-continuous clouds with high spatial heterogeneity; adding the standard deviation of the 1st band as the discriminant condition during the day to extract broken clouds.

[0088]

[0089]

[0090] In the formula, and are the standard deviations calculated by the 15th band and the 1st band of the tile, respectively.

[0091] ​​​​​​​The determination method is to collect different types of cloud and underlying surface samples in the research area, and to count the mean and standard deviation of each band of different size tiles, and to compare the obtained empirical values. This method uses less features and is simple, only 15 bands and 1 band are used; after dividing the tiles, the processing unit is reduced, and the processing efficiency is improved; the mean and standard deviation of the DN values of the two bands are directly used, and the threshold setting is relatively loose, and a little larger or a little smaller does not affect the final result. If the low-temperature cloud extraction and high spatial heterogeneity cloud extraction threshold is set small, then the cloud edge extraction iterative times can be slightly larger; and the seed cloud extraction and iterative growth method are combined to solve the problems of fragmented cloud and missed detection of cloud edge existing in the existing algorithms for a long time.

[0092] Step 2-2-3, taking the extracted large low-temperature cloud and non-continuous high spatial heterogeneity cloud as seeds, using the iterative growth method, multiple cycle iterations are performed until the edge of the extracted cloud is obtained.

[0093] In an embodiment, the number of iterations can be selected to be 10 times until the edge of the cloud is extracted. The discrimination condition is that there is a common boundary with the existing cloud and the standard deviation of the 15th band is greater than 100. .

[0094] According to the characteristics that the temperature difference of the forest and grass area under clear sky is generally small, and the standard deviation of the 15th band is generally low, a certain tile is taken as a cloud when the common boundary with the existing cloud tile is greater than zero and the standard deviation of the 15th band of the tile is greater than 100. This discrimination condition extracts thick clouds and thin clouds in the transition area of clear sky, and does not misdetect the forest and grass under clear sky.

[0095] Further, the hot infrared remote sensing data is masked based on the extracted cloud, and the mid-infrared band and the far-infrared band of the masked data are segmented to obtain a homothermal spot layer. The hot infrared remote sensing data can be masked based on the extracted cloud, and the homothermal spot layer is determined by segmenting the remote sensing image brightness value of the 7th, 14th and 15th bands of the masked data. The homothermal spot layer is in the lower layer of the tile layer, and the upper and lower layers have a topological association relationship.

[0096] The homothermal spot segmentation is to solve the problem of large-scale temperature heterogeneity. Through clustering of the DN values of the three infrared bands, pixels with adjacent positions and similar temperatures are clustered into the same spot. Each spot is composed of one or more pixels, but will not exceed the range of the tile to which it belongs. As shown in FIG. 6, the results of homothermal spot segmentation of four adjacent tiles are shown. Figure 2

[0097] Further, the homothermal spot layer can be masked by water body and building area corresponding to the land cover product. The homothermal spot layer can be water body masked and building area masked by the land cover product; wherein the discrimination condition for water body masking is​ Water body, 1 is the code of water body; the discrimination condition of building area mask is Building area, 7 is the code of building area, and the building area includes fixed heat source.

[0098] Further, the fire point comprehensive identification is carried out on the mask thermogram spot based on the statistical characteristics of the thermogram spot and the comparison characteristics of the thermogram spot and adjacent background spot, and the fire point detection result in the forest and grass monitoring area is obtained, which can include the following steps 3-1 and 3-2:

[0099] Step 3-1, calculating the difference between the mean values of the 7th band and the 14th band of each thermogram spot , the weighted average of the difference between the mean values of the 7th band of the target spot and the background spot with common boundary , the difference between the mean values of the 7th band of the target spot and the upper tile to which it belongs , and the calculation formula is as follows:

[0100] (1)

[0101] (2)

[0102] In the formula, is the mean value of all n pixel values of the jth band constituting the thermogram spot or tile; and are the mean values of the 7th band and the 14th band of the thermogram spot, respectively;

[0103] (3) In the formula,

[0104] is the length of the common boundary between the target spot and the adjacent background spot; is the length of the common boundary between the target spot and the jth adjacent background spot; is the mean value of the 7th band of the target spot; is the mean value of the 7th band of the jth adjacent background spot; is the number of adjacent background spots. (4)

[0105] In the formula,

[0106] In the formula, is the mean value of the 7th band of the target spot; is the mean value of the 7th band of the tile to which the target spot belongs.

