Multi-type anomaly instant remote sensing detection method and system

Abstract

The present application relates to the technical field of remote sensing feature extraction, in particular to a multi-type abnormality instant remote sensing detection method and system, comprising the following steps: calling a multi-channel remote sensing image, screening disturbance pixels to judge the aggregation and connected boundary, calculating the mean of spectral difference and eliminating abnormal pixels, screening a thermal wave band to analyze energy path structure, extracting a red edge spectrum to compare slope difference and identify the ground state, and reorganizing field segment order to generate a mapping result, in the present application, by constructing a disturbance change trend basis, combining direction consistency judgment and heat source path structure identification, the extraction capability of the spatial distribution and evolution characteristics of the ground abnormality is enhanced, the reflection difference and radiation feature fusion mode among multiple source channels are adopted, high-density identification and dynamic structure restoration of the abnormal region are realized, the expression structure and transmission efficiency of the abnormal information are optimized by using the field semantic reorganization strategy, and the spatial precision, semantic integrity and communication adaptability of the ground abnormality detection are simultaneously improved.

Description

Technical Field

[0001] This invention relates to the field of remote sensing feature extraction technology, and in particular to a method and system for real-time remote sensing detection of multiple types of anomalies. Background Technology

[0002] The field of remote sensing feature extraction technology encompasses methods and systems for feature extraction and recognition of remote sensing images from aerospace platforms. It focuses on analyzing and processing optical remote sensing data to extract geospatial information such as land cover type, water distribution, vegetation status, and geological morphology. This information is then used to perform quantitative or qualitative identification and classification using specific physical models or spectral indicators. Remote sensing feature extraction is applied in various fields, including land use analysis, environmental monitoring, disaster identification, and resource surveys. Its overall technical system includes image preprocessing, spectral feature modeling, spatial distribution extraction, time series analysis, and rule-based or statistical model-based classification and diagnosis. Common data sources include multispectral, hyperspectral, and radar remote sensing images. Processing methods include exponential inversion-based methods, spatial clustering analysis, rule-based representation, and model-driven feature extraction. Among them, the multi-type anomaly real-time remote sensing detection method refers to the process of identifying and expressing specific surface anomalies by extracting spatial information and analyzing semantic structure from multispectral or hyperspectral remote sensing data from optical remote sensing payloads. It covers the identification of surface anomaly change areas, the diagnosis of surface anomaly types, and the semantic expression and structured coding of anomaly features. Specifically, it is accomplished by an anomaly mask extraction method driven by prior knowledge, a surface category change map construction method combined with clustering algorithms, anomaly diagnosis method based on feature indices and rule bases, and a text generation and compression method based on fixed formats. Its processing flow includes constructing a low-dimensional spatial expression, applying density-based clustering to detect anomaly areas, calling the rule base to complete the classification of disasters or environmental events, and completing the extraction of structured parameters and organization of BeiDou short message content according to the target type.

[0003] Traditional real-time remote sensing detection technology constructs anomaly masks on remote sensing images using prior rules and calls a fixed rule base to determine anomaly types. When processing changes in surface disturbance features, it does not consider the temporal continuity and directional characteristics of reflectivity changes, resulting in an inability to effectively reflect the dynamic correlation between pixels in disturbance identification. In particular, pixel anomalies are easily isolated in complex areas, leading to blurred boundary recognition. Existing structured expression processes fail to perform difference judgment and field optimization based on semantic relationships between fields. They organize and transmit field content using fixed templates, resulting in low transmission efficiency in high-frequency anomaly areas and high repetition rate of structured expression. This makes it unable to support the multi-type expression needs and short message capacity control requirements in complex scenarios. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a real-time remote sensing detection method for multiple types of anomalies, including the following steps: To achieve the above objectives, the present invention adopts the following technical solution: a multi-type anomaly real-time remote sensing detection method, comprising the following steps: S1: Call up multi-channel satellite remote sensing images, analyze the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filter out reverse-variable pixels and record the frequency, determine the degree of spatial clustering, and extract the perturbation boundary in combination with connectivity to obtain channel direction perturbation distribution information; S2: Using the channel direction perturbation distribution information, compare the angle between the spectral vectors of the boundary pixels and the neighboring pixels, construct a direction consistency set, calculate the mean spectral difference within the set, determine the stability of the direction change pattern within the set, remove edge abnormal pixels, and obtain nested neighboring spectral vector data. S3: Based on the nested data of the neighborhood spectral vectors, filter the radiation channels of the main thermal band and the secondary thermal band, analyze the energy path offset trend and spatial distribution characteristics, mark the positional relationship of the pixels in the path structure, and obtain the spatial heat source path analysis results. S4: Based on the spatial heat source path analysis results, call the red edge spectral sequence, analyze the changing trends of the rising and falling segments, compare the slope differences between the two segments, screen the set of pixels with abnormal curve structures, identify the reflection breakpoint area and determine the surface state type, and obtain the surface anomaly type determination information. S5: Based on the surface anomaly type determination information, analyze the semantic composition of the fields in the information structure, compare the semantic overlap between fields, determine the role switching conditions, filter the replaceable field segments, calculate the content overlap ratio between segments and reorganize the field segment order, and obtain the segment reorganization field mapping result.

[0005] As a further aspect of the present invention, the channel direction disturbance distribution information includes pixel reflectance direction difference characteristics, reverse change frequency, and disturbance clustering boundary structure; the neighborhood spectral vector nesting data specifically includes spectral angle consistency grouping results, spectral vector direction mean index, and disturbance anomaly removal results; the spatial heat source path analysis results include main heat band channel number, secondary heat band energy distribution information, and energy path label structure; the surface anomaly type determination information specifically refers to breakpoint slope structure, reflection trend direction, and surface state label; and the segment reorganization field mapping results include semantic field cross structure, field role switching identifier, and field arrangement order.

