Onboard fire point de-aliasing method, device and equipment based on terrain combustibility constraint

Abstract

The application discloses a kind of on-satellite fire point false elimination method, device and equipment based on ground object combustibility constraint, belong to remote sensing technical field, method includes obtaining the candidate fire point pixel set detected based on the temperature characteristic detection of remote sensing image, and determine the target area corresponding to candidate fire point pixel set;Obtain the ground object combustibility binary image of target area;Candidate fire point pixel set and ground object combustibility binary image are carried out spatial overlay analysis, and the pixel value of each candidate fire point pixel in the corresponding position of ground object combustibility binary image is obtained;In the case where pixel value is 0, corresponding candidate fire point pixel is judged as false fire point and is eliminated, in the case where pixel value is 1, corresponding candidate fire point pixel is judged as real fire point and is retained.The application can efficiently eliminate false fire point in a lightweight, low resource-dependent manner, improve the accuracy and real-time performance of on-satellite fire point detection.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing technology, and specifically relates to a method, apparatus and equipment for despoofing on-board fire points based on ground object flammability constraints. Background Technology

[0002] Remote sensing monitoring of fire points is an important technical means for monitoring forest fires, grassland fires and other surface burning phenomena. Existing fire point detection methods are usually based on the brightness and temperature anomalies in the mid-infrared and thermal infrared bands, and candidate fire points are obtained through threshold determination or context analysis.

[0003] However, in practical applications, non-combustible ground features such as bare ground, impermeable urban surfaces, and industrial facilities may also produce high brightness temperature responses. Relying solely on brightness temperature characteristics for fire point determination can easily lead to a large number of false fire points.

[0004] To reduce the number of false fire points, various methods for removing false fire points have been proposed in the existing technology, including secondary screening based on brightness and temperature thresholds, posterior judgment based on multi-temporal consistency, re-identification methods based on complex land cover classification or machine learning models, and manual rule correction methods.

[0005] The methods described above can suppress false fire points to some extent, but they generally suffer from high computational complexity, strong dependence on multi-temporal or multi-band data, complex implementation, or high resource consumption, making them difficult to adapt to the limited computing and storage resources available on satellites. Summary of the Invention

[0006] This application provides a method, apparatus, and device for removing false fire points on satellites based on the flammability constraints of ground features. It can efficiently eliminate false fire points in a lightweight and low-resource-dependent manner, thereby improving the accuracy and real-time performance of on-satellite fire point detection.

[0007] On one hand, embodiments of this application provide a method for debunking on-board fire points based on ground feature flammability constraints, including: Obtain a set of candidate fire point pixels based on brightness temperature feature detection of remote sensing images, and determine the target area corresponding to the set of candidate fire point pixels; Obtain a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image is 1, and the pixel value of non-flammable ground feature pixels is 0. Spatial overlay analysis is performed on the candidate fire point pixel set and the binary image of ground feature flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground feature flammability; When the pixel value is 0, the corresponding candidate fire point pixel is determined to be a false fire point and removed. When the pixel value is 1, the corresponding candidate fire point pixel is determined to be a real fire point and retained.

[0008] Furthermore, the construction of the binary image of the flammability of the ground features includes: Acquire remote sensing images of the target area; The remote sensing image is analyzed based on a preset analysis model to obtain the component coverage of each pixel. The component coverage includes photosynthetic vegetation coverage, non-photosynthetic vegetation coverage and bare soil / impermeable surface coverage, and the sum of photosynthetic vegetation coverage, non-photosynthetic vegetation coverage and bare soil / impermeable surface coverage is 1. If the non-photosynthetic vegetation coverage is greater than a first preset threshold, it is determined to be a combustible feature pixel, and the pixel value of the location corresponding to the combustible feature pixel is set to 1. If the photosynthetic vegetation coverage is less than the second preset threshold and the bare soil / impermeable surface coverage is greater than the third preset threshold, the image is determined to be a non-combustible feature pixel, and the pixel value at the location corresponding to the non-combustible feature pixel is set to 0. If neither of the above two conditions is met, and the sum of the photosynthetic vegetation coverage and the non-photosynthetic vegetation coverage is greater than the fourth preset threshold, the image is determined to be a combustible feature pixel, and the pixel value of the location corresponding to the combustible feature pixel is set to 1. If none of the above three conditions are met, the pixel is determined to be a non-combustible feature pixel, and the pixel value of the corresponding position of the non-combustible feature pixel is set to 0. Among them, the areas corresponding to photosynthetic vegetation and non-photosynthetic vegetation are combustible areas, while the areas corresponding to bare soil and impermeable surfaces are non-combustible areas.

[0009] Furthermore, the analysis of the remote sensing image based on the preset analysis model includes: The remote sensing image is decomposed based on a linear spectral mixture analysis model to obtain the photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage of each pixel. The spectral fitting residual after decomposition of each pixel is calculated. The spectral fitting residual is used to characterize the degree of difference between the actual spectrum of the pixel and the linear combination result of the endmember spectrum. If the spectral fitting residual is greater than a preset residual threshold, the corresponding pixel is taken as a mixed pixel. For the mixed pixel, obtain the component coverage of the neighboring pixels of the mixed pixel; The component coverage of the mixed pixels is recalculated based on the component coverage of the neighboring pixels and the spatial interpolation method to eliminate the uncertainty of the mixed pixels in the flammability determination.

[0010] Furthermore, after generating the binary image of the ground feature's flammability based on the pixel values, the method further includes: Based on row priority, the pixels in the binary image of the flammability of the ground features are mapped to a one-dimensional index sequence using the formula k=i×W+j, where i is the row number, j is the column number, W is the width of the binary image of the flammability of the ground features, and k is the index value. The pixel values ​​of every N pixels in the one-dimensional index sequence are combined into a storage unit for storage, wherein the pixel value with index value k is stored in the first storage unit. In a storage unit, where N is a positive integer, and in the storage unit, each cell value is arranged sequentially from the least significant bit to the most significant bit.

[0011] The determination of the value of N includes: Obtain seasonal variation information and climate zoning information for the target area; The spatial distribution density of combustible ground features is determined based on the seasonal variation information and climate zoning information. The spatial distribution density is used to characterize the number of pixels with a pixel value of 1 per unit area. The value of N is determined based on the spatial distribution density of the combustible ground features. When the spatial distribution density of the combustible ground features is higher than a preset density threshold, N takes a first value. When the spatial distribution density of the combustible ground features is lower than the preset density threshold, N takes a second value, and the first value is greater than the second value.

[0012] Furthermore, the step of performing spatial overlay analysis on the candidate fire point pixel set and the binary image of ground cover flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground cover flammability includes: Obtain the geographic coordinates of each candidate fire point pixel in the candidate fire point pixel set; Based on the spatial coordinate system and pixel size of the binary image of flammability of ground features, the geographic coordinates are mapped to the pixel grid of the binary image of flammability of ground features to obtain the grid position corresponding to each candidate fire point pixel, wherein the grid position includes row number i and column number j; The index value k is calculated based on the grid location using the formula k=i×W+j, where W is the width of the binary image of the flammability of the ground features. The bit corresponding to the index value is found from the one-dimensional index sequence of the binary image of the flammability of the ground features with constant time complexity, and the pixel value of the bit is used as the pixel value of the candidate fire point pixel at the corresponding position in the binary image of the flammability of the ground features.