[0107] Step 3-2, the fire point comprehensive identification, and the following three difference conditions are met simultaneously for the fire point spot:

[0108] ​​ (5)

[0109] wherein, is the difference between the average of the 7th band and the 14th band of each isothermic spot; is the weighted average of the difference between the average of the 7th band of the target spot and the background spot with common boundary; is the difference between the average of the 7th band of the target spot and the upper tile to which the target spot belongs.

[0110] In addition, after obtaining the above detection result, the detected fire point can be superimposed with the forest and grass range, and the fire point detection result in the forest and grass range is packaged and exported, including infrared composite image map in GeoTiff format, fire point map in shapefile format and attribute data thereof, for analysis and decision of forest fire management personnel. When superimposing, the forest and grass distribution range composed of a series of spatial coordinates is superimposed, as shown in Figure 3 , the irregular polygon is the working area, the gray spot is the forest and grass area, and the rectangular frame is the fire point spot.

[0111] The present application will be further described in detail below with Himawari8 / 9 data as an example, in combination with the drawings and specific embodiments.

[0112] The radiation imager (AHI) carried by the stationary meteorological satellite Himawari-8 / 9 includes 16 channels, which are composed of 3 visible light channels, 3 near-infrared channels and 10 infrared channels. Among them, the resolution of the visible red band is 0.5km, the resolution of the blue, green and near-infrared bands is 1km, and the resolution of the remaining near-infrared and infrared bands is 2km. The full observation frequency is once every 10 minutes, and the two satellites are expected to continue to operate until 2029.

[0113] The Planck law gives the quantitative relationship between the exitance of black body radiation and temperature, wavelength. The Wien displacement law and the Stefan-Boltzmann law derived from the Planck formula together build a solid physical foundation for infrared remote sensing fire point monitoring. The Wien displacement law gives the quantitative relationship between the emission peak wavelength of a black body and the temperature. With the increase of the temperature of the black body, the emission peak wavelength moves to the short wave direction, and the two are inversely proportional The temperature of the surface object is generally between -40℃ and +40℃, the average background temperature is 27℃ (equivalent to 300K), and the peak wavelength of its radiation is in the far infrared band of 9.7um, corresponding to the 14th band of Himawari-8 / 9 AHI, the center wavelength is 11.2um, and the quantization level is 12bit or 4096 gray levels; while the temperature of the forest and grass burning reaches about 500-1000K, and the peak wavelength of its radiation is in the mid-infrared band of 3-5um, corresponding to the 7th band of Himawari-8 / 9 AHI, the center wavelength is 3.9um, and the quantization level is 14bit or 16384 gray levels. The Stefan-Boltzmann law describes that the total emission radiation increases rapidly with the increase of the temperature of the black body, that is, the radiation intensity of the black body is proportional to the 4th power of the temperature That is, even if the fire point area only accounts for a very small proportion of the pixel area, it will cause the temperature of the entire pixel to surge, especially in the mid-infrared band, which forms a strong contrast with the surrounding background and becomes the main basis for fire point identification.