[0006] As a further aspect of the present invention, the step of obtaining the channel direction disturbance distribution information specifically includes: S101: Acquire multi-channel satellite remote sensing images, detect the reflectance of the red band and near-infrared band in the target area between consecutive time slices, analyze the reflectance change direction of each pixel in multiple time slices, compare the channel reflectance direction differences of corresponding pixels in adjacent time slices, determine the consistency of reflectance change direction between channels, and generate a direction change consistency coefficient. S102: Based on the directional change consistency coefficient, filter out pixels with reverse changes, count the frequency of reverse changes in each pixel, determine the degree of clustering of target pixels in spatial distribution, and establish spatial clustering distribution parameters. S103: Based on the spatial aggregation distribution parameters, analyze the distribution relationship of adjacent perturbation pixels within the target area, determine spatial connectivity, identify the connectivity boundary of adjacent perturbation pixels, and obtain channel direction perturbation distribution information.

[0007] As a further aspect of the present invention, the step of obtaining the nested neighborhood spectral vector data specifically includes: S201: Call the channel direction perturbation distribution information, compare the angle between the spectral vectors of each boundary pixel and the neighboring pixels, group the pixels, construct a set of pixels with consistent spectral change direction, and generate the pixel set grouping result; S202: Based on the pixel set grouping results, the mean spectral difference within the set is used as a benchmark to calculate the deviation between the variation difference of each pixel within the group and the mean spectral difference. The stability of the directional change pattern of each group of pixels is judged based on the distribution and dispersion of the deviation. The pixel combination with abnormal edge disturbance is removed according to the directional fluctuation amplitude that exceeds the mean constraint of the deviation, and the directional pattern stability parameter is obtained. S203: Call the directional pattern stability parameter to reconstruct the directional consistency partition of the target region spectral vector, analyze the pixel distribution relationship within the partition, establish a nested partition structure, and obtain nested data of the neighborhood spectral vector.

[0008] As a further aspect of the present invention, the steps for obtaining the spatial heat source path analysis results are specifically as follows: S301: Call the nested data of the neighborhood spectral vector, filter the pixel radiation channels of the main thermal band and the secondary thermal band in the target area, analyze the radiation energy value of the pixel in each thermal band, combine and compare the radiation energy distribution relationship between each pair of pixels, and generate energy channel combination coefficients. S302: Based on the energy channel combination coefficient, determine the spatial continuity of energy value changes along the energy conduction path, compare the continuous distribution of energy offset trends in the path, analyze the distribution pattern of energy changes, and obtain the energy path continuity parameters. S303: Call the energy path continuity parameter to identify the path separation characteristics of the heat concentration area and the dispersion area, optimize the energy path label of the pixel in the spatial location, and obtain the spatial heat source path analysis results.

[0009] As a further aspect of the present invention, the step of obtaining the surface anomaly type determination information specifically includes: S401: Call the spatial heat source path analysis results, collect the red edge spectral reflectance sequence in the same area, analyze the rising and falling trends of the reflectance curve of each pixel, compare the reflectance change directions of the two segments, and obtain the curve change direction comparison parameters. S402: Based on the curve change direction comparison parameters, determine the morphological extension characteristics of the curve at the breakpoint, calculate the slope ratio between the rising and falling segments of each pixel, analyze the distribution difference of the slope ratio among different pixels, and generate the curve slope difference coefficient. S403: Call the curve slope difference coefficient to filter the set of pixels with abnormal curve structure, identify the reflection breakpoint area, determine the surface state type of the pixel set, and obtain the surface anomaly type determination information.

[0010] As a further aspect of the present invention, the process of determining the land surface condition type of the pixel set specifically involves: based on the curve slope difference coefficient, obtaining the symmetry index formed by the upward and downward slopes of each pixel at the breakpoint; calculating the dispersion of the index between pixels; constructing a type boundary benchmark interval based on the dispersion; defining the type determination boundary for distinguishing land surface conditions according to the benchmark interval; determining the segment to which the symmetry index corresponding to each pixel belongs in the type determination boundary; and determining the land surface condition type of each pixel in the pixel set based on the type label corresponding to the target segment, thereby obtaining the land surface anomaly type determination information.

[0011] As a further aspect of the present invention, the acquisition step specifically includes: S501: Call the surface anomaly type determination information, analyze the semantic composition of the fields in the information structure, compare the semantic overlap of each field in the anomaly category, level and time description, and generate field semantic overlap parameters; S502: Based on the semantic overlap parameter of the fields, determine the role switching conditions between multiple fields, filter the fields whose semantic expression direction can be replaced, calculate the crossover ratio of the content between the fields, and obtain the crossover coefficient of the field content. S503: Call the cross coefficient of the field content, adjust the field arrangement according to the degree of overlap and reorganize the field segment order to obtain the segment reorganization field mapping result.