[0013] Furthermore, after determining and retaining the candidate fire point pixels corresponding to a pixel value of 1 as real fire points, the process also includes: Obtain the retained candidate fire point pixels and use the retained candidate fire point pixels as fire point pixels to be processed; Obtain the connected regions of combustible features in the binary image of the combustibility of the ground features. The connected regions of combustible features are composed of pixels with a pixel value of 1 and eight neighboring pixels connected. Based on the connected regions of the combustible features, the fire point pixels to be processed are clustered, and the fire point pixels to be processed located in the same connected region of combustible features are divided into the same fire point cluster, and the fire point pixels to be processed in different connected regions of combustible features belong to different fire point clusters. For each fire cluster, obtain the pixel values ​​of all fire pixels to be processed within the fire cluster in the binary image of ground feature flammability. If the number of fire point pixels with a value of 1 in the fire point cluster is less than a preset threshold, all fire point pixels in the fire point cluster will be identified as false fire points and removed.

[0014] Furthermore, after determining the corresponding candidate fire point pixels as false fire points and removing them, the process also includes: Calculate the ratio of the number of rejected false fire points to the total number of candidate fire point pixels in the candidate fire point pixel set, and use the ratio as the false fire point rejection rate. If the false fire point rejection rate exceeds a preset rejection threshold, it is determined that the flammability classification result of the binary image of ground features deviates from the actual surface conditions of the target area, triggering the process of reconstructing the binary image of ground features in the target area.

[0015] On the other hand, embodiments of this application provide a satellite fire point despoofing device based on ground object flammability constraints, the device comprising: The first acquisition module is used to acquire a set of candidate fire point pixels obtained based on brightness temperature feature detection of remote sensing images, and to determine the target area corresponding to the set of candidate fire point pixels. The second acquisition module is used to acquire a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image of flammable ground features is 1, and the pixel value of non-flammable ground feature pixels is 0. The analysis module is used to perform spatial overlay analysis on the candidate fire point pixel set and the binary image of ground cover flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground cover flammability. The determination module is used to determine the corresponding candidate fire point pixel as a false fire point and remove it when the pixel value is 0, and to determine the corresponding candidate fire point pixel as a real fire point and retain it when the pixel value is 1.

[0016] In another aspect, embodiments of this application provide an electronic device, the device comprising: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements a method for debunking on-board fire points based on ground object flammability constraints as described in the first aspect.

[0017] Compared with existing technologies, the on-board fire point despoofing method, apparatus, and device based on ground feature flammability constraints of this application embodiment can acquire the ground feature flammability binary image of the target area corresponding to the candidate fire point pixel set, and perform spatial overlay analysis on the candidate fire point pixel set and the ground feature flammability binary image. It can directly determine the authenticity of fire points in the candidate fire point pixels based on the pixel values, and realize the rapid elimination of false fire points by utilizing the essential attribute differences between flammable and non-flammable ground features. This avoids the high computational complexity and resource consumption problems caused by relying on complex threshold secondary screening, multi-temporal consistency judgment, or machine learning models in existing technologies. It makes it possible to real-time despoofing of large-scale fire points under the condition of limited on-board computing and storage resources. To a certain extent, it reduces the false fire point misjudgment rate caused by non-flammable ground features such as bare land, urban impermeable surfaces, and industrial facilities, thereby improving the accuracy of fire point detection and on-board processing efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings. Figure 1 This is a flowchart illustrating a method for identifying false launch points on a satellite based on the flammability constraints of ground features in this embodiment. Figure 2 This is a structural block diagram illustrating a satellite fire point defacement device based on ground object flammability constraints in this embodiment; Figure 3 This is a structural block diagram illustrating an electronic device in this embodiment. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] To facilitate understanding of the embodiments of this application, further explanation and description will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of this application. In the drawings, the dimensions and relative dimensions of components may be exaggerated for clarity and / or descriptive purposes. When exemplary embodiments can be implemented differently, a specific process sequence may be performed in a different order than that described. For example, two consecutively described processes may be performed substantially simultaneously or in the reverse order of their description. Furthermore, the same reference numerals denote the same components.

[0021] The terminology used herein is for the purpose of describing particular embodiments and is not intended to be limiting. As used herein, unless the context clearly indicates otherwise, the singular forms “a” and “the” are intended to include the plural forms as well. Furthermore, when the terms “comprising” and / or “including” and variations thereof are used in this specification, it indicates the presence of the stated features, integrals, steps, operations, parts, components, and / or groups thereof, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, parts, components, and / or groups thereof. It should also be noted that, as used herein, the terms “substantially,” “about,” and other similar terms are used as approximate terms rather than as terms of degree, thus explaining the inherent biases in measurements, calculated values, and / or provided values ​​that would be recognized by one of ordinary skill in the art.

[0022] Existing fire detection methods primarily rely on the brightness temperature characteristics of remotely sensed images. However, non-combustible features such as bare ground and impermeable surfaces in urban areas can also generate high brightness temperature responses, leading to a large number of false fire points. To suppress false fire points, existing methods generally suffer from high computational complexity, strong dependence on multi-temporal or multi-band data, complex implementation, or high resource consumption, making them unsuitable for the limited computing and storage resources available on satellites.

[0023] To address the problems in the prior art, this application provides a method, apparatus, and device for despoofing on-board fire points based on ground object flammability constraints.

[0024] The following section first introduces a method for despoofing on-board fire points based on ground object flammability constraints, provided in the embodiments of this application.

[0025] Figure 1 This diagram illustrates a flowchart of a method for debunking on-board fire points based on ground feature flammability constraints, according to an embodiment of this application. Figure 1 As shown, a method for debunking on-board fire points based on ground feature flammability constraints includes the following steps: S101. Obtain the set of candidate fire point pixels based on brightness temperature feature detection of remote sensing images, and determine the target area corresponding to the set of candidate fire point pixels.

[0026] In this embodiment, pixels with abnormally high brightness temperature characteristics in remote sensing images can be identified manually and used as candidate fire pixels. Based on the spatial distribution of the candidate fire pixels, a rectangular or polygonal region containing all candidate fire pixels can be determined as the target region. Alternatively, a preset brightness temperature threshold method can be used to determine the set of candidate fire pixels. Specific bands such as mid-infrared and thermal infrared of the remote sensing image can be analyzed, and pixels with brightness temperature values ​​exceeding the preset brightness temperature threshold can be used as candidate fire pixels. For example, a global brightness temperature threshold can be preset, and all pixels with brightness temperature values ​​exceeding the global brightness temperature threshold can be assigned to the candidate fire pixel set. After obtaining the candidate fire pixels, the target region can be delineated based on the smallest bounding rectangle or polygon of the candidate fire pixel set. The target region can also be determined by combining preset geographic administrative divisions, thereby forming a complete region containing all candidate fire pixels.

[0027] S102. Obtain a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image of ground feature flammability is 1, and the pixel value of non-flammable ground feature pixels is 0.