[0114] Since the high-temperature radiation of the fire point cannot penetrate the cloud layer, the prerequisite for fire point monitoring is clear sky without clouds. First, cloud monitoring is required. The thickness and shape of clouds in nature change rapidly, and the boundary is blurred. Cloud detection is mainly based on the radiation characteristics and geometric texture features of the cloud. Compared with the underlying surface, the cloud has strong reflection in the visible and near-infrared bands, and low radiation in the thermal infrared band, and its reflection and radiation spatial differentiation is high. The cloud shows low temperature in the 15th band with a center wavelength of 12.4um, therefore, the 7th, 14th and 15th bands of Himawari-8 / 9 AHI are selected for fire point monitoring and cloud identification at night. The RGB color synthesis of the 7th, 14th and 15th bands of Himawari-8 / 9 AHI, the fire point pixel is pink, the surrounding background is blue-gray, and the cloud is dark red or dark brown or dark blue mottled color. During the day, some small clouds will also appear pink, but the cloud is distributed in groups and drifts, and the fire point position is relatively fixed. During the day, the edges of the small clouds and the clouds do not show low temperature characteristics, but they have strong reflection in the visible and near-infrared bands, therefore, the blue, green, red and near-infrared bands (1, 2, 3 and 4 bands respectively) are added as auxiliary bands to identify the edges of the small clouds and the clouds. Forest and grass burning often causes smoke and dust in the fire area and the downwind area, and the visible and near-infrared band images can clearly reflect the smoke and dust information, which is used for the human eye to further confirm the fire point.

[0115] In addition to cloud pollution, water, bare ground, rock, and glass buildings, etc. cause abnormal thermal infrared. In addition, fixed heat sources such as steel mills and power plants are also misdetected as fire points. The exclusion of water, high-brightness buildings, and fixed heat sources requires the help of land cover products. The ESRI global land cover product has been updated once a year since 2017, and is divided into 11 categories, including water, forest, grassland, waterlogged vegetation, farmland, shrubs, building area, bare land, glaciers and permanent snow, clouds and others, based on 10m resolution Sentinel-2 optical satellite 4 visible light bands of blue, green, red, near-infrared and 2 short-wave infrared bands, using 5 billion different seasonal pixels around the world as samples, training deep learning models to produce, and the overall accuracy is more than 85%.

[0116] In order to illustrate the near real-time dynamic monitoring effect of the embodiments of the present application, the monitoring results of the embodiments of the present application are compared with the traditional fire point monitoring algorithm and the monitoring results of JAXA, see Table 1. It can be seen that the traditional fire point monitoring algorithm only monitors the fire at two time periods of UTC 8:30 and 8:40, and cannot realize dynamic monitoring throughout the process. The fire point detection algorithm of JAXA, which is not disclosed, detects fire points from 8:20, and continues until 12:30, with a small part of the time period not detected, which can realize dynamic monitoring of the whole process of fire points. The embodiments of the present application detect fire points for the first time from 8:00, and continue until 12:20, which not only can realize near real-time dynamic monitoring of the whole process of fire points, but also can detect fire points 20 minutes in advance, which can save valuable time for timely rescue measures.

[0117] From the detection results of the embodiments of the present application, the fire time is before UTC 6:50, but the fire point temperature is always lower than the detection condition for more than 1 hour after that. The fire point is detected for the first time at 8:00, the fire spreads rapidly to the east and northeast at 8:20, the fire decreases at 8:30, and the fire attacks again to the east and northeast at 8:50. The fire decreases at 9:10, the fire goes to the north at 9:20, the fire retreats to the west at 9:30, the fire reaches the maximum range at 9:40, and the fire continues for 90 minutes. The fire retreats to the west at 11:20, the fire increases to the maximum range again at 12:00, and continues for 20 minutes, and the fire retreats to the east at 12:20, and the fire point temperature is lower than the detection condition thereafter until 13:00.

[0118] Table 1 Detection results of different ways of fire point detection

[0119]

[0120]

[0121]

[0122] Compared with the existing fire point monitoring algorithm, the advantages of the embodiments of the present application mainly include:

[0123] 1) The inversion steps of physical quantities such as brightness temperature and reflectivity are saved, and the timeliness of fire point monitoring is improved.

[0124] 2) The tile-based cloud mask extraction technology greatly reduces the processing units and improves the processing efficiency, and is not only suitable for large low-temperature clouds, but also solves the problems of broken clouds and cloud edge missed detection in the existing cloud mask extraction algorithm.