[0012] Real-time remote sensing detection system for multiple types of anomalies, including: The directional disturbance detection module calls up multi-channel satellite remote sensing images, analyzes the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filters out reverse-changing pixels and records their frequency, determines the degree of spatial clustering, and extracts disturbance boundaries by combining connectivity to obtain channel directional disturbance distribution information; The spectral vector recognition module uses the channel direction perturbation distribution information to compare the angle between the spectral vectors of boundary pixels and neighboring pixels, constructs a direction consistency set, calculates the mean spectral difference within the set, judges the stability of the direction change pattern within the set, removes abnormal edge pixels, and obtains nested neighboring spectral vector data. The heat source path construction module filters the radiation channels of the main heat band and the secondary heat band based on the nested data of the neighborhood spectral vector, analyzes the energy path offset trend and spatial distribution characteristics, marks the positional relationship of pixels in the path structure, and obtains the spatial heat source path analysis results. Based on the spatial heat source path analysis results, the reflection breakpoint identification module calls the red edge spectral sequence, analyzes the changing trends of the rising and falling segments, compares the slope differences between the two segments, filters the set of pixels with abnormal curve structures, identifies the reflection breakpoint area and determines the surface state type, and obtains surface anomaly type determination information. The field structure mapping module analyzes the semantic composition of fields in the information structure based on the surface anomaly type determination information, compares the semantic overlap between fields, determines the role switching conditions, filters replaceable field segments, calculates the content overlap ratio between segments and reorganizes the field segment order, and obtains the segment reorganization field mapping result.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by constructing a foundation for perturbation change trends and combining directional consistency judgment with heat source path structure identification, the ability to extract spatial distribution and evolution characteristics of surface anomalies is enhanced. By adopting a fusion method of reflection differences and radiation characteristics between multi-source channels, high-density identification and dynamic structure reconstruction of anomaly areas are achieved. The expression structure and transmission efficiency of anomaly information are optimized by using a field semantic recombination strategy, thereby simultaneously improving the spatial accuracy, semantic integrity, and communication adaptability of surface anomaly detection. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0017] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0018] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0019] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0021] Please see Figure 1 This invention provides a real-time remote sensing detection method for multiple types of anomalies, including the following steps: S1: Call up multi-channel satellite remote sensing images, analyze the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filter out reverse-variable pixels and record the frequency, determine the degree of spatial clustering, and extract the perturbation boundary in combination with connectivity to obtain channel direction perturbation distribution information; S2: Utilize the channel direction perturbation distribution information, compare the angle between the spectral vectors of boundary pixels and neighboring pixels, construct a direction consistency set, calculate the mean spectral difference within the set, determine the stability of the direction change pattern within the set, remove edge abnormal pixels, and obtain nested neighboring spectral vector data. S3: Based on the nested data of the neighborhood spectral vector, filter the radiation channels of the main thermal band and the secondary thermal band, analyze the energy path offset trend and spatial distribution characteristics, mark the positional relationship of the pixels in the path structure, and obtain the spatial heat source path analysis results. S4: Based on the spatial heat source path analysis results, call the red edge spectral sequence, analyze the changing trends of the rising and falling segments, compare the slope differences between the two segments, screen the set of pixels with abnormal curve structures, identify the reflection breakpoint area and determine the surface state type, and obtain the surface anomaly type determination information. S5: Based on the information on the type of surface anomaly, analyze the semantic composition of the fields in the information structure, compare the semantic overlap between fields, determine the role switching conditions, filter the replaceable field segments, calculate the content overlap ratio between segments and reorganize the field segment order, and obtain the segment reorganization field mapping results.

[0022] The channel directional disturbance distribution information includes pixel reflectance directional difference characteristics, reverse change frequency, and disturbance clustering boundary structure. The neighboring spectral vector nesting data specifically includes spectral angle consistency grouping results, spectral vector directional mean index, and disturbance anomaly removal results. The spatial heat source path analysis results include the main heat band channel number, secondary heat band energy distribution information, and energy path label structure. The surface anomaly type determination information specifically refers to the breakpoint slope structure, reflection trend direction, and surface state label. The segment reorganization field mapping results include semantic field cross structure, field role switching identifier, and field arrangement order.

[0023] Please see Figure 2 The specific steps for obtaining channel directional disturbance distribution information are as follows: S101: Acquire multi-channel satellite remote sensing images, detect the reflectance of the red band and near-infrared band in the target area between consecutive time slices, analyze the reflectance change direction of each pixel in multiple time slices, compare the channel reflectance direction differences of corresponding pixels in adjacent time slices, determine the consistency of reflectance change direction between channels, and generate a direction change consistency coefficient. By using multi-channel satellite remote sensing imagery and focusing on 10 specific pixels in a farmland area, data was collected. (June 1, 2025) and (June 8, 2025) Reflectance values ​​in the red band (Band 4) and near-infrared band (Band 8) for two time pieces. These reflectance values ​​are from atmospherically corrected surface reflectance data, and the range is [range missing]. For each cell First, calculate the direction of reflectivity change in the two wavebands. and For example, for pixel P104, its red band reflectance is from... Descending to The direction of change is Near-infrared reflectivity from Descending to The direction of change is also Next, by calculating the product of the two directions... To compare the differences in the direction of change between channels. When the product is... When this occurs, it is determined to be a reverse change, for example, the red band of pixel P103 rises (direction). The near-infrared band decreases (direction). The product is Finally, using the formula Calculate the consistency coefficient of directional change When the directions are consistent When the change is reversed For example, on page 104... And P103 This process is repeated for each pixel to generate a consistency coefficient for directional changes across all monitored pixels, providing a quantitative basis for subsequent steps to screen out anomalous pixels.