[0028] The construction of the binary image of ground feature combustibility includes: acquiring remote sensing images of the target area; analyzing the remote sensing images based on a preset analysis model to obtain the component coverage of each pixel, which includes photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage, and the sum of photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage is 1; if the non-photosynthetic vegetation coverage is greater than a first preset threshold, it is identified as a combustible ground feature pixel, and the pixel value of the corresponding location of the combustible ground feature pixel is set to 1; if the non-photosynthetic vegetation coverage is less than a second preset threshold, and the bare soil / impermeable surface coverage is greater than a third preset threshold... If a pixel is identified as a non-combustible feature, its corresponding pixel value is set to 0. If neither of the above two conditions is met, and the sum of photosynthetic vegetation coverage and non-photosynthetic vegetation coverage is greater than a fourth preset threshold, the pixel is identified as a combustible feature, and its corresponding pixel value is set to 1. If none of the above three conditions are met, the pixel is identified as a non-combustible feature, and its corresponding pixel value is set to 0. A binary image of the combustibility of features is generated based on the pixel values. The areas corresponding to photosynthetic and non-photosynthetic vegetation are combustible feature areas, while the areas corresponding to bare soil and impermeable surfaces are non-combustible feature areas.

[0029] In this embodiment, multispectral or hyperspectral image data of the target area can be acquired through sensors mounted on satellite remote sensing platforms such as MODIS, VIIRS, or Landsat; pre-processed remote sensing image products of the target area, such as remote sensing images that have undergone radiometric, atmospheric, and geometric corrections, can also be downloaded from remote sensing data centers or geospatial information platforms.

[0030] Among them, the preset analysis model can adopt a linear spectral mixture analysis model. The spectral response of the linear spectral mixture analysis model is a linear combination of the spectra of various land cover components within it. The coverage of each component is inverted through optimization algorithms such as the least squares method. Alternatively, a nonlinear spectral mixture model can be adopted, such as machine learning methods based on neural networks or support vector machines. The nonlinear relationship between pixel spectra and component coverage is learned through training samples, thereby performing component inversion. Component coverage obtained by inverting the NSSI-NDVI triangular feature space can also be adopted. For example, the component coverage of each pixel in the target area can be obtained by inverting remote sensing image data: non-photosynthetic vegetation, photosynthetic vegetation, and bare soil / impermeable surface. Among them, the areas corresponding to photosynthetic vegetation and non-photosynthetic vegetation are combustible land cover areas, and the areas corresponding to bare soil and impermeable surface are non-combustible land cover areas.

[0031] The sum of the proportions of the three components of each pixel satisfies: ; Among them, f NPV f represents the percentage of non-photosynthetic vegetation cover. PV f represents the percentage of photosynthetic vegetation cover. BS This represents the percentage of bare soil / impermeable surface coverage.

[0032] The first preset threshold is used to determine whether the non-photosynthetic vegetation cover reaches a combustible level. In this embodiment, the first preset threshold can be 0.25, and can also be adjusted within the range of 0.15 to 0.40 according to the remote sensing image resolution and land cover characteristics. The second preset threshold is used to identify the upper bound where the non-photosynthetic vegetation cover is negligible. The second preset threshold can also be determined by the 99% confidence upper limit of typical non-combustible land cover samples or the 1st percentile of the entire area. In this embodiment, the value range of the second preset threshold can be 0.01 to 0.05. The three preset thresholds are used to determine whether the coverage of bare soil / impermeable surface has reached the level of non-combustible dominance. In this embodiment, the third preset threshold can be 0.60. The third preset threshold can also be adjusted within the range of 0.45 to 0.80 according to the remote sensing image resolution and surface cover characteristics. The fourth preset threshold is used to determine whether the total vegetation coverage has reached the level of combustibility. It can be dynamically adjusted by the 50th percentile of historical fire point samples or the 70th percentile of the vegetation coverage distribution in the whole area. In this embodiment, the value range of the fourth preset threshold can be 0.35 to 0.60.

[0033] In this embodiment, image processing software assigns values ​​to each pixel based on its determined pixel value and saves it as a standard raster image format.

[0034] In this embodiment, remote sensing images are analyzed using a preset analysis model to obtain the photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage of each pixel. The sum of these three values ​​is 1, ensuring the scientific validity and completeness of the decomposition of ground features. When the non-photosynthetic vegetation coverage is greater than a first preset threshold, the pixel is identified as a combustible ground feature. When the non-photosynthetic vegetation coverage is less than a second preset threshold and the bare soil / impermeable surface coverage is greater than a third preset threshold, the pixel is identified as a non-combustible ground feature. When the sum of the photosynthetic vegetation coverage and the non-photosynthetic vegetation coverage is greater than a fourth preset threshold, the pixel is identified as a combustible ground feature. This approach comprehensively identifies different types of combustible and non-combustible ground features, thereby not only identifying combustible areas with high vegetation coverage but also effectively excluding non-combustible areas such as bare soil and impermeable surfaces that are prone to brightness and temperature anomalies. At the same time, the combustible characteristics of non-photosynthetic vegetation are taken into account, avoiding missed detections caused by insufficient coverage of a single vegetation type.

[0035] The analysis of remote sensing images based on a preset analysis model includes: decomposing the remote sensing images based on a linear spectral mixture analysis model to obtain the photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage of each pixel, and calculating the spectral fitting residual after decomposition of each pixel. The spectral fitting residual is used to characterize the degree of difference between the actual spectrum of the pixel and the linear combination result of the endmember spectrum. If the spectral fitting residual is greater than a preset residual threshold, the corresponding pixel is regarded as a mixed pixel. For mixed pixels, the component coverage of the neighboring pixels of the mixed pixel is obtained. The component coverage of the mixed pixel is recalculated based on the component coverage of the neighboring pixels and the spatial interpolation method to eliminate the uncertainty of mixed pixels in the flammability determination.

[0036] In this embodiment, the preset analysis model adopts the linear spectral mixture analysis model. The linear spectral mixture analysis model is a remote sensing image processing technology. Its basic assumption is that the spectral signal of each pixel received by the sensor is a linear combination of the spectral signals of the various land cover components it contains. The linear spectral mixture analysis model decomposes the mixed spectrum of a pixel into different land cover components, such as photosynthetic vegetation, non-photosynthetic vegetation, and bare soil / impermeable surface, and obtains the proportion of the components in the pixel, i.e., the coverage.

[0037] The difference between the actual observed spectrum of the pixel quantized by the spectral fitting residual and the synthesized spectrum obtained by weighting and summing the endmember spectra obtained by decomposition through the linear spectral mixture model according to their coverage is considered. The magnitude of the spectral fitting residual reflects the accuracy and reliability of the decomposition result of the linear spectral mixture model. The larger the residual, the greater the deviation between the decomposition result and the actual situation, indicating the existence of nonlinear mixing effects or unconsidered endmembers. In this embodiment, the coverage of each component is solved and the residual is calculated using the least squares method or the constrained least squares method. The constrained least squares method obtains a reasonable decomposition result by adding the constraint that the coverage is non-negative and the sum of the coverage is 1.

[0038] In this embodiment, a preset residual threshold is used to determine the reliability of the spectral decomposition results. When the spectral fitting residual of a pixel is greater than the preset residual threshold, it indicates that the decomposition result of that pixel has a large uncertainty or error, and its component coverage value cannot accurately reflect the actual ground cover composition. At this time, such pixels are marked as mixed pixels, and the initial decomposition results of mixed pixels need to be further corrected or verified to avoid making flammability determination based on inaccurate component coverage. The preset residual threshold can be determined based on historical data or expert experience. For example, when the residual exceeds a certain percentage, such as 5% or 10%, it is determined to be a mixed pixel. Alternatively, statistical analysis can be performed on the residuals of the entire target area. For example, the mean and standard deviation of the residuals can be calculated, and pixels with residuals greater than the mean plus twice the standard deviation are determined to be mixed pixels.

