An earthquake disaster area analysis and prediction method and device based on remote sensing data and a medium

Abstract

This invention discloses a method, equipment, and medium for earthquake disaster area analysis and prediction based on remote sensing data. The method includes: acquiring hyperspectral remote sensing data of the earthquake-stricken area after an earthquake, performing cross-modal mutual information calibration, and outputting cross-modal standardized data of the hyperspectral remote sensing image; extracting reflectance vectors corresponding to different spectral wavelengths, constructing a linear spectral mixture model, and calculating the vegetation destruction rate; classifying pixels, calculating the building collapse rate of the earthquake-stricken area, and then using the building collapse rate to calculate the post-earthquake building stability coefficient; calculating the post-earthquake surface geological structure stability coefficient; using the post-earthquake building stability coefficient to evaluate the priority of post-earthquake rescue efforts; and using the post-earthquake surface geological structure stability coefficient to plan the types of transportation vehicles entering the earthquake-stricken area. Based on remote sensing data analysis technology, this method enables accurate prediction and analysis of post-earthquake risks in earthquake-stricken areas, providing reasonable technical support for rescue teams.

Description

Technical Field

[0001] This invention relates to the field of earthquake disaster analysis, specifically to a method, equipment, and medium for earthquake disaster area analysis and prediction based on remote sensing data. Background Technology

[0002] Earthquakes, as highly destructive natural disasters, are characterized by their suddenness and uncertainty. Accurately predicting the impact range, destructive intensity, and secondary disaster risks of earthquake-stricken areas is crucial for reducing casualties and property losses. Remote sensing technology, with its advantages of wide coverage, rapid data acquisition, and non-contact measurement, can be used for timely analysis and prediction of disaster conditions in earthquake-stricken areas.

[0003] Due to the complex terrain features in earthquake-stricken areas, and the low accuracy of existing technologies for identifying terrain features using hyperspectral remote sensing images of these areas, different features only exhibit good spectral characteristics under specific wavelengths. However, spectral characteristics at other wavelengths also possess certain spectral information, leading to the omission of feature information and low accuracy in terrain feature identification. Furthermore, existing technologies are poor at analyzing and predicting secondary disasters caused by aftershocks in the disaster area, hindering the rational allocation of rescue resources and resulting in key areas not receiving timely rescue before secondary disasters occur. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides a method, equipment, and medium for earthquake disaster area analysis and prediction based on remote sensing data. It utilizes remote sensing technology to predict and analyze earthquake disaster areas, solving problems such as low accuracy in identifying ground cover types and poor predictive ability for secondary disasters after earthquakes in existing technologies.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0006] A method for earthquake disaster area analysis and prediction based on remote sensing data is provided, which includes:

[0007] Step S1: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, perform cross-modal mutual information calibration on the hyperspectral remote sensing data, eliminate the bias of cross-modal data, and output cross-modal standardized data of hyperspectral remote sensing images.

[0008] Step S2: Extract reflectance vectors corresponding to different spectral wavelengths from cross-modal normalized data, construct a linear spectral mixture model, and calculate vegetation cover; use vegetation cover to calculate the vegetation damage rate after the earthquake.

[0009] Step S3: Based on the reflectance vectors of the reference end-members corresponding to intact building end-members and collapsed ruin end-members, classify the pixels in the hyperspectral remote sensing image, calculate the building collapse rate in the earthquake-stricken area based on the number of pixels of the collapsed ruins classified, and then use the building collapse rate to calculate the building stability coefficient characterized by the earthquake.

[0010] Step S4: Calculate the comprehensive slope stability coefficient of the area where the earthquake-stricken area is located before the earthquake, and calculate the surface geological structure stability coefficient characterized by the earthquake in combination with the vegetation destruction rate after the earthquake.

[0011] Step S5: Use the post-earthquake building stability coefficient to evaluate the priority of post-earthquake rescue, and use the post-earthquake surface geological structure stability coefficient to plan the type of transportation to enter the earthquake-stricken area.

[0012] Further, step S1 includes:

[0013] Step S11: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, and obtain data features of different modes in the hyperspectral remote sensing data. The data features of different modes include hyperspectral remote sensing images of different spectral wavelengths. The hyperspectral remote sensing images contain spectral features in both spatial and spectral dimensions.