[0024] S102: Based on the consistency coefficient of directional change, filter pixels with reverse changes, count the frequency of reverse changes in each pixel, determine the degree of clustering of target pixels in spatial distribution, and establish spatial clustering distribution parameters. Based on the consistency coefficient of directional change First, filter and count the pixels that show reverse changes. The pixels form a set of reverse-shifted pixels. According to the example in S101, the values ​​of pixels P103 and P106... All Therefore, they were included Since this example only analyzes... arrive For a given time slice, the frequency of each reverse-shifted pixel is [missing information]. Next, for each reverse-shifted pixel... Statistics Within the neighborhood Number of pixels in the set And by calculating the proportion of reverse pixels in the neighborhood This is used to determine the degree of clustering in its spatial distribution. For example, assuming that there are no other reverse-shifted pixels in the neighborhood of pixel P103, then... The proportion of its neighboring reverse pixels Subsequently, based on the spatial clustering threshold... An evaluation was conducted, and the threshold was set as follows: .if If the pixel is found to be in a clustered state, then it is determined that the pixel is in a clustered state. Finally, the percentage of neighboring pixels of each pixel is calculated. As a spatial aggregation distribution parameter .

[0025] S103: Based on the spatial aggregation distribution parameters, analyze the distribution relationship of adjacent perturbation pixels within the target area, determine spatial connectivity, identify the connectivity boundary of adjacent perturbation pixels, and obtain channel direction perturbation distribution information; Based on spatial clustering distribution parameters The distribution relationship between adjacent reverse-shifted pixels is analyzed, and the connected boundaries are determined. First, based on... For the reverse-variable pixel set Traverse the pixels in the array and search for their... All belonging to the neighborhood Find the neighboring cells and establish adjacency relationships. For example, if pixels P107 and P110 are adjacent and both belong to... Then establish Relationships. Then, a graph search algorithm (such as breadth-first search) is used to traverse all cells with established adjacency relationships, grouping interconnected cells into a single connected component. In this example, we assume that the connected components are obtained. (size 1) and (Size is 3). Next, the size of the connected component is compared with the minimum connected component threshold. The threshold is set based on the smallest identifiable abnormal area in the remote sensing image, and in this example, it is set to [value missing]. .because and The size is smaller than Therefore, they are all determined to be disconnected and do not constitute perturbation distribution information. If... Set as ,but These will be retained. For retained connected components, it is determined whether their internal cells are adjacent to external cells to define the connectivity boundary. Finally, all connected components that satisfy the minimum connected component threshold and their boundary information are integrated into channel direction perturbation distribution information. .

[0026] Please see Figure 3The specific steps for obtaining nested neighborhood spectral vector data are as follows: S201: Call the channel direction perturbation distribution information, compare the angle between the spectral vectors of each boundary pixel and the neighboring pixels, group the pixels, construct a set of pixels with consistent spectral change direction, and generate the pixel set grouping result; Call channel direction disturbance distribution information Focusing on pixels within the perturbation region, pixels are grouped by comparing the angles between spectral vectors. First, using... Based on the spectral reflectance at time t, for each pixel within the perturbation region and its neighboring pixels Extract its spectral vector and This vector is composed of the reflectance of the blue band (B2), green band (B3), red band (B4), and near-infrared band (B8). Subsequently, the spectral angles of the two vectors are calculated. For example, suppose the spectral vectors of pixel P107 and its neighboring pixel P111 are respectively and The calculated spectral angle Subsequently, the calculated spectral angles are compared with the spectral similarity threshold. The threshold is set for comparison. .if Then the pixel and The spectral changes are determined to be in the same direction, and a connection is established. In this example, the spectral angles of P107 and P111... Greater than the threshold Therefore, they are not grouped into the same set. Finally, using connected component analysis, all interconnected cells are clustered into one or more sets of cells with consistent spectral variation directions. It also records the list of pixels it contains.

[0027] S202: Based on the pixel set grouping results, the mean spectral difference within the set is used as a benchmark to calculate the deviation between the variation difference of each pixel within the group and the mean spectral difference. The stability of the directional change pattern of each group of pixels is judged based on the distribution and dispersion of the deviation. The pixel combination with abnormal edge disturbance is removed according to the directional fluctuation amplitude that exceeds the mean constraint of the deviation, and the directional pattern stability parameter is obtained. Based on pixel set grouping results The spectral orientation variation differences of pixels within a set over continuous time are calculated, and edge anomalous pixels are removed. For each set... Each cell in Calculate its continuous time series Normalized Differential Vegetation Index (NDVI) change rate For example, the NDVI of pixel P104 is from... of Descending to of The rate of change Next, calculate the values ​​of all cells within the set. mean and standard deviation Standard deviation This is the directional mode stability parameter; the smaller the value, the more stable the mode. Maximum tolerable fluctuation threshold. In this example, it is set to This value is referenced to the NDVI fluctuation range of normal homogeneous materials over time. Assume a set... of If the pattern is stable, then it is determined to be stable. Subsequently, outlier pixels at the edges are removed based on the magnitude of directional fluctuations; for example, pixels with fluctuating amplitudes are removed. Cells exceeding twice the standard deviation. For example, if the fluctuation range of P104 is... less than Therefore, they are not removed. Finally, the stability standard deviation of each set is obtained. And the list of cells after removing anomalies .