[0039] In this context, neighboring pixels are spatially adjacent pixels surrounding the target mixed pixel. Since the distribution of ground features usually has spatial continuity, the composition of a pixel is often correlated with its neighboring pixels. Obtaining the component coverage of neighboring pixels provides reference information for correcting the component coverage of mixed pixels. In this embodiment, neighboring pixels are determined by selecting a fixed-size window centered on the mixed pixel, such as 3×3 or 5×5 pixels, and obtaining the component coverage of all non-mixed pixels within this window. The size or shape of the neighboring window can also be adjusted according to the complexity or spatial heterogeneity of the ground feature distribution around the mixed pixel. For example, when the neighboring pixel is a mixed pixel, the range of the neighboring window can be expanded or a more robust spatial interpolation method can be used to obtain the most representative component coverage of the neighboring pixels.

[0040] Spatial interpolation is a method for estimating mixed pixel data using data from neighboring pixels. It corrects or re-estimates the uncertain component coverage of mixed pixels by combining the component coverage information of neighboring pixels around the mixed pixel. Spatial interpolation methods can include inverse distance weighted interpolation, kriging interpolation, nearest neighbor interpolation, or bilinear interpolation. Inverse distance weighted interpolation assigns different weights to neighboring pixels and mixed pixels based on their distances, with closer pixels receiving greater weights. Kriging interpolation considers spatial autocorrelation and determines the weights through a variogram model, thereby providing the optimal unbiased estimate.

[0041] In this embodiment, the spectral fitting residual is used as a quantitative indicator of the reliability of the decomposition result, thereby identifying the pixels whose initial component coverage is inaccurate due to the spectral mixing effect. For mixed pixels, the component coverage information of its neighboring pixels is obtained by utilizing the principle of spatial continuity of ground feature distribution. The component coverage of the mixed pixels is recalculated using spatial interpolation methods, thereby correcting the uncertainty in the original decomposition value and making the component coverage closer to the actual ground feature composition, which improves the accuracy of the binary image construction of ground feature flammability to a certain extent.

[0042] After generating a binary image of ground cover flammability based on pixel values, the process further includes: mapping the pixels in the binary image of ground cover flammability to a one-dimensional index sequence using the formula k=i×W+j based on row priority, where i is the row number, j is the column number, W is the width of the binary image of ground cover flammability, and k is the index value; combining the pixel values ​​of every N pixels in the one-dimensional index sequence into a storage unit for storage, wherein the pixel value with index value k is stored in the first... In a storage unit, where N is a positive integer, and each cell value is arranged sequentially from the least significant bit to the most significant bit.

[0043] In a two-dimensional image, each cell is uniquely identified by row number i and column number j. By adopting a row-first traversal method, that is, visiting cells sequentially from left to right and from top to bottom, and combining the image width, any two-dimensional coordinate (i, j) is mapped to a one-dimensional index value. In this embodiment, the calculated index value is used as the subscript of the one-dimensional array to store or retrieve cell values. Specific memory addressing logic can also be designed to calculate the corresponding one-dimensional physical address in real time based on the input two-dimensional coordinates and image width, thereby realizing direct access to the storage unit.

[0044] A binary image has a pixel value of 0 or 1, and each pixel only needs 1 bit to represent it. Computers usually use bytes as the smallest storage unit. When allocating one byte for each pixel, it will cause a waste of storage space. In this embodiment, the 1-bit value of 8 pixels is stored in one byte, which can reduce the actual storage space occupied. One storage unit includes M bytes.

[0045] The method for determining the value of N includes: acquiring seasonal variation information and climate zoning information of the target area; determining the spatial distribution density of combustible ground features based on the seasonal variation information and climate zoning information, wherein the spatial distribution density is used to characterize the number of pixels with a pixel value of 1 per unit area; determining the value of N based on the spatial distribution density of combustible ground features, wherein when the spatial distribution density of combustible ground features is higher than a preset density threshold, N takes a first value, and when the spatial distribution density of combustible ground features is lower than the preset density threshold, N takes a second value, and the first value is greater than the second value.

[0046] In this embodiment, seasonal variation information and climate zoning information of the target area can be obtained in various ways. For example, seasonal variation characteristics of the target area, such as vegetation growth cycle and precipitation patterns, can be extracted by combining periodic observation data from spaceborne sensors with timestamp information. Climate zoning map data at the global or regional scale can be used to obtain information such as the climate zone and climate type to which the target area belongs. Alternatively, historical meteorological data, phenological data and existing climate zoning vector data of the target area can be queried through a geographic information system database to obtain the required seasonal variation information and climate zoning information.

[0047] In this embodiment, by combining seasonal variation information and climate zoning information, a trained machine learning model, such as support vector machine or random forest, is used to predict the distribution of combustible ground features in the target area. The proportion of pixels with a value of 1 in the prediction results is then counted to obtain the spatial distribution density of combustible ground features. The seasonal variation information includes, but is not limited to, the vegetation index, precipitation, and climate zoning information of the current season. The climate zoning information includes, but is not limited to, the influence of climate type on vegetation type. Alternatively, the spatial distribution density of combustible ground features in the current target area can be estimated through statistical analysis based on the seasonal variation information, climate zoning information, and historical remote sensing image data. For example, a lookup table or empirical formula can be established based on the typical ground feature cover types and their combustibility in different seasons and climate zones. By inputting the seasonal and climate information, the corresponding spatial distribution density of combustible ground features can be output.

[0048] In this embodiment, when the spatial distribution density is higher than the preset density threshold, N takes a first value; when the spatial distribution density is lower than the preset density threshold, N takes a second value, and the first value is greater than the second value. For example, when the spatial distribution density is 0.7 and the preset density threshold is 0.5, since 0.7 is higher than 0.5, the larger first value is selected as the value of N. For example, N=64, indicating that the pixel values ​​of every 64 pixels are stored in one storage unit. When the spatial feature distribution density is 0.3, since 0.3 is lower than 0.5, the smaller second value is selected as the value of N. For example, N=32, indicating that the pixel values ​​of every 32 pixels are stored in one storage unit.

[0049] In this embodiment, the number N of pixel values ​​in the storage unit is adjusted according to the dynamic changes in the spatial distribution density of combustible ground features in the target area. In areas with high combustible ground feature density, a larger N value is used to reduce the total number of storage units, reduce the addressing overhead during data access, and thus improve data reading efficiency and avoid affecting data access speed due to insufficient storage units. In areas with low combustible ground feature density, a smaller N value is used to avoid redundancy of storage units and thus save on-board storage resources.

[0050] In this embodiment, since every N pixels are combined and stored in one unit, the index of the storage unit where the pixel with index value k is located is determined by dividing k by N and rounding down. Given the pixel index k, the address of its corresponding storage unit is calculated, thereby achieving efficient random access. In this embodiment, the address offset is calculated by arithmetic logic units or bit shift operations.