[0014] Step S12: Construct a cross-modal mutual information calibration model, perform cross-modal mutual information calibration on the data features of different modes, eliminate the bias of cross-modal data, and output the calibrated cross-modal spectral feature matrix;

[0015] ;

[0016] in, This represents the spectral feature matrix of the hyperspectral remote sensing image within the modality. The spectral feature matrix to be calibrated, M For the number of modes, These are the numbers for two different modes. Modal The corresponding spectral feature matrix, For mutual information functions, For modality The joint probability density, For modality marginal probability density, For modality Marginal probability density, To calibrate the cross-modal spectral feature matrix, The regularization coefficient is . Norm operations;

[0017] Step S13: Extract spectral features from the calibrated cross-modal spectral feature matrix to obtain cross-modal normalized data of the hyperspectral remote sensing image. The cross-modal normalized data includes different spectral wavelengths. The corresponding reflectivity vector .

[0018] Further, step S2 includes:

[0019] Step S21: Extract different spectral wavelengths from cross-modal normalized data The corresponding reflectivity vector A linear spectral mixture model was constructed to calculate the vegetation component abundance of the vegetation endmembers corresponding to each pixel. Then, the abundance of vegetation components corresponding to each pixel is used. Calculate the vegetation coverage rate in the earthquake-stricken area V ;

[0020] ;

[0021] in, The coordinates of the pixel, These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Abundance of vegetation components and non-vegetation components under certain conditions. These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Reflectivity vector under the given conditions To decompose the residuals, and satisfy the following conditions: , For the number of spectral wavelengths, These represent the height and width of the hyperspectral remote sensing image, respectively. J The number of spectral wavelengths;

[0022] Step S22: Based on the vegetation cover rate represented in the hyperspectral remote sensing image of the earthquake-stricken area before the earthquake... Calculate the vegetation damage rate after the earthquake .

[0023] Further, step S3 includes:

[0024] Step S31: Based on the reference end-member reflectance vectors corresponding to the intact building end-member and the collapsed ruin end-member, calculate the spectral angle between the pixel and the intact building end-member, and the spectral angle between the pixel and the collapsed ruin end-member, and use the spectral angle to classify the pixel;

[0025] ;

[0026] in, These are the reference end-member reflectivity vectors corresponding to the intact building end-member and the collapsed rubble end-member, respectively. Let be the reflectance vector of a pixel. The spectral angles are those of the pixel and the reference endmember spectrum of an intact building, and the reference endmember spectrum of collapsed ruins, respectively. For pixel classification functions, The spectral angle classification threshold is used, and the classification result 1 represents the pixel of the intact building, 2 represents the pixel of the collapsed ruins, and 0 represents the pixel of the non-building area;

[0027] Step S32: Based on the number of pixels of collapsed ruins classified in the hyperspectral remote sensing images of the earthquake-stricken area. N Calculate the building collapse rate in the earthquake-stricken area ;

[0028] ;

[0029] in, s The area of ​​a single pixel. The area of ​​intact buildings as represented in pre-earthquake hyperspectral remote sensing images;

[0030] Step S33: Based on the building collapse rate Calculate the building stability coefficient after an earthquake ;

[0031] ;

[0032] in, The average seismic resistance designed for buildings in earthquake-stricken areas.

[0033] Further, step S4 includes:

[0034] Step S41: Based on the average soil-rock cohesion of the slopes in the earthquake-stricken area collected before the earthquake. c Average length of a single slope slip surface L Average self-weight of a single slope W Average slope of the slope β and average internal friction angle of soil and rock Calculate the comprehensive slope stability coefficient of the earthquake-stricken area before the earthquake. F ;

[0035] ;

[0036] By slope stability coefficient F The slope stability coefficient is used to characterize the slope stability state before an earthquake. F The larger the slope, the more stable the slope; conversely, the smaller the slope, the less stable the slope.