[0028] S203: Call the directional pattern stability parameter to reconstruct the directional consistency partition of the target region's spectral vector, analyze the pixel distribution relationship within the partition, establish a nested partition structure, and obtain nested data of the neighborhood spectral vector; Calling direction mode stability parameters and the list of cells after removing anomalies Reconstruct the regional partitioning structure and establish nested data. First, stabilize all patterns (i.e., ) pixel set Merge to form the restructured partition structure For example, when of At that time, it was included in the reconstructed partition. Subsequently, the pixel distribution relationship within the partitions was analyzed to establish a nested partition structure. The specific method involved: reconstructing the entire region... Defined as a first-level partition, and within it identified those with high spatial connectivity and lower... Values ​​(e.g.) subregions of ) These sub-regions are defined as second-level partitions, nested within the first-level partitions. In. For example, if in Internal identification of one subregion This sub-region will be defined as a second-level partition. Ultimately, the partition structure will be... (Level 1 partition) and nested partitions The pixel list and spatial geometric information of the (secondary partition) are integrated into nested neighborhood spectral vector data, which describes the core and edge distribution of the anomalous region in a hierarchical structure.

[0029] Please see Figure 4 The specific steps for obtaining the results of the space heat source path analysis are as follows: S301: Call the nested data of the neighborhood spectral vector, filter the pixel radiation channels of the main thermal band and the secondary thermal band in the target area, analyze the radiation energy value of the pixel in each thermal band, combine and compare the radiation energy distribution relationship between each pair of pixels, and generate energy channel combination coefficients. The system calls up nested neighbor spectral vector data to filter the primary and secondary thermal bands within the target area, and then compares the radiant energy distribution relationships between pixels. This embodiment uses MuDuo-1B shortwave infrared (SWIR) data, selecting radiant energy values ​​(radiance, units of...) for Band 11 (primary thermal band) and Band 12 (secondary thermal band). First, extract and reconstruct the partitions. Each pixel radiance value and As shown in Table 2, the radiance values ​​of pixels P115 and P116 are respectively and Subsequently, for adjacent pixel pairs Calculate the energy difference ratio between the primary and secondary thermal bands. For example, for pixels P115 and P116, the energy difference in the main thermal band Energy difference in the secondary thermal band Therefore, the ratio Finally, adjacent pixel pairs are... Energy difference ratio As the energy channel combination coefficient .

[0030] S302: Based on the energy channel combination coefficient, determine the spatial continuity of energy value changes along the energy conduction path, compare the continuous distribution of energy offset trends in the path, analyze the distribution pattern of energy changes, and obtain the energy path continuity parameters. Based on energy channel combination coefficient To determine the spatial continuity of energy value changes along the energy conduction path, we first define the energy conduction path as a chain of adjacent pixels along the direction of increasing radiance value in the main thermal band. Subsequently, the continuous distribution of energy shift trends along the path is determined, specifically by calculating the path... Energy channel combination coefficients of all adjacent pixel pairs Standard deviation To achieve this. This refers to the energy path continuity parameter. This value reflects the distribution of energy offset along the path; the lower the value, the smoother and more continuous the energy change. When determining continuity, it is also necessary to compare the angle between adjacent gradient vectors along the path. Consistency threshold with energy gradient (Set as) ),like If so, the path is determined to be continuous at that point. Through analysis... The size can distinguish between uniform variation (low) The heat changes caused by local heat sources (high) and the heat changes caused by local heat sources (high) The mutation caused by ) . Finally, the output path Standard deviation As a parameter for energy path continuity.

[0031] S303: Call the energy path continuity parameter to identify the path separation characteristics of heat concentration areas and dispersion areas, optimize the energy path label of the pixel in spatial location, and obtain the spatial heat source path analysis results; Calling the energy path continuity parameter This method identifies path separation features between areas of concentrated and dispersed heat, and optimizes the energy path labels of pixels in spatial location. First, a discontinuity threshold is set... Set as This threshold is referenced to the minimum fluctuation value of the standard deviation of the rate of change of the heat gradient in the conduction path of the artificial heat source. According to... and By comparing the paths, we can identify the path separation characteristics between areas of concentrated and dispersed heat. If a certain path... If the radiance at the starting point is higher than the surrounding average, it is determined to be a region of concentrated heat. If the endpoint radiance is lower than the surrounding average, it is identified as a heat dispersion area. Next, based on the identified features, the energy path labels of the pixels in spatial location are optimized. For example, the pixel label for a heat concentration area is optimized to "heat source core pixel," the pixel label for a heat dispersion area is optimized to "heat dissipation pixel," and the pixel label for the remaining paths is "heat conduction pixel." Finally, the spatial heat source path analysis results are obtained, which include the optimized energy path label for each pixel and the geometric information of the corresponding path. For example, if the path of P116 is identified as a heat concentration area, its label is optimized to "heat source core pixel."

[0032] Please see Figure 5 The specific steps for obtaining information on the type of surface anomaly are as follows: S401: Call the spatial heat source path analysis results, collect the red edge spectral reflectance sequence in the same area, analyze the changing trend of the rising and falling segments of the reflectance curve of each pixel, compare the two segments of reflectance change direction, and obtain the curve change direction comparison parameter. The spatial heat source path analysis results are used to collect red-edge spectral reflectance sequences within the same region and analyze the changing trends of the red-edge curves. This embodiment uses MuDuo-1B hyperspectral data. arrive Red-edge reflectance data were collected within the specified range. Taking pixel P104 as an example, its red-edge reflectance sequence is shown in Table 3. First, the rising and falling trends of the reflectance curve for each pixel were analyzed, specifically by calculating the reflectance difference between adjacent bands. To determine the trend direction. For example, for pixel P104, the reflectance difference between bands B50 and B51 is... The trend is upward. Since the red-edge reflectance curve of normal vegetation should maintain a continuous upward trend, a plateau effect or downward trend at the end of the upward segment indicates that the vegetation is under stress. Subsequently, the directions of reflectance changes in the upward and downward segments of the red edge are compared. and By calculating the product of the two To quantify the difference in direction. In this example, since the curve of P104 does not have a clear descending segment, its product is positive. Ultimately, As a parameter for comparing the direction of curve change This provides a foundation for subsequent slope analysis.