[0051] The order from least significant bit to most significant bit indicates that the first cell value is stored in the least significant bit of the storage unit, the second cell value is stored in the second least significant bit, and so on, until the Nth cell value is stored in the most significant bit.

[0052] In this embodiment, the two-dimensional binary image of ground feature flammability is converted into a one-dimensional bitmap storage structure, reducing the storage space requirement of the binary image of ground feature flammability. By calculating the one-dimensional index and bit operations, the efficiency of locating and accessing the pixel value of any pixel is improved, avoiding the additional overhead of traditional two-dimensional array access. While ensuring data integrity, the storage space is minimized and the access performance is maximized, making it easier to adapt to the harsh conditions of limited computing and storage resources on the satellite.

[0053] S103. Perform spatial overlay analysis on the candidate fire point pixel set and the binary image of ground feature flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground feature flammability.

[0054] Specifically, the geographic coordinates of each candidate fire point pixel in the candidate fire point pixel set are obtained; the geographic coordinates are mapped to the pixel grid of the binary image of flammability based on the spatial coordinate system and pixel size of the binary image of flammability, to obtain the grid position corresponding to each candidate fire point pixel, where the grid position includes row number i and column number j; the index value k is calculated based on the grid position using the formula k=i×W+j, where W is the width of the binary image of flammability; the bit corresponding to the index value is found from the one-dimensional index sequence of the binary image of flammability with constant time complexity, and the pixel value of the bit is used as the pixel value of the candidate fire point pixel at the corresponding position in the binary image of flammability.

[0055] In this embodiment, geographic coordinates can be calculated by combining the metadata of remote sensing images with the pixel row and column numbers, or they can be obtained by the geographic coordinates contained in the pre-stored fire detection results. During on-board processing, the pixel positions in the sensor coordinate system can also be converted into geographic coordinates by using the satellite's attitude, orbital parameters, and sensor parameters.

[0056] In this embodiment, the binary image of ground feature flammability has a specific spatial coordinate system and pixel size. The pixel size represents the actual ground distance of each pixel. The spatial coordinate system includes, but is not limited to, UTM and WGS84 projection. In this embodiment, the geographic coordinates are converted into image coordinates, i.e., row and column numbers, through an affine transformation matrix or a polynomial transformation model. The coordinate transformation function provided by the geographic information system library is also used to project the geographic coordinates onto the projection coordinate system of the image. Then, the row and column numbers are calculated based on the pixel size and origin position of the image. During on-board processing, the geographic reference parameters of the binary image of ground feature flammability are pre-stored, and the geographic coordinates are converted into image grid positions through simple linear calculation.

[0057] In this embodiment, the constant time complexity is O(1) time complexity. O(1) time complexity means that the execution time of the lookup operation is constant and does not change with the increase of data size. For example, the pixel values ​​of every N pixels are combined into a storage unit. The storage unit storing the pixel value is directly located by the index value k obtained in advance. For example, the index value of the storage unit is obtained by dividing k by N. Then, bit operations are performed, such as taking the modulo of k with N to obtain the bit offset in the storage unit. Then, shift and bit AND operations are performed to extract the corresponding bit value.

[0058] The spatial index and memory address can be mapped using the following formula:

[0059] Where (r, c) represents the pixel position of the candidate fire point in the remote sensing image, r is the row number, and c is the column number; W represents the total number of pixels in a single row of the preset ground feature flammability background map, i.e., the image width; Base_Addr is the starting base address of the single-bit ground feature flammability background map in the on-board memory. The offset is rounded down to the nearest integer and used to determine the byte offset to which the cell belongs. N is the number of cells contained in each storage unit.

[0060] In this embodiment, by mapping geographic coordinates to an image grid and using a row-first one-dimensional index sequence to store binary images of ground feature flammability, the complex two-dimensional spatial query is transformed into a one-dimensional index lookup. Pixel values ​​are directly obtained from the one-dimensional index sequence with constant time complexity, avoiding the time-consuming coordinate transformation, complex interpolation calculations, and iterative search processes in traditional methods. This reduces computational load and memory access overhead, improving the efficiency and real-time performance of spatial overlay analysis to a certain extent. It also enables the rapid completion of despoofing processing of a large number of candidate fire point pixels under the limited computing and storage resources on the satellite.

[0061] S104. When the pixel value is 0, the corresponding candidate fire point pixel is determined to be a false fire point and removed. When the pixel value is 1, the corresponding candidate fire point pixel is determined to be a real fire point and retained.

[0062] In this embodiment, the flammability of ground features can be determined in the following way: ; in, This represents an array of ground feature flammability background maps stored in bytes; >> represents the logical right shift operator, used to shift the target bit to the least significant bit; (modN) represents the modulo operation, used to determine the bit offset of the pixel within the corresponding byte; & represents the bitwise AND operator, used to mask other bits and extract the final binary result; (r, c) represents the pixel position of the candidate fire point in the remote sensing image, where r is the row number and c is the column number; W represents the total number of pixels per row in the preset ground feature flammability background map, i.e., the image width; N is the number of pixels contained in each storage unit; if If the location has sufficient flammable material, it is determined to be a candidate for a true ignition point; if If the location is determined to be a non-combustible feature, it is identified as a false fire point and removed.

[0063] In this embodiment, all pixels identified as real fire points can be recorded in a list, and all pixels identified as false fire points can be removed from the original list.

[0064] After determining and retaining candidate fire points corresponding to a pixel value of 1 as real fire points, the process further includes: acquiring the retained candidate fire point pixels and using them as fire point pixels to be processed; acquiring the connected regions of combustible land cover in the binary image of land cover flammability, wherein the connected regions of combustible land cover are composed of pixels with a pixel value of 1 and eight neighboring connected pixels; clustering the fire point pixels to be processed based on the connected regions of combustible land cover, dividing the fire point pixels to be processed located in the same connected region of combustible land cover into the same fire point cluster, and dividing the fire point pixels to be processed in different connected regions of combustible land cover into different fire point clusters; for each fire point cluster, acquiring the pixel values ​​of all fire point pixels to be processed in the binary image of land cover flammability within the fire point cluster; if the number of fire point pixels to be processed with a pixel value of 1 in the fire point cluster is less than a preset threshold, all fire point pixels to be processed in the fire point cluster are determined as false fire points and removed.

[0065] In this embodiment, after the initial determination is completed, the row number, column number or unique identifier of all the pixels marked as real fire points are added to a list to form a set of fire point pixels to be processed; a status flag can also be set for each pixel, and the status of the pixels initially determined to be real fire points is updated to pending processing, and only the pixels with the pending processing status are processed subsequently.

[0066] In this context, the connected regions of combustible features are sets of pixels with a value of 1 that are spatially connected by an eight-neighbor rule. In this embodiment, a connected component labeling algorithm from the field of image processing is used to determine the connected regions of combustible features. For example, when traversing the binary image of combustibility, if a pixel with a value of 1 is encountered that has not yet been labeled, starting from that pixel, a breadth-first search or depth-first search algorithm is used to recursively search for all connected pixels with a value of 1 in its eight neighborhoods, and assign them the same connected region identifier. This process is repeated until all pixels with a value of 1 are classified into the corresponding connected regions.