[0037] Step S42: Utilize the vegetation destruction rate after the earthquake Comprehensive stability coefficient of slope before earthquake F Calculate the stability coefficient of the surface geological structure characterized after the earthquake. ;

[0038] ;

[0039] in, The magnitude of the earthquake. To assess the intensity of earthquake energy release, This represents the measured intensity of earthquake energy release. The threshold value for the difference between the measured and assessed seismic energy release intensity. The weighting represents the impact of post-earthquake vegetation destruction rate on the stability assessment of surface geological structures.

[0040] A computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, performs the aforementioned method for earthquake disaster area analysis and prediction based on remote sensing data.

[0041] A terminal device is provided, which is used to call and run a computer program from a memory, so that the terminal device can execute the above-mentioned earthquake disaster area analysis and prediction method based on remote sensing data.

[0042] The beneficial effects of this invention are as follows: By performing cross-modal mutual information calibration on hyperspectral remote sensing data, this invention effectively enriches the spectral feature information in hyperspectral remote sensing images, resulting in higher accuracy in identifying ground features in earthquake-stricken areas. Through high-precision identification of vegetation endmembers in hyperspectral remote sensing images, the vegetation coverage rate of earthquake-stricken areas is calculated. Combined with the comprehensive slope stability before the earthquake, the stability of the surface geology of earthquake-stricken areas is comprehensively analyzed, providing data support for the types of vehicles used by subsequent rescue teams entering the disaster area. Simultaneously, this invention obtains accurate building collapse data through hyperspectral remote sensing data, thereby predicting the stability of buildings after an earthquake, providing data support for rescue teams to select key rescue areas. This invention, based on remote sensing data analysis technology, achieves accurate prediction and analysis of post-earthquake risks in earthquake-stricken areas, providing reasonable technical support for rescue teams. Attached Figure Description

[0043] Figure 1 This is a flowchart of an earthquake disaster area analysis and prediction method based on remote sensing data. Detailed Implementation

[0044] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0045] like Figure 1 As shown, a method for earthquake disaster area analysis and prediction based on remote sensing data includes:

[0046] Step S1: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, perform cross-modal mutual information calibration on the hyperspectral remote sensing data to eliminate cross-modal data bias, and output cross-modal standardized data of the hyperspectral remote sensing image. Step S1 specifically includes:

[0047] Step S11: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, and obtain data features of different modes in the hyperspectral remote sensing data. The data features of different modes include hyperspectral remote sensing images of different spectral wavelengths. The hyperspectral remote sensing images contain spectral features in both spatial and spectral dimensions.

[0048] Step S12: Construct a cross-modal mutual information calibration model, perform cross-modal mutual information calibration on the data features of different modes, eliminate the bias of cross-modal data, and output the calibrated cross-modal spectral feature matrix;

[0049] ;

[0050] in, This represents the spectral feature matrix of the hyperspectral remote sensing image within the modality. The spectral feature matrix to be calibrated, M For the number of modes, These are the numbers for two different modes. Modal The corresponding spectral feature matrix, For mutual information functions, For modality The joint probability density, For modality Marginal probability density, For modality Marginal probability density, To calibrate the cross-modal spectral feature matrix, The regularization coefficient is . Norm operations;

[0051] Step S13: Extract spectral features from the calibrated cross-modal spectral feature matrix to obtain cross-modal normalized data of the hyperspectral remote sensing image. The cross-modal normalized data includes different spectral wavelengths. The corresponding reflectivity vector .

[0052] Hyperspectral remote sensing data is a three-dimensional feature. Each cell in the cube has three attributes: spatial row dimension, column dimension, and spectral dimension. In the spatial dimension, each row and column forms a continuous spectral curve in the spectral dimension. This curve can characterize the spectral features of ground features or targets at the image point.

[0053] Hyperspectral remote sensing data consists of images arranged in order of wavelength, forming dozens or hundreds of images. These images are arranged according to wavelength, with the first image having the shortest wavelength and the last image having the longest wavelength. Each pixel in a hyperspectral remote sensing image contains information from dozens to hundreds of consecutive spectral bands, providing rich spectral details for identifying and distinguishing subtle differences in surface materials.

[0054] This invention achieves accurate calibration of hyperspectral remote sensing data by associating information on the spectral features of different modes, making the correlation of spectral features between different modes stronger and eliminating asynchronous deviations caused by different spectral wavelengths.