[0125] 3) The fire point recognition method based on the same temperature map spot, combined with the statistical characteristics of the target spot and the comparison characteristics between the adjacent background spots, comprehensively identifies the algorithm, which is simple, has less calculation amount and high efficiency, and a general notebook computer runs 100 million km 2 The fire point detection algorithm takes less than 5 seconds, and the near real-time dynamic monitoring of the whole process from the beginning to the final extinguishing of the fire point is realized.

[0126] Based on the above method embodiments, the embodiments of the present application also provide a near real-time fire point dynamic monitoring device, as shown in Figure 4 The device mainly includes the following parts:

[0127] The data acquisition and preprocessing module 410 is used for acquiring thermal infrared remote sensing data and land cover product of the forest and grass monitoring area, and performing data preprocessing on the thermal infrared remote sensing data and the land cover product;

[0128] The cloud extraction module 420 is used for segmenting the thermal infrared remote sensing data to obtain tile data, and extracting clouds based on the tile data;

[0129] The cloud mask module 430 is used for masking the thermal infrared remote sensing data based on the extracted clouds, and segmenting the mid-infrared band and the far-infrared band of the masked data to obtain a same temperature map spot layer;

[0130] The water body and building mask module 440 is used for masking the same temperature map spot layer through the water body class and the building area class corresponding to the land cover product;

[0131] The fire point monitoring module 450 is used for comprehensively identifying the fire point of the masked same temperature map spot based on the statistical characteristics of the same temperature map spot and the comparison characteristics between the same temperature map spot and the adjacent background spot, and obtaining the fire point detection result in the forest and grass monitoring area.

[0132] Optionally, the data acquisition and preprocessing module 410 is used for:

[0133] Radiation correction and geometric correction are performed on the thermal infrared remote sensing data, so that the thermal infrared remote sensing data and the land cover product are in the same coordinate system;

[0134] Crop monitoring area according to the monitoring range of data pruning;

[0135] Extract the wave band data of the preset wave band, and divide the wave band data into time periods to obtain first time period monitoring data and second time period monitoring data.

[0136] Optionally, the cloud extraction module 420 is also used for:

[0137] The thermal infrared remote sensing data is divided into tile data according to the preset size;

[0138] Based on the statistical value of each tile in the preset wave band, large low-temperature clouds are extracted, and the cloud edges are extracted through iterative growth.

[0139] Optionally, the cloud extraction module 420 is also used for:

[0140] For tile data in the night period, the first statistical value is determined based on the mean and standard deviation of the first preset number of pixels in each tile in the 15th wave band, and for tile data in the daytime period, the second statistical value is determined based on the mean and standard deviation of the second preset number of pixels in each tile in the 1st wave band based on the night data;

[0141] For tile data in the night period, large low-temperature clouds and clouds with spatial differentiation exceeding the threshold value are extracted through the first statistical value, and for tile data in the daytime period, fragmented clouds with spatial differentiation exceeding the preset threshold value are extracted by adding the second statistical value corresponding to the 1st wave band;

[0142] Using the extracted large low-temperature clouds and non-continuous high spatial differentiation clouds as seeds, an iterative growth method is used for multiple cycles of iteration until the edges of the extracted clouds are obtained.

[0143] Optionally, the water body building mask module 440 is used for:

[0144] Based on the extracted clouds, the thermal infrared remote sensing data is masked, and the same temperature spot layer is determined by performing same temperature spot segmentation on the brightness values of the remote sensing image pixels of the 7th, 14th and 15th wave bands on the masked data. The same temperature spot layer is in the lower layer of the tile layer, and the top and bottom layers have a topological association relationship.

[0145] Optionally, the same temperature spot layer is masked by the water body class and the building area class corresponding to the land cover product, including:

[0146] The same temperature spot layer is masked by the water body and the building area by the land cover product;

[0147] The discrimination condition of the water body mask is Water, 1 is the water body code; the discrimination condition of the building area mask is Building area, 7 is the building area code, and the building area includes fixed heat sources.