[0067] In this embodiment, fire point pixels located within the same combustible land cover connected region are grouped into a fire point cluster to aggregate fire point pixels belonging to the same real combustion event and distinguish independent combustion events located in different combustible land cover regions. The clustering method based on spatial connectivity can reflect the actual spatial distribution characteristics of fire points. In this embodiment, for each fire point pixel to be processed, the combustible land cover connected region identifier corresponding to its position in the binary image of combustible land cover is queried. All sets of fire point pixels to be processed with the same connected region identifier are divided into a fire point cluster. For example, a hash table or dictionary is constructed, with the connected region identifier as the key and the list of all fire point pixels to be processed in that region as the value.

[0068] In this embodiment, each formed fire cluster is traversed. For each fire cell to be processed in the cluster, its corresponding cell value is read from the pre-acquired binary image of ground feature flammability according to its spatial coordinates. The cell value is stored in the data structure of each fire cluster, for example, as an attribute of the cell list or a separate list.

[0069] If the number of unprocessed fire point pixels with a value of 1 within a fire point cluster is less than a preset threshold, all unprocessed fire point pixels within the fire point cluster are identified as false fire points and removed. This is to eliminate false fire points caused by isolated noise points or extremely small-scale, meaningless combustion events. By setting a preset threshold, it is ensured that only sufficiently large and stable fire point clusters are considered real fire points. For example, a fire point cluster consisting of a single or a few pixels, even if its pixel value is 1, may only be a false alarm caused by sensor noise or edge effects in ground cover classification. In this embodiment, after removing pixels with a value of 0 within the fire point cluster, the number of unprocessed fire point pixels with a value of 1 remaining in the fire point cluster is counted. When this number is less than a preset threshold, all pixels in the fire point cluster, including previously unremoved pixels with a value of 1, are identified as false fire points and removed from the final result. The preset threshold can be determined according to the remote sensing image resolution, fire point characteristics, and application requirements. For example, the preset threshold can be 2, 3, or 5.

[0070] In this embodiment, the initially retained candidate fire point pixels are used as the objects to be processed. The connected regions of combustible ground features are identified, and fire point pixels in the same connected region are divided into the same fire point cluster to ensure that the fire point clustering is consistent with the actual distribution of combustible ground features. Then, each fire point cluster is subjected to double screening: pixels with a value of 0 in the cluster are removed to correct false fire points caused by classification noise; when the number of pixels with a value of 1 in the fire point cluster is lower than a preset threshold, the entire fire cluster is judged as a false fire point and removed to eliminate isolated noise or minimal combustion events, thereby reducing false alarms to a certain extent and improving the reliability of fire point detection.

[0071] After identifying and removing the corresponding candidate fire point pixels as false fire points, the process also includes: calculating the ratio of the number of removed false fire points to the total number of candidate fire point pixels in the candidate fire point pixel set, and using the ratio as the false fire point removal rate; if the false fire point removal rate exceeds the preset removal threshold, it is determined that the flammability classification result of the ground feature flammability binary image deviates from the actual surface conditions of the target area, triggering the process of reconstructing the ground feature flammability binary image of the target area.

[0072] In this embodiment, when performing the false fire point removal operation, the number of false fire point pixels removed is counted in real time, and the initial total number of candidate fire point pixels is recorded. The false fire point removal rate is obtained by division. Alternatively, after the entire false fire point removal process is completed, the pixels that are judged as false fire points and real fire points can be distinguished by traversing the set of processed candidate fire point pixels, and then the false fire point removal rate can be calculated.

[0073] In this embodiment, when the false fire point rejection rate abnormally increases or decreases, it indicates that the binary image of ground feature flammability reflects the actual surface flammability in a biased manner. For example, when the rejection rate is too high, it indicates that the binary image of ground feature flammability incorrectly marks a large number of actual flammable areas as non-flammable, resulting in the false rejection of real fire points. When the rejection rate is too low, it indicates that the binary image of ground feature flammability fails to effectively identify non-flammable areas, resulting in the insufficient rejection of false fire points. In this embodiment, the rejection rate is directly compared with a preset rejection threshold. Once the threshold is exceeded, a bias judgment is triggered. Alternatively, statistical methods can be used. For example, a threshold range can be set. When the false fire point rejection rate falls outside this range, a bias is determined. The threshold range can be set as needed and is not specifically limited here.

[0074] When the flammability classification result of the binary image of flammable ground features deviates from the actual surface conditions of the target area, the process of reconstructing the binary image of flammable ground features in the target area is triggered, thereby ensuring the dynamic updating and correction of the binary image of flammable ground features. In this embodiment, an instruction is sent to the module responsible for generating the binary image of flammable ground features, requesting it to reacquire the latest remote sensing image data, and to re-perform the ground feature component coverage analysis and flammability binarization processing based on the updated remote sensing image data or the optimized model.

[0075] The preset rejection threshold can be determined based on historical false fire detection statistics. For example, by analyzing the average false fire rejection rate and its fluctuation range in the target area under different seasons and surface conditions over a period of time, values ​​exceeding the normal fluctuation range can be set as the preset rejection threshold. Alternatively, the threshold can be determined based on the accuracy of land cover classification. For example, if the accuracy of the land cover classification model used to generate binary images of land cover flammability is known, such as overall accuracy, kappa coefficient, or producer accuracy and user accuracy for each category, a reasonable false fire rejection rate threshold can be set based on the above indicators and the tolerance for false fire rejection effects in practical applications.

[0076] In this embodiment, when the false fire point removal rate is abnormal, the reconstruction process of the binary image of ground object flammability is triggered, thereby effectively correcting the problem of false fire point misjudgment caused by the inaccuracy of the binary image of ground object flammability. This enables the binary image of ground object flammability to maintain consistency with the actual ground surface conditions, thereby improving the accuracy of the on-board fire point despoofing method to a certain extent. Especially when the ground environment is complex and changeable or the quality of remote sensing images is unstable, it effectively avoids the failure of despoofing caused by outdated or erroneous binary images of ground object flammability, ensuring the reliability of fire point monitoring.

[0077] Based on the above embodiments, this application also provides a specific implementation of a satellite fire point despoofing device based on ground object flammability constraints. Please refer to the following embodiments.

[0078] First see Figure 2 The on-board fire point debunking device 200 based on ground object flammability constraints provided in this application embodiment may include: The first acquisition module 201 is used to acquire a set of candidate fire point pixels obtained based on the brightness temperature feature detection of remote sensing images, and to determine the target area corresponding to the set of candidate fire point pixels. The second acquisition module 202 is used to acquire a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image of flammable ground features is 1, and the pixel value of non-flammable ground feature pixels is 0. Analysis module 203 is used to perform spatial overlay analysis on the candidate fire point pixel set and the binary image of ground feature flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground feature flammability; The determination module 204 is used to determine the corresponding candidate fire point pixel as a false fire point and remove it when the pixel value is 0, and to determine the corresponding candidate fire point pixel as a real fire point and retain it when the pixel value is 1.