[0055] Step S2: Extract reflectance vectors corresponding to different spectral wavelengths from the cross-modal normalized data, construct a linear spectral mixture model, and calculate vegetation cover; use the vegetation cover to calculate the vegetation damage rate after the earthquake. Step S2 specifically includes:

[0056] Step S21: Extract different spectral wavelengths from cross-modal normalized data The corresponding reflectivity vector A linear spectral mixture model was constructed to calculate the vegetation component abundance of the vegetation endmembers corresponding to each pixel. Then, the abundance of vegetation components corresponding to each pixel is used. Calculate the vegetation coverage rate in the earthquake-stricken area V ;

[0057] ;

[0058] in, The coordinates of the pixel, These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Abundance of vegetation components and non-vegetation components under certain conditions. These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Reflectivity vector under the given conditions To decompose the residuals, and satisfy the following conditions: , For the number of spectral wavelengths, These represent the height and width of the hyperspectral remote sensing image, respectively. J The number of spectral wavelengths;

[0059] This invention extracts the spectra of endmembers in densely vegetated areas within hyperspectral remote sensing images of earthquake-stricken areas, calculates the abundance of vegetation components and non-vegetation components in vegetation and non-vegetation endmembers under different spectral wavelength conditions, removes anomalous pixels corresponding to the decomposition residuals, and fills anomalous pixels with the mean abundance of vegetation components from neighboring pixels. Finally, it averages the vegetation component abundance corresponding to all pixels under different spectral wavelength conditions to obtain the vegetation coverage rate of the earthquake-stricken area.

[0060] A pixel represents a point in a hyperspectral remote sensing image of an earthquake-stricken area. It is the smallest unit of data collected by the sensor when scanning ground objects in the earthquake-stricken area. An endmember contains only one type of ground feature information. Generally, pixels are mixed pixels, which include multiple ground features. When performing mixed pixel decomposition, the endmembers included in a pixel can be quantitatively described to obtain the percentage of the area of ​​each type of endmember in that pixel. This percentage is the abundance of the corresponding ground feature endmember.

[0061] Step S22: Based on the vegetation cover rate represented in the hyperspectral remote sensing image of the earthquake-stricken area before the earthquake... Calculate the vegetation damage rate after the earthquake .

[0062] During an earthquake, it can trigger secondary geological disasters such as landslides, mudslides, and debris flows, resulting in broken, fallen, and buried trees in the earthquake-stricken area. Therefore, the vegetation destruction rate is closely related to the slope stability of the earthquake-stricken area. This invention uses the post-earthquake vegetation destruction rate to indirectly reflect the surface geological structure stability of the earthquake-stricken area. The larger the area, the greater the area prone to secondary geological disasters such as landslides, mudslides, and debris flows.

[0063] Step S3: Based on the reflectance vectors of the reference end-members corresponding to intact building end-members and collapsed rubble end-members, the pixels in the hyperspectral remote sensing image are classified. The building collapse rate in the earthquake-stricken area is calculated based on the number of classified collapsed rubble pixels. Then, the building collapse rate is used to calculate the post-earthquake building stability coefficient. Step S3 specifically includes:

[0064] Step S31: Based on the reference end-member reflectance vectors corresponding to the intact building end-member and the collapsed ruin end-member, calculate the spectral angle between the pixel and the intact building end-member, and the spectral angle between the pixel and the collapsed ruin end-member, and use the spectral angle to classify the pixel;

[0065] ;

[0066] in, These are the reference end-member reflectivity vectors corresponding to the intact building end-member and the collapsed rubble end-member, respectively. Let be the reflectance vector of a pixel. The spectral angles are those of the pixel and the reference endmember spectrum of an intact building, and the reference endmember spectrum of collapsed ruins, respectively. For pixel classification functions, The spectral angle classification threshold is used, and the classification result 1 represents the pixel of the intact building, 2 represents the pixel of the collapsed ruins, and 0 represents the pixel of the non-building area;

[0067] Step S32: Based on the number of pixels of collapsed ruins classified in the hyperspectral remote sensing images of the earthquake-stricken area. N Calculate the building collapse rate in the earthquake-stricken area ;

[0068] ;

[0069] in, s The area of ​​a single pixel. The area of ​​intact buildings as represented in pre-earthquake hyperspectral remote sensing images;

[0070] Step S33: Based on the building collapse rate Calculate the building stability coefficient after an earthquake ;

[0071] ;

[0072] in, The average seismic resistance designed for buildings in earthquake-stricken areas.