[0033] S402: Based on the curve change direction comparison parameters, determine the morphological extension characteristics of the curve at the break point, calculate the slope ratio between the rising and falling segments of each pixel, analyze the distribution difference of the slope ratio among different pixels, and generate the curve slope difference coefficient. Based on the comparison parameters of curve change direction The slope ratio between the rising and falling segments of the red edge is calculated, and the difference distribution is analyzed. First, the breakpoint of the curve is determined as the wavelength where the second derivative equals zero, i.e., the red edge position. Next, the maximum slope of the rising and falling segments of the red edge for each pixel is calculated. and And calculate its slope ratio. For example, suppose the maximum slope of the rising red edge of cell P104 is... Maximum slope of the descending segment Then its slope ratio This ratio reflects the symmetry of the curve's shape, and is typical of normal, healthy vegetation. near Subsequently, the distribution differences of slope ratios among different pixels were analyzed, that is, the mean and standard deviation of the slope ratios of all pixels in the pixel set were calculated. Standard deviation This reflects the spatial consistency of the red-edge morphology within the abnormal region. Ultimately, it will... As the curve slope difference coefficient .

[0034] S403: Call the curve slope difference coefficient, filter the set of pixels with abnormal curve structure, identify the reflection breakpoint area, determine the surface state type of the pixel set, and obtain the surface anomaly type determination information. Call the curve slope difference coefficient The process involves filtering pixel sets with abnormal curve structures, identifying reflection discontinuity regions, and determining the surface condition type of the pixel sets. First, slope ratios are selected. Greater than the abnormality ratio threshold The pixels form a set of curved structure anomalous pixels. Due to P104 Therefore, P104 was selected for inclusion in this set. Subsequently, the surface condition type of the pixels within this set was determined. The specific process was as follows: First, the symmetry index of each pixel was calculated. For example, on page 104... Next, a type boundary baseline interval is constructed, which is based on the symmetry index distribution of the labeled samples and the average offset of the weight distribution range. Build as Subsequently, this interval is equidistantly defined as the type determination boundary, such as... (water bodies, ), (vegetation, )and (bare land, Finally, the symmetry index of each pixel is determined to belong to the segment within the type determination boundary, and the surface condition type is determined based on the corresponding type label. For example, P104... Therefore, its surface condition was determined to be "water body". Ultimately, the surface anomaly type determination information was obtained. This includes each anomalous cell and its corresponding surface condition type.

[0035] Please see Figure 6 The specific steps to obtain are as follows: S501: Call the surface anomaly type determination information, analyze the semantic composition of the fields in the information structure, compare the semantic overlap of each field in the anomaly category, level and time description, and generate field semantic overlap parameters. Calling surface anomaly type determination information This involves analyzing the semantic composition of the fields within the information structure. In this example, the information structure includes... (Surface condition type) (Severity of abnormality) Fields such as (confidence level) are used. First, each field is mapped to three semantic dimensions: anomaly category (A), anomaly level (L), and time description (T). Then, a semantic weight (such as sovereign weight) is assigned to each field on each dimension. Secondary weight ).For example, The weights in the L dimension are ,and The weights in the L dimension are Subsequently, by calculating each pair of fields... Shared weights across three semantic dimensions To compare their semantic overlap. For example, and semantic overlap This indicates a moderate overlap between the two in terms of anomaly level. Ultimately, the semantic overlap of each pair of fields will be determined. As a parameter for semantic overlap of fields, it provides a quantitative basis for subsequent field reorganization.

[0036] S502: Based on the semantic overlap parameter of the fields, determine the role switching conditions between multiple fields, filter the fields whose semantic expression direction can be replaced, calculate the crossover ratio of the content between the fields, and obtain the crossover coefficient of the field content. Based on field semantic overlap parameters This involves determining the role switching conditions between multiple fields and calculating the overlap ratio of content between different tiers. First, it compares the semantic weight difference between the fields to be judged with the field role switching threshold range. To determine the conditions for character switching. Threshold range. Set as It is referenced to the average offset of the weight distribution range of the sample fields under the same semantic dimension. For example, fields and The weight difference along the L dimension is The value is at the upper limit of the threshold range, therefore this field is determined to meet the role-switching condition. Subsequently, replaceable field segments are selected, i.e., parameters that meet the role-switching condition and have semantic overlap. Above the threshold Field pairs, in this example and This condition is met. Next, the content overlap ratio between replaceable segments is calculated. ,in This refers to the information entropy of the field. For example, suppose... and ,but Ultimately, As the cross coefficient of field content The result indicates and The content has Cross-relationships.

[0037] S503: Call the cross coefficient of field content, adjust the field arrangement according to the degree of overlap and reorganize the field segment order, and obtain the segment reorganization field mapping result; Cross coefficient of field content The arrangement of fields is adjusted and the order of field segments is reorganized based on the degree of overlap. The goal of this reorganization is to optimize the information structure to meet the priority requirements of BeiDou short message transmission for core information. The reorganization principle is as follows: fields with high priority and low crossover coefficients are placed at the beginning of the information packet; for fields with medium priority but high crossover coefficients with another field, the more concise field is placed at the beginning, and the other is used as a supplementary segment. In this example, because... and The cross coefficient is ,and The semantics are more direct, so it is retained in the core section and... Downgraded to an extended tier. The final field reorganization order is as follows: (Abnormal type) (Geographic coordinates) (Time of occurrence) (Abnormality Level) (Event Number) and (Confidence level). This reorganization ensures the core components of the information packet (e.g., the previous...) are secure. (Bytes) maximize the preservation of key, independent information, greatly improving the transmission efficiency and effectiveness of information through bandwidth-constrained communication links. Finally, the segment reassembly field mapping result is obtained. This refers to the reorganized field order and its mapping position in the information packet.