[0148] Optionally, the fire detection module 450 is also used for:

[0149] Calculate the difference between the mean values ​​of band 7 and band 14 for each isothermal patch. The weighted average of the differences in the 7th band between the target patch and the background patch with a common boundary. The difference between the target patch and the mean value of the 7th band of its upper layer tile. The calculation formula is as follows:

[0150] (1)

[0151] (2)

[0152] In the formula, For all n pixel values ​​in the j-th band that constitute the isothermal patch or tile The average value; and These are the average values ​​of bands 7 and 14 of the isothermal plot, respectively.

[0153] (3)

[0154] In the formula, The length of the common boundary between the target patch and its adjacent background patches; For the target patch and the first The length of the common boundary of adjacent background patches; The mean value of the 7th band of the target image patch; For the first The average value of the 7th band of adjacent background patches; This represents the number of adjacent background patches;

[0155] (4)

[0156] In the formula, The mean value of the 7th band of the target image patch; The mean value of the 7th band of the tile to which the target image patch belongs;

[0157] Fire spot identification is based on the comprehensive judgment of fire spots, which simultaneously meet the following three difference conditions:

[0158] (5)

[0159] In the formula, This represents the difference between the mean values ​​of band 7 and band 14 of the isothermal patch. It is a weighted average of the differences between the target patch and the background patch with a common boundary in the 7th band. The difference between the target patch and the mean value of the 7th band of the upper tile to which the target patch belongs.

[0160] The near real-time fire point dynamic monitoring device provided by the embodiments of the present application has the same implementation principle and technical effects as the foregoing method embodiments. For brevity, the embodiments of the near real-time fire point dynamic monitoring device are not mentioned in the foregoing embodiments of the near real-time fire point dynamic monitoring method, and the corresponding content can be referred to in the foregoing embodiments of the near real-time fire point dynamic monitoring method.

[0161] The embodiments of the present application further provide an electronic device, as shown in the accompanying drawings, which is a structural schematic diagram of the electronic device. The electronic device 100 includes a processor 51 and a memory 50. The memory 50 stores computer executable instructions capable of being executed by the processor 51. The processor 51 executes the computer executable instructions to implement any of the foregoing near real-time fire point dynamic monitoring methods. Figure 5

[0162] In the embodiment shown in the accompanying drawings, the electronic device further includes a bus 52 and a communication interface 53. The processor 51, the communication interface 53 and the memory 50 are connected through the bus 52. Figure 5 The memory 50 can include a high-speed random access memory (RAM) and can further include a non-volatile memory, for example, at least one disk memory. The communication connection between the system network element and at least one other network element is implemented through at least one communication interface 53 (which can be wired or wireless). The communication connection can use the Internet, a wide area network, a local area network, a metropolitan area network, etc. The bus 52 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 52 can be divided into an address bus, a data bus, a control bus, etc. For brevity,

[0163] In the accompanying drawings, only one bidirectional arrow is used to represent the bus, but it does not mean that there is only one bus or only one type of bus. Figure 5

[0164] ​​The processor 51 can be an integrated circuit chip with processing capability. In implementation process, each step of the above method can be completed by integrated logic circuit of hardware in the processor 51 or by instructions in the form of software. The processor 51 described above can be a general processor, including a central processing unit (CPU), a network processor (NP), etc.; can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as a hardware code processor to execute, or be executed by a combination of hardware and software modules in the code processor. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, or other mature storage medium in the art. The storage medium is located in the storage, and the processor 51 reads the information in the storage, and combines the hardware to complete the steps of the near real-time fire point dynamic monitoring method of the foregoing embodiments.

[0165] The embodiment of the present application further provides a computer readable storage medium, which stores computer executable instructions. When the computer executable instructions are called and executed by a processor, the computer executable instructions cause the processor to implement the near real-time fire point dynamic monitoring method described above. For specific implementation, reference can be made to the foregoing method embodiments, and details are not described herein.