[0073] Calculate the stability coefficient of a building During the process, This is the proportional term representing the magnitude of the actual earthquake exceeding the average seismic resistance of the building's design. This indicates that the earthquake magnitude was too large, exceeding the building's earthquake-resistant capacity. Such a large magnitude earthquake weakens the stability of the building's conventional seismic resistance and increases the likelihood of collapse. Therefore, by introducing a proportional term to amplify the building's stability coefficient after a large-magnitude earthquake, the error impact of large-magnitude earthquakes on building stability assessments is reduced. The building collapse rate in this embodiment... satisfy The variables within the exp function change with the building collapse rate. The building stability coefficient increases with the increase of [something]. The larger the value, the better the stability of the building after an earthquake, and vice versa.

[0074] Step S4: Calculate the comprehensive slope stability coefficient of the earthquake-stricken area before the earthquake, and combine it with the post-earthquake vegetation destruction rate to calculate the surface geological structure stability coefficient characterized by the earthquake. Step S4 specifically includes:

[0075] Step S41: Based on the average soil-rock cohesion of the slopes in the earthquake-stricken area collected before the earthquake. c Average length of a single slope slip surface L Average self-weight of a single slope W Average slope of the slope β and average internal friction angle of soil and rock Calculate the comprehensive slope stability coefficient of the earthquake-stricken area before the earthquake. F ;

[0076] ;

[0077] By slope stability coefficient F The slope stability coefficient is used to characterize the slope stability state before an earthquake. F The larger the slope, the more stable the slope; conversely, the smaller the slope, the less stable the slope.

[0078] Step S42: Utilize the vegetation destruction rate after the earthquake Comprehensive stability coefficient of slope before earthquake F Calculate the stability coefficient of the surface geological structure characterized after the earthquake. ;

[0079] ;

[0080] in, The magnitude of the earthquake. To assess the intensity of earthquake energy release, This represents the measured intensity of earthquake energy release. The threshold value for the difference between the measured and assessed seismic energy release intensity. The weighting represents the impact of post-earthquake vegetation destruction rate on the stability assessment of surface geological structures.

[0081] This invention comprehensively calculates the surface geological structure stability coefficient by combining the post-earthquake vegetation destruction rate and the slope comprehensive stability coefficient measured and calculated before the earthquake, thereby assessing the stability of the surface geological structure in the earthquake zone. The larger the value, the more stable it is; conversely, the smaller the value, the less stable it is. This embodiment uses... The actual rate of vegetation destruction after an earthquake It can better reflect the stability of the surface geological structure. This is an amplification term based on the intensity of earthquake energy release. When the actual earthquake energy release intensity is greater than the estimated earthquake energy release intensity, the amplification term... The rate of vegetation destruction By magnifying the data, the effects of earthquakes on vegetation in earthquake-stricken areas can be amplified, thereby providing a more accurate reflection of the stability of the surface geological structure.

[0082] Step S5: Evaluate post-earthquake rescue priorities using building stability coefficients characterized after the earthquake, and plan the types of transportation vehicles entering the earthquake-stricken area using surface geological structure stability coefficients characterized after the earthquake. Specifically:

[0083] Set a threshold for the building stability coefficient characterized after an earthquake. and the threshold of surface geological structure stability coefficient ;

[0084] like If the earthquake-stricken area is determined to have poor post-earthquake stability, there is a high probability that aftershocks will cause buildings to collapse again, and the post-earthquake rescue is of high priority. It is necessary to promptly arrange corresponding rescue personnel to the earthquake-stricken area for rescue.

[0085] like If the earthquake-stricken area is deemed to have good post-earthquake stability and a low probability of buildings collapsing again due to aftershocks, then post-earthquake rescue priority is the lowest; core rescue forces can be prioritized and deployed to areas with high post-earthquake rescue priority.