[0038] Please see Figure 7 Real-time remote sensing detection system for multiple types of anomalies, including: The directional disturbance detection module calls up multi-channel satellite remote sensing images, analyzes the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filters out reverse-changing pixels and records their frequency, determines the degree of spatial clustering, and extracts disturbance boundaries by combining connectivity to obtain channel directional disturbance distribution information; The spectral vector recognition module utilizes channel direction perturbation distribution information to compare the angle between the spectral vectors of boundary pixels and neighboring pixels, constructs a set of directional consistency, calculates the mean spectral difference within the set, judges the stability of the direction change pattern within the set, removes abnormal edge pixels, and obtains nested neighboring spectral vector data. The heat source path construction module filters the radiation channels of the main heat band and the secondary heat band based on the nested data of the neighborhood spectral vector, analyzes the energy path offset trend and spatial distribution characteristics, marks the positional relationship of pixels in the path structure, and obtains the spatial heat source path analysis results. Based on the spatial heat source path analysis results, the reflection breakpoint identification module calls the red edge spectral sequence to analyze the changing trends of the rising and falling segments, compares the slope differences between the two segments, filters the set of pixels with abnormal curve structures, identifies the reflection breakpoint area and determines the surface state type, and obtains surface anomaly type determination information. The field structure mapping module determines information based on the type of surface anomaly, analyzes the semantic composition of fields in the information structure, compares the semantic overlap between fields, determines the role switching conditions, filters replaceable field segments, calculates the content overlap ratio between segments, reorganizes the field segment order, and obtains the segment reorganization field mapping result.

[0039] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A real-time remote sensing detection method for multiple types of anomalies, characterized in that, Includes the following steps: S1: Call up multi-channel satellite remote sensing images, analyze the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filter out reverse-variable pixels and record the frequency, determine the degree of spatial clustering, and extract the perturbation boundary in combination with connectivity to obtain channel direction perturbation distribution information; S2: Using the channel direction perturbation distribution information, compare the angle between the spectral vectors of the boundary pixels and the neighboring pixels, construct a direction consistency set, calculate the mean spectral difference within the set, determine the stability of the direction change pattern within the set, remove edge abnormal pixels, and obtain nested neighboring spectral vector data. The nested neighborhood spectral vector data specifically includes the spectral angle consistency grouping results, the spectral vector direction mean index, and the perturbation anomaly removal results. S3: Based on the nested data of the neighborhood spectral vectors, filter the radiation channels of the main thermal band and the secondary thermal band, analyze the energy path offset trend and spatial distribution characteristics, mark the positional relationship of the pixels in the path structure, and obtain the spatial heat source path analysis results. S4: Based on the spatial heat source path analysis results, call the red edge spectral sequence, analyze the changing trends of the rising and falling segments, compare the slope differences between the two segments, screen the set of pixels with abnormal curve structures, identify the reflection breakpoint area and determine the surface state type, and obtain the surface anomaly type determination information.

2. The real-time remote sensing detection method for multiple types of anomalies according to claim 1, characterized in that, The channel directional disturbance distribution information includes pixel reflectivity directional difference characteristics, reverse change frequency, and disturbance aggregation boundary structure. The spatial heat source path analysis results include the main heat band channel number, secondary heat band energy distribution information, and energy path label structure. The surface anomaly type determination information specifically refers to the breakpoint slope structure, reflection trend direction, and surface state label.

3. The real-time remote sensing detection method for multiple types of anomalies according to claim 1, characterized in that, The specific steps for obtaining the channel directional disturbance distribution information are as follows: S101: Acquire multi-channel satellite remote sensing images, detect the reflectance of the red band and near-infrared band in the target area between consecutive time slices, analyze the reflectance change direction of each pixel in multiple time slices, compare the channel reflectance direction differences of corresponding pixels in adjacent time slices, determine the consistency of reflectance change direction between channels, and generate a direction change consistency coefficient. S102: Based on the directional change consistency coefficient, filter out pixels with reverse changes, count the frequency of reverse changes in each pixel, determine the degree of clustering of target pixels in spatial distribution, and establish spatial clustering distribution parameters. S103: Based on the spatial aggregation distribution parameters, analyze the distribution relationship of adjacent perturbation pixels within the target area, determine spatial connectivity, identify the connectivity boundary of adjacent perturbation pixels, and obtain channel direction perturbation distribution information.

4. The real-time remote sensing detection method for multiple types of anomalies according to claim 3, characterized in that, The specific steps for obtaining the nested neighborhood spectral vector data are as follows: S201: Call the channel direction perturbation distribution information, compare the angle between the spectral vectors of each boundary pixel and the neighboring pixels, group the pixels, construct a set of pixels with consistent spectral change direction, and generate the pixel set grouping result; S202: Based on the pixel set grouping results, the mean spectral difference within the set is used as a benchmark to calculate the deviation between the variation difference of each pixel within the group and the mean spectral difference. The stability of the directional change pattern of each group of pixels is judged based on the distribution and dispersion of the deviation. The pixel combination with abnormal edge disturbance is removed according to the directional fluctuation amplitude that exceeds the mean constraint of the deviation, and the directional pattern stability parameter is obtained. S203: Call the directional pattern stability parameter to reconstruct the directional consistency partition of the target region spectral vector, analyze the pixel distribution relationship within the partition, establish a nested partition structure, and obtain nested data of the neighborhood spectral vector.