[0079] As an optional implementation of this embodiment, the on-board fire point debunking device 200 based on ground object flammability constraints further includes: The image acquisition module is used to acquire remote sensing images of the target area; The analysis module is used to analyze remote sensing images based on a preset analysis model to obtain the component coverage of each pixel. The component coverage includes photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage, and the sum of photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage is 1. The first threshold determination module is used to determine a combustible ground feature pixel when the non-photosynthetic vegetation coverage is greater than the first preset threshold, and set the pixel value of the location corresponding to the combustible ground feature pixel to 1. The second threshold determination module is used to determine a non-combustible ground feature pixel when the non-photosynthetic vegetation coverage is less than the second preset threshold and the bare soil / impermeable surface coverage is greater than the third preset threshold, and set the pixel value of the location corresponding to the non-combustible ground feature pixel to 0. The third threshold determination module is used to determine a flammable feature pixel when neither of the above two conditions is met and the sum of photosynthetic vegetation coverage and non-photosynthetic vegetation coverage is greater than the fourth preset threshold, and to set the pixel value of the location corresponding to the flammable feature pixel to 1; when neither of the above three conditions is met, it is determined to be a non-flammable feature pixel, and to set the pixel value of the location corresponding to the non-flammable feature pixel to 0. The generation module is used to generate a binary image of ground feature flammability based on pixel values; wherein, the areas corresponding to photosynthetic vegetation and non-photosynthetic vegetation are flammable ground feature areas, and the areas corresponding to bare soil and impermeable surfaces are non-flammable ground feature areas.

[0080] As an optional implementation of this embodiment, the module is specifically used for: The remote sensing image is decomposed based on a linear spectral fusion analysis model to obtain the photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage for each pixel. The spectral fitting residual after decomposition of each pixel is calculated, which is used to characterize the degree of difference between the actual spectrum of the pixel and the linear combination result of the endmember spectrum. If the spectral fitting residual is greater than a preset residual threshold, the corresponding pixel is regarded as a fusion pixel. For fusion pixels, the component coverage of the neighboring pixels is obtained. The component coverage of the fusion pixels is recalculated based on the component coverage of the neighboring pixels and the spatial interpolation method to eliminate the uncertainty of fusion pixels in flammability determination.

[0081] As an optional implementation of this embodiment, the on-board fire point debunking device 200 based on ground object flammability constraints further includes: The mapping module is used to map the pixels in the binary image of flammable land cover into a one-dimensional index sequence based on row priority using the formula k=i×W+j after generating the binary image of flammable land cover based on the pixel value. Here, i is the row number, j is the column number, W is the width of the binary image of flammable land cover, and k is the index value. The combination module is used to combine the pixel values ​​of every N pixels in the one-dimensional index sequence into a storage unit for storage, wherein the pixel value with index value k is stored in the first storage unit. In a storage unit, where N is a positive integer, and each cell value is arranged sequentially from the least significant bit to the most significant bit.

[0082] As an optional implementation of this embodiment, the on-board fire point debunking device 200 based on ground object flammability constraints further includes: The information acquisition module is used to acquire seasonal variation information and climate zoning information for the target area. The first determination module is used to determine the spatial distribution density of combustible ground features based on seasonal variation information and climate zoning information. The spatial distribution density is used to characterize the number of pixels with a pixel value of 1 per unit area. The second determining module is used to determine the value of N based on the spatial distribution density of combustible ground features. When the spatial distribution density of combustible ground features is higher than a preset density threshold, N takes a first value. When the spatial distribution density of combustible ground features is lower than the preset density threshold, N takes a second value, and the first value is greater than the second value.

[0083] As an optional implementation of this embodiment, the analysis module 203 is specifically used for: Obtain the geographic coordinates of each candidate fire point pixel in the candidate fire point pixel set; map the geographic coordinates to the pixel grid of the binary image of flammability based on the spatial coordinate system and pixel size of the binary image of flammability, to obtain the grid position corresponding to each candidate fire point pixel, where the grid position includes row number i and column number j; calculate the index value k based on the grid position using the formula k=i×W+j, where W is the width of the binary image of flammability; find the bit corresponding to the index value from the one-dimensional index sequence of the binary image of flammability with constant time complexity, and use the pixel value of the bit as the pixel value of the candidate fire point pixel at the corresponding position in the binary image of flammability.

[0084] As an optional implementation of this embodiment, the on-board fire point debunking device 200 based on ground object flammability constraints further includes: The pixel acquisition module is used to acquire the retained candidate fire point pixels after determining the candidate fire point pixels with a pixel value of 1 as real fire points and retaining them, and to use the retained candidate fire point pixels as fire point pixels to be processed. The region acquisition module is used to acquire the connected regions of combustible features in the binary image of combustibility. The connected regions of combustible features are composed of pixels with a value of 1 and eight neighboring pixels connected. The clustering module is used to cluster the fire point pixels to be processed based on the connected areas of combustible features. It divides the fire point pixels to be processed located in the same connected area of ​​combustible features into the same fire point cluster, and divides the fire point pixels to be processed in different connected areas of combustible features into different fire point clusters. The pixel value acquisition module is used to acquire the corresponding pixel values ​​of all fire point pixels to be processed in the binary image of ground cover flammability for each fire point cluster. The rejection module is used to identify all fire pixels in the fire cluster as false fires and reject them when the number of fire pixels with a value of 1 in the fire cluster is less than a preset threshold.

[0085] As an optional implementation of this embodiment, the on-board fire point debunking device 200 based on ground object flammability constraints further includes: The calculation module is used to calculate the ratio of the number of eliminated false fire points to the total number of candidate fire point pixels in the candidate fire point pixel set after the corresponding candidate fire point pixels are identified as false fire points and eliminated. The ratio is used as the false fire point elimination rate. The determination and triggering module is used to determine that the flammability classification result of the binary image of ground features deviates from the actual surface conditions of the target area when the false fire point rejection rate exceeds the preset rejection threshold, and to trigger the process of reconstructing the binary image of ground features in the target area.

[0086] Figure 3 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0087] It may include a processor 301 and a memory 302 storing computer program instructions.

[0088] Specifically, the processor 301 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0089] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. In one instance, memory 302 may include removable or non-removable (or fixed) media, or memory 302 may be non-volatile solid-state memory. Memory 302 may be internal or external to the integrated gateway disaster recovery device.

[0090] In one instance, memory 302 may be read-only memory (ROM). In one instance, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0091] Memory 302 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform operations described with reference to a method for despoofing on-board fire points based on ground flammability constraints according to a first aspect of this disclosure.

[0092] The processor 301 reads and executes computer program instructions stored in the memory 302 to achieve... Figure 1 The illustrated embodiment presents a method for debunking on-board fire points based on ground feature flammability constraints.

[0093] In one example, the electronic device may also include a communication interface 303 and a bus 304. For example, Figure 3 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 304 and complete communication with each other.

[0094] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0095] Bus 304 includes hardware, software, or both, that couples components of an electronic device together. For example, and not as a limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 304 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0096] This electronic device can execute a method for debunking on-board fire points based on ground object flammability constraints, as described in this application embodiment, thereby achieving a combination of... Figures 1-2 This paper describes a method and apparatus for despoofing on-board fire points based on the flammability constraints of ground features.

[0097] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for debunking on-board fire points based on ground object flammability constraints, characterized in that, include: Obtain a set of candidate fire point pixels based on brightness temperature feature detection of remote sensing images, and determine the target area corresponding to the set of candidate fire point pixels; Obtain a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image is 1, and the pixel value of non-flammable ground feature pixels is 0. Spatial overlay analysis is performed on the candidate fire point pixel set and the binary image of ground feature flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground feature flammability; When the pixel value is 0, the corresponding candidate fire point pixel is determined to be a false fire point and removed. When the pixel value is 1, the corresponding candidate fire point pixel is determined to be a real fire point and retained.

2. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 1, characterized in that, The construction of the binary image of the flammability of ground features includes: Acquire remote sensing images of the target area; The remote sensing image is analyzed based on a preset analysis model to obtain the component coverage of each pixel. The component coverage includes photosynthetic vegetation coverage, non-photosynthetic vegetation coverage and bare soil / impermeable surface coverage, and the sum of photosynthetic vegetation coverage, non-photosynthetic vegetation coverage and bare soil / impermeable surface coverage is 1. If the non-photosynthetic vegetation coverage is greater than a first preset threshold, it is determined to be a combustible feature pixel, and the pixel value of the location corresponding to the combustible feature pixel is set to 1. If the non-photosynthetic vegetation coverage is less than the second preset threshold and the bare soil / impermeable surface coverage is greater than the third preset threshold, the pixel is determined to be a non-combustible feature pixel, and the pixel value of the location corresponding to the non-combustible feature pixel is set to 0. If neither of the above two conditions is met, and the sum of the photosynthetic vegetation coverage and the non-photosynthetic vegetation coverage is greater than the fourth preset threshold, the image is determined to be a combustible feature pixel, and the pixel value of the location corresponding to the combustible feature pixel is set to 1. If none of the above three conditions are met, the pixel is determined to be a non-combustible feature pixel, and the pixel value of the corresponding position of the non-combustible feature pixel is set to 0. Generate a binary image of the flammability of the ground features based on the pixel values; Among them, the areas corresponding to photosynthetic vegetation and non-photosynthetic vegetation are combustible areas, while the areas corresponding to bare soil and impermeable surfaces are non-combustible areas.

3. The method for debunking on-board fire points based on ground object flammability constraints according to claim 2, characterized in that, The analysis of the remote sensing image based on the preset analysis model includes: The remote sensing image is decomposed based on a linear spectral mixture analysis model to obtain the photosynthetic vegetation coverage, non-photosynthetic vegetation coverage, and bare soil / impermeable surface coverage of each pixel. The spectral fitting residual after decomposition of each pixel is calculated. The spectral fitting residual is used to characterize the degree of difference between the actual spectrum of the pixel and the linear combination result of the endmember spectrum. If the spectral fitting residual is greater than a preset residual threshold, the corresponding pixel is taken as a mixed pixel. For the mixed pixel, obtain the component coverage of the neighboring pixels of the mixed pixel; The component coverage of the mixed pixels is recalculated based on the component coverage of the neighboring pixels and the spatial interpolation method to eliminate the uncertainty of the mixed pixels in the flammability determination.

4. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 2, characterized in that, After generating the binary image of the ground feature flammability based on the pixel values, the method further includes: Based on row priority, the pixels in the binary image of the flammability of the ground features are mapped to a one-dimensional index sequence using the formula k=i×W+j, where i is the row number, j is the column number, W is the width of the binary image of the flammability of the ground features, and k is the index value. The pixel values ​​of every N pixels in the one-dimensional index sequence are combined into a storage unit for storage, wherein the pixel value with index value k is stored in the first storage unit. In each storage cell, N is a positive integer, and in each storage cell, the pixel value is arranged sequentially from the least significant bit to the most significant bit.

5. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 4, characterized in that, The determination of the value of N includes: Obtain seasonal variation information and climate zoning information for the target area; The spatial distribution density of combustible ground features is determined based on the seasonal variation information and climate zoning information. The spatial distribution density is used to characterize the number of pixels with a pixel value of 1 per unit area. The value of N is determined based on the spatial distribution density of the combustible ground features. When the spatial distribution density of the combustible ground features is higher than a preset density threshold, N takes a first value. When the spatial distribution density of the combustible ground features is lower than the preset density threshold, N takes a second value, and the first value is greater than the second value.

6. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 4, characterized in that, The step of performing spatial overlay analysis on the candidate fire point pixel set and the binary image of ground cover flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground cover flammability includes: Obtain the geographic coordinates of each candidate fire point pixel in the candidate fire point pixel set; Based on the spatial coordinate system and pixel size of the binary image of flammability of ground features, the geographic coordinates are mapped to the pixel grid of the binary image of flammability of ground features to obtain the grid position corresponding to each candidate fire point pixel, wherein the grid position includes row number i and column number j; The index value k is calculated based on the grid location using the formula k=i×W+j, where W is the width of the binary image of the flammability of the ground features. The bit corresponding to the index value is found from the one-dimensional index sequence of the binary image of the flammability of the ground features with constant time complexity, and the pixel value of the bit is used as the pixel value of the candidate fire point pixel at the corresponding position in the binary image of the flammability of the ground features.

7. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 1, characterized in that, After identifying and retaining candidate fire pixels with a value of 1 as real fire pixels, the process also includes: Obtain the retained candidate fire point pixels and use the retained candidate fire point pixels as fire point pixels to be processed; Obtain the connected regions of combustible features in the binary image of the combustibility of the ground features. The connected regions of combustible features are composed of pixels with a pixel value of 1 and eight neighboring pixels connected. Based on the connected regions of the combustible features, the fire point pixels to be processed are clustered, and the fire point pixels to be processed located in the same connected region of combustible features are divided into the same fire point cluster, and the fire point pixels to be processed in different connected regions of combustible features belong to different fire point clusters. For each fire cluster, obtain the pixel values ​​of all fire pixels to be processed within the fire cluster in the binary image of ground feature flammability. If the number of fire point pixels with a value of 1 in the fire point cluster is less than a preset threshold, all fire point pixels in the fire point cluster will be identified as false fire points and removed.

8. The method for debunking on-board fire points based on ground feature flammability constraints according to claim 1, characterized in that, After determining the corresponding candidate fire point pixels as false fire points and removing them, the process further includes: Calculate the ratio of the number of rejected false fire points to the total number of candidate fire point pixels in the candidate fire point pixel set, and use the ratio as the false fire point rejection rate. If the false fire point rejection rate exceeds a preset rejection threshold, it is determined that the flammability classification result of the binary image of ground features deviates from the actual surface conditions of the target area, triggering the process of reconstructing the binary image of ground features in the target area.

9. A satellite fire point debunking device based on ground object flammability constraints, characterized in that, The device includes: The first acquisition module is used to acquire a set of candidate fire point pixels obtained based on brightness temperature feature detection of remote sensing images, and to determine the target area corresponding to the set of candidate fire point pixels. The second acquisition module is used to acquire a binary image of the flammability of ground features in the target area, wherein the pixel value of flammable ground feature pixels in the binary image of flammable ground features is 1, and the pixel value of non-flammable ground feature pixels is 0. The analysis module is used to perform spatial overlay analysis on the candidate fire point pixel set and the binary image of ground cover flammability to obtain the pixel value of each candidate fire point pixel at the corresponding position in the binary image of ground cover flammability. The determination module is used to determine the corresponding candidate fire point pixel as a false fire point and remove it when the pixel value is 0, and to determine the corresponding candidate fire point pixel as a real fire point and retain it when the pixel value is 1.

10. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements a method for removing false launch points on a satellite based on ground object flammability constraints as described in any one of claims 1-8.

Images