[0086] like If the slopes in the earthquake-stricken area are stable, then land transportation can be used to enter the earthquake-stricken area for rescue operations.

[0087] like If the slope in the earthquake-stricken area is unstable, it is determined that air transportation is required to enter the earthquake-stricken area for rescue, and land transportation is not allowed to enter the earthquake-stricken area for rescue.

[0088] A computer-readable storage medium storing a computer program, which, when executed by a processor, performs the aforementioned method for earthquake disaster area analysis and prediction based on remote sensing data.

[0089] A terminal device is provided for calling and running a computer program from a memory, enabling the terminal device to execute the aforementioned earthquake disaster area analysis and prediction method based on remote sensing data.

[0090] This invention effectively enriches the spectral feature information in hyperspectral remote sensing images by performing cross-modal mutual information calibration on hyperspectral remote sensing data, resulting in higher accuracy in identifying ground features in earthquake-stricken areas. Through high-precision identification of vegetation end-members in hyperspectral remote sensing images, the vegetation coverage rate of earthquake-stricken areas is calculated. Combined with the comprehensive slope stability before the earthquake, the stability of the surface geology of earthquake-stricken areas is comprehensively analyzed, providing data support for the types of vehicles used by subsequent rescue teams to enter the disaster area. Simultaneously, this invention obtains accurate building collapse data through hyperspectral remote sensing data, thereby predicting the stability of buildings after an earthquake, providing data support for rescue teams to select key rescue areas. This invention achieves accurate prediction and analysis of post-earthquake risks in earthquake-stricken areas based on remote sensing data analysis technology, providing reasonable technical support for rescue teams.

Claims

1. A method for earthquake disaster area analysis and prediction based on remote sensing data, characterized in that, include: Step S1: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, perform cross-modal mutual information calibration on the hyperspectral remote sensing data, eliminate the bias of cross-modal data, and output cross-modal standardized data of hyperspectral remote sensing images. Step S2: Extract reflectance vectors corresponding to different spectral wavelengths from cross-modal normalized data, construct a linear spectral mixture model, and calculate vegetation cover; use vegetation cover to calculate the vegetation damage rate after the earthquake. Step S3: Based on the reflectance vectors of the reference end-members corresponding to intact building end-members and collapsed ruin end-members, classify the pixels in the hyperspectral remote sensing image, calculate the building collapse rate in the earthquake-stricken area based on the number of pixels of the collapsed ruins classified, and then use the building collapse rate to calculate the building stability coefficient characterized by the earthquake. Step S4: Calculate the comprehensive slope stability coefficient of the area where the earthquake-stricken area is located before the earthquake, and calculate the surface geological structure stability coefficient characterized by the earthquake in combination with the vegetation destruction rate after the earthquake. Step S5: Use the post-earthquake building stability coefficient to evaluate the priority of post-earthquake rescue, and use the post-earthquake surface geological structure stability coefficient to plan the type of transportation to enter the earthquake-stricken area.

2. The earthquake disaster area analysis and prediction method based on remote sensing data according to claim 1, characterized in that, Step S1 includes: Step S11: Collect hyperspectral remote sensing data of the earthquake-stricken area after the earthquake, and obtain data features of different modes in the hyperspectral remote sensing data. The data features of different modes include hyperspectral remote sensing images of different spectral wavelengths. The hyperspectral remote sensing images contain spectral features in both spatial and spectral dimensions. Step S12: Construct a cross-modal mutual information calibration model, perform cross-modal mutual information calibration on the data features of different modes, eliminate the bias of cross-modal data, and output the calibrated cross-modal spectral feature matrix; ； in, This represents the spectral feature matrix of the hyperspectral remote sensing image within the modality. The spectral feature matrix to be calibrated, M For the number of modes, These are the numbers for two different modes. Modal The corresponding spectral feature matrix, For mutual information functions, For modality The joint probability density, For modality Marginal probability density, For modality Marginal probability density, To calibrate the cross-modal spectral feature matrix, The regularization coefficient is . Norm operations; Step S13: Extract spectral features from the calibrated cross-modal spectral feature matrix to obtain cross-modal normalized data of the hyperspectral remote sensing image. The cross-modal normalized data includes different spectral wavelengths. The corresponding reflectivity vector .