5. The real-time remote sensing detection method for multiple types of anomalies according to claim 4, characterized in that, The specific steps for obtaining the spatial heat source path analysis results are as follows: S301: Call the nested data of the neighborhood spectral vector, filter the pixel radiation channels of the main thermal band and the secondary thermal band in the target area, analyze the radiation energy value of the pixel in each thermal band, combine and compare the radiation energy distribution relationship between each pair of pixels, and generate energy channel combination coefficients. S302: Based on the energy channel combination coefficient, determine the spatial continuity of energy value changes along the energy conduction path, compare the continuous distribution of energy offset trends in the path, analyze the distribution pattern of energy changes, and obtain the energy path continuity parameters. S303: Call the energy path continuity parameter to identify the path separation characteristics of the heat concentration area and the dispersion area, optimize the energy path label of the pixel in the spatial location, and obtain the spatial heat source path analysis results.

6. The real-time remote sensing detection method for multiple types of anomalies according to claim 5, characterized in that, The specific steps for obtaining the surface anomaly type determination information are as follows: S401: Call the spatial heat source path analysis results, collect the red edge spectral reflectance sequence in the same area, analyze the rising and falling trends of the reflectance curve of each pixel, compare the reflectance change directions of the two segments, and obtain the curve change direction comparison parameters. S402: Based on the curve change direction comparison parameters, determine the morphological extension characteristics of the curve at the breakpoint, calculate the slope ratio between the rising and falling segments of each pixel, analyze the distribution difference of the slope ratio among different pixels, and generate the curve slope difference coefficient. S403: Call the curve slope difference coefficient to filter the set of pixels with abnormal curve structure, identify the reflection breakpoint area, determine the surface state type of the pixel set, and obtain the surface anomaly type determination information.

7. The real-time remote sensing detection method for multiple types of anomalies according to claim 6, characterized in that, The process of determining the land surface condition type of the pixel set specifically involves: based on the curve slope difference coefficient, obtaining the symmetry index formed by the upward and downward slopes of each pixel at the breakpoint; calculating the dispersion of the index between pixels; constructing a type boundary benchmark interval based on the dispersion; defining the type determination boundary for distinguishing land surface conditions according to the benchmark interval; determining the segment to which the symmetry index corresponding to each pixel belongs in the type determination boundary; and determining the land surface condition type of each pixel in the pixel set based on the type label corresponding to the target segment, thereby obtaining the land surface anomaly type determination information.

8. The real-time remote sensing detection method for multiple types of anomalies according to claim 1, characterized in that, The method further includes: S5: Based on the surface anomaly type determination information, analyze the semantic composition of the fields in the information structure, compare the semantic overlap between fields, determine the role switching conditions, filter the replaceable field segments, calculate the content overlap ratio between segments and reorganize the field segment order, and obtain the segment reorganization field mapping result. The segment reorganization field mapping result includes semantic field cross structure, field role switching identifier, and field arrangement order.

9. The real-time remote sensing detection method for multiple types of anomalies according to claim 8, characterized in that, The specific steps for obtaining the segment reorganization field mapping result are as follows: S501: Call the surface anomaly type determination information, analyze the semantic composition of the fields in the information structure, compare the semantic overlap of each field in the anomaly category, level and time description, and generate field semantic overlap parameters; S502: Based on the semantic overlap parameter of the fields, determine the role switching conditions between multiple fields, filter the fields whose semantic expression direction can be replaced, calculate the crossover ratio of the content between the fields, and obtain the crossover coefficient of the field content. S503: Call the cross coefficient of the field content, adjust the field arrangement according to the degree of overlap and reorganize the field segment order to obtain the segment reorganization field mapping result.

10. A real-time remote sensing detection system for multiple types of anomalies, characterized in that, The system is used to implement the multi-type anomaly real-time remote sensing detection method according to any one of claims 1-9, the system comprising: The directional disturbance detection module calls up multi-channel satellite remote sensing images, analyzes the difference in reflectance direction between the red band and the near-infrared band in continuous time slices, filters out reverse-changing pixels and records their frequency, determines the degree of spatial clustering, and extracts disturbance boundaries by combining connectivity to obtain channel directional disturbance distribution information; The spectral vector recognition module uses the channel direction perturbation distribution information to compare the angle between the spectral vectors of boundary pixels and neighboring pixels, constructs a direction consistency set, calculates the mean spectral difference within the set, judges the stability of the direction change pattern within the set, removes abnormal edge pixels, and obtains nested neighboring spectral vector data. The heat source path construction module filters the radiation channels of the main heat band and the secondary heat band based on the nested data of the neighborhood spectral vector, analyzes the energy path offset trend and spatial distribution characteristics, marks the positional relationship of pixels in the path structure, and obtains the spatial heat source path analysis results. Based on the spatial heat source path analysis results, the reflection breakpoint identification module calls the red edge spectral sequence, analyzes the changing trends of the rising and falling segments, compares the slope differences between the two segments, filters the set of pixels with abnormal curve structures, identifies the reflection breakpoint area and determines the surface state type, and obtains surface anomaly type determination information. The field structure mapping module analyzes the semantic composition of fields in the information structure based on the surface anomaly type determination information, compares the semantic overlap between fields, determines the role switching conditions, filters replaceable field segments, calculates the content overlap ratio between segments and reorganizes the field segment order, and obtains the segment reorganization field mapping result.

Images