[0166] The near real-time fire point dynamic monitoring method, device, equipment and medium computer program product provided by the embodiment of the present application include a computer readable storage medium storing program codes. The instructions included in the program codes can be used to execute the method described in the foregoing method embodiments. For specific implementation, reference can be made to the method embodiments, and details are not described herein.

[0167] Unless otherwise specifically stated, the relative steps, numerical expressions and values of the components and steps set forth in these embodiments do not limit the scope of the present application.

[0168] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a nonvolatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present application or the part of the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0169] In the description of the present application, it should be noted that the terms "center", "upper", "lower", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship commonly used when the product is used, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0170] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A near real-time dynamic monitoring method for fire points, characterized in that, include: Acquire thermal infrared remote sensing data and land cover products of the forest and grassland monitoring area, and perform data preprocessing on the thermal infrared remote sensing data and the land cover products; The thermal infrared remote sensing data is divided into tile data according to a preset size; based on the statistical values ​​of a preset band in each tile, large areas of low-temperature clouds are extracted, and the cloud edges are extracted iteratively by increasing the size of the cloud. Based on the extracted cloud data, the thermal infrared remote sensing data is masked, and the mid-infrared and far-infrared bands of the masked data are segmented to obtain isothermal patch layers. The isothermal patch layer is masked by the water body type and building area type corresponding to the land cover product; Based on the statistical characteristics of isothermal patches and the comparative characteristics of the isothermal patches with adjacent background patches, fire points are comprehensively identified in the masked isothermal patches to obtain the fire point detection results in the forest and grassland monitoring area. Based on the statistical characteristics of isothermal patches and the comparative characteristics between the isothermal patches and adjacent background patches, a comprehensive fire point identification is performed on the masked isothermal patches to obtain the fire point detection results in the forest and grassland monitoring area, including: Calculate the difference between the mean values ​​of band 7 and band 14 for each isothermal patch. The weighted average of the differences in the 7th band between the target patch and the background patch with a common boundary. The difference between the target patch and the mean value of the 7th band of its upper layer tile. The calculation formula is as follows: （1） （2） In the formula, For all n pixel values ​​in the j-th band that constitute the isothermal patch or tile The mean; and These are the average values ​​of bands 7 and 14 of the isothermal plot, respectively. （3） In the formula, The length of the common boundary between the target patch and its adjacent background patches; For the target patch and the first The length of the common boundary of adjacent background patches; The mean value of the 7th band of the target image patch; For the first The average value of the 7th band of adjacent background patches; This represents the number of adjacent background patches; （4） In the formula, The mean value of the 7th band of the target image patch; The mean value of the 7th band of the tile to which the target image patch belongs; Fire spot identification is based on the comprehensive judgment of fire spots, which simultaneously meet the following three difference conditions: （5） In the formula, This represents the difference between the mean values ​​of band 7 and band 14 of the isothermal patch. It is a weighted average of the differences between the target patch and the background patch with a common boundary in the 7th band. It is the difference between the target patch and the mean value of the 7th band of the upper layer tile to which it belongs.

2. The near real-time dynamic fire point monitoring method according to claim 1, characterized in that, Data preprocessing of the thermal infrared remote sensing data and the land cover product includes: The thermal infrared remote sensing data is subjected to radiometric and geometric corrections to ensure that the thermal infrared remote sensing data and the land cover product are in the same coordinate system. Data was cropped according to the monitoring range of the forest and grassland monitoring area. Extract band data from preset bands and divide the band data into time periods to obtain monitoring data for the first time period and monitoring data for the second time period.

3. The near real-time dynamic fire point monitoring method according to claim 1, characterized in that, Based on the statistical values ​​of preset bands within each tile, large areas of low-temperature clouds are extracted, and the cloud edges are iteratively extracted through growth, including: For tile data during the nighttime period, a first statistical value is determined based on the mean and standard deviation of a first preset number of pixels in each tile of the 15th band. For tile data during the daytime period, a second statistical value is determined based on the mean and standard deviation of a second preset number of pixels in each tile of the 1st band, on the basis of the nighttime data. For tile data during the nighttime period, large areas of low-temperature clouds and clouds with spatial heterogeneity exceeding a threshold are extracted using the first statistical value. For tile data during the daytime period, fragmented clouds with spatial heterogeneity exceeding a preset threshold are extracted by adding the second statistical value corresponding to the first band. Using the extracted large areas of low-temperature clouds and non-contiguous high-spatial-differentiated clouds as seeds, an iterative growth method is employed, iterating multiple times until the edge of the extracted cloud is reached.