3. The earthquake disaster area analysis and prediction method based on remote sensing data according to claim 2, characterized in that, Step S2 includes: Step S21: Extract different spectral wavelengths from cross-modal normalized data The corresponding reflectivity vector A linear spectral mixture model was constructed to calculate the vegetation component abundance of the vegetation endmembers corresponding to each pixel. Then, the abundance of vegetation components corresponding to each pixel is used. Calculate the vegetation coverage rate in the earthquake-stricken area V ; ； in, The coordinates of the pixel, These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Abundance of vegetation components and non-vegetation components under certain conditions. These represent vegetation endmembers and non-vegetation endmembers at different spectral wavelengths on a pixel. Reflectivity vector under the given conditions To decompose the residuals, and satisfy the following conditions: , For the number of spectral wavelengths, These represent the height and width of the hyperspectral remote sensing image, respectively. J The number of spectral wavelengths; Step S22: Based on the vegetation cover rate represented in the hyperspectral remote sensing image of the earthquake-stricken area before the earthquake. Calculate the vegetation damage rate after the earthquake .

4. The earthquake disaster area analysis and prediction method based on remote sensing data according to claim 3, characterized in that, Step S3 includes: Step S31: Based on the reference end-member reflectance vectors corresponding to the intact building end-member and the collapsed ruin end-member, calculate the spectral angle between the pixel and the intact building end-member, and the spectral angle between the pixel and the collapsed ruin end-member, and use the spectral angle to classify the pixel; ； in, These are the reference end-member reflectivity vectors corresponding to the intact building end-member and the collapsed rubble end-member, respectively. Let be the reflectance vector of a pixel. The spectral angles are those of the pixel and the reference endmember spectrum of an intact building, and the reference endmember spectrum of collapsed ruins, respectively. For pixel classification functions, The spectral angle classification threshold is used, and the classification result 1 represents the pixel of the intact building, 2 represents the pixel of the collapsed ruins, and 0 represents the pixel of the non-building area; Step S32: Based on the number of pixels of collapsed ruins classified in the hyperspectral remote sensing images of the earthquake-stricken area. N Calculate the building collapse rate in the earthquake-stricken area ; ； in, s The area of ​​a single pixel. The area of ​​intact buildings as represented in the hyperspectral remote sensing image before the earthquake; Step S33: Based on the building collapse rate Calculate the building stability coefficient after an earthquake ; ； in, The average seismic resistance designed for buildings in earthquake-stricken areas. The magnitude of the earthquake that actually occurred in the earthquake-stricken area.

5. The earthquake disaster area analysis and prediction method based on remote sensing data according to claim 4, characterized in that, Step S4 includes: Step S41: Based on the average soil-rock cohesion of the slopes in the earthquake-stricken area collected before the earthquake. c Average length of a single slope slip surface L Average self-weight of a single slope W Average slope of the slope β and average internal friction angle of soil and rock Calculate the comprehensive slope stability coefficient of the earthquake-stricken area before the earthquake. F ; ； By slope stability coefficient F The slope stability coefficient is used to characterize the slope stability state before an earthquake. F The larger the slope, the more stable the slope; conversely, the smaller the slope, the less stable the slope. Step S42: Utilize the vegetation destruction rate after the earthquake Comprehensive stability coefficient of slope before earthquake F Calculate the stability coefficient of the surface geological structure characterized after the earthquake. ; ； in, The magnitude of the earthquake. To assess the intensity of earthquake energy release, This represents the measured intensity of earthquake energy release. The threshold value for the difference between the measured and assessed seismic energy release intensity. The weighting represents the impact of post-earthquake vegetation destruction rate on the stability assessment of surface geological structures.

6. A computer-readable storage medium, characterized in that, The system contains a computer program that, when executed by a processor, performs the earthquake disaster area analysis and prediction method based on remote sensing data as described in any one of claims 1-5.

7. A terminal device, characterized in that, The terminal device is used to call and run a computer program from a memory, so that the terminal device performs the earthquake disaster area analysis and prediction method based on remote sensing data as described in any one of claims 1-5.

Images