4. The near-real-time dynamic fire point monitoring method according to claim 1 or 3, characterized in that, Based on the extracted cloud data, the thermal infrared remote sensing data is masked, and the mid-infrared and far-infrared bands of the masked data are segmented to obtain isothermal patch layers, including: Based on the extracted cloud data, the thermal infrared remote sensing data is masked, and the masked data is segmented into isothermal patches using the pixel brightness values ​​of the remote sensing images in the 7th, 14th, and 15th bands to determine the isothermal patch layer; the isothermal patch layer is below the tile layer, and there is a topological relationship between the upper and lower layers.

5. The near real-time dynamic fire point monitoring method according to claim 1, characterized in that, Masking the isothermal patch layer using the water body type and building area type corresponding to the land cover product includes: The land cover product is used to mask the isothermal patch layer for water bodies and building areas; Among them, the discrimination criteria for water body mask are: For water bodies, 1 represents the water body code; the discrimination criteria for building area masks are: 7 represents the building area, and 7 is the building area code. The building area includes fixed heat sources.

6. A near real-time dynamic fire point monitoring device, characterized in that, include: The data acquisition and preprocessing module is used to acquire thermal infrared remote sensing data and land cover products of the forest and grassland monitoring area, and to perform data preprocessing on the thermal infrared remote sensing data and the land cover products. The cloud extraction module is used to divide the thermal infrared remote sensing data into tile data according to a preset size; based on the statistical values ​​of a preset band within each tile, large areas of low-temperature clouds are extracted, and the cloud edges are extracted iteratively through growth. The cloud masking module is used to mask the thermal infrared remote sensing data based on the extracted cloud, and to segment the mid-infrared and far-infrared bands of the masked data to obtain isothermal patch layers. A water body and building masking module is used to mask the isothermal patch layer based on the water body type and building area type corresponding to the land cover product. The fire detection module is used to comprehensively identify fire points in the masked isothermal patches based on the statistical characteristics of the isothermal patches and the comparative characteristics between the isothermal patches and adjacent background patches, so as to obtain the fire detection results in the forest and grassland monitoring area. The fire detection module is specifically used to calculate the difference between the mean values ​​of the 7th and 14th bands of each isothermal patch. The weighted average of the differences in the 7th band between the target patch and the background patch with a common boundary. The difference between the target patch and the mean value of the 7th band of its upper layer tile. The calculation formula is as follows: （1） （2） In the formula, For all n pixel values ​​in the j-th band that constitute the isothermal patch or tile The mean; and These are the average values ​​of bands 7 and 14 of the isothermal plot, respectively. （3） In the formula, The length of the common boundary between the target patch and its adjacent background patches; For the target patch and the first The length of the common boundary of adjacent background patches; The mean value of the 7th band of the target image patch; For the first The average value of the 7th band of adjacent background patches; This represents the number of adjacent background patches; （4） In the formula, The mean value of the 7th band of the target image patch; The mean value of the 7th band of the tile to which the target image patch belongs; Fire spot identification is based on the comprehensive judgment of fire spots, which simultaneously meet the following three difference conditions: （5） In the formula, This represents the difference between the mean values ​​of band 7 and band 14 of the isothermal patch. It is a weighted average of the differences between the target patch and the background patch with a common boundary in the 7th band. It is the difference between the target patch and the mean value of the 7th band of the upper layer tile to which it belongs.

7. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the near real-time fire point dynamic monitoring method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when invoked and executed by a processor, cause the processor to implement the near-real-time fire point dynamic monitoring method according to any one of claims 1 to 5.

Images