Disaster area loss analysis method and device based on unmanned aerial vehicle and medium

Abstract

The invention relates to a disaster detection technology, and discloses a disaster area loss analysis method and device based on an unmanned aerial vehicle and a computer readable storage medium, and the method comprises the steps: carrying out the cruise shooting of a to-be-detected disaster area through employing the unmanned aerial vehicle, and obtaining real-time laser data, real-time radar data and real-time aerial photographing data, performing point cloud difference analysis, point cloud repair and three-dimensional modeling on the real-time laser data and the historical laser data to obtain a variable model group; and performing differential interference analysis on the real-time radar data and the historical radar data to obtain an interference phase diagram group, performing parallax analysis on the real-time aerial photography data and the historical aerial photography data, and performing comprehensive loss analysis on the to-be-detected disaster area according to the change diagram block group, the interference phase diagram group and the change diagram block group to obtain a damage result of the disaster area. The method can be applied to the field of disaster detection, can achieve triple-contrast disaster detection according to the three-dimensional model, the radar data and the two-dimensional image, and improves the efficiency of disaster detection.

Description

Technical Field

[0001] This invention relates to the field of disaster detection technology, and in particular to a method, apparatus and computer-readable storage medium for disaster area loss analysis based on unmanned aerial vehicles (UAVs). Background Technology

[0002] With the rapid development of drone technology, drones are becoming increasingly common in disaster management and rescue. Drones have advantages such as flexibility, rapid deployment, low cost, and high-resolution imaging, making them an important tool for disaster response.

[0003] Existing UAV-based disaster loss analysis methods are mostly based on global identification. They use UAVs to acquire images of disaster areas and perform disaster detection and identification on the images to achieve disaster loss analysis. In practical applications, for large-scale disasters, the amount of data acquired is large and the calculation steps are relatively complex. The data acquired by UAVs is affected by weather and environment and has a certain amount of data error, which may lead to low efficiency in loss analysis. Summary of the Invention

[0004] This invention provides a method, apparatus, and computer-readable storage medium for disaster area loss analysis based on unmanned aerial vehicles (UAVs), with the main purpose of solving the problem of low efficiency in loss analysis.

[0005] To achieve the above objectives, this invention provides a disaster area loss analysis method based on unmanned aerial vehicles (UAVs), comprising:

[0006] Drones were used to patrol and photograph the disaster area under test, obtaining real-time laser data, real-time radar data, and real-time aerial photography data. Historical laser data, historical radar data, and historical aerial photography data were extracted from historical patrol data.

[0007] Point cloud difference analysis is performed on the real-time laser data and the historical laser data to obtain a variable point cloud data group. Point cloud repair and 3D modeling are performed on the variable point cloud data group to obtain a variable model group.

[0008] Differential interferometry analysis is performed on the real-time radar data and the historical radar data to obtain an interferometric phase map group. Parallax analysis is performed on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometry analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: using the real-time radar data to perform radar registration on the historical radar data to obtain registration and matching data; and using the following phase decomposition algorithm to calculate the decomposition deformation between the real-time radar data and the registered radar data:

[0009]

[0010] Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map.

[0011] Model loss identification is performed on the variable model group to obtain the laser disaster data group; phase loss identification is performed on the interferometric phase map group to obtain the radar disaster data group; and image loss identification is performed on the variable map block group to obtain the aerial disaster data group.

[0012] A comprehensive loss analysis is performed on the disaster area to be tested based on the laser disaster data set, the radar disaster data set, and the aerial disaster data set to obtain the damage results of the disaster area.

[0013] Optionally, the use of drones to cruise and photograph the disaster area to obtain real-time laser data, real-time radar data, and real-time aerial photography data includes:

[0014] The drone is used to conduct a comprehensive patrol of the disaster area under test according to a preset patrol route.

[0015] The laser scanning equipment of the UAV is used to perform laser scanning on the disaster area to be measured during the coverage patrol to obtain laser patrol data;

[0016] The radar equipment of the UAV is used to perform radar scanning on the disaster area to be measured during the coverage patrol to obtain radar patrol data.

[0017] The camera equipment of the drone is used to take images of the disaster area to be tested during the coverage patrol, and aerial patrol data is obtained.

[0018] Based on the stated cruise route, the laser cruise data is stitched together to form real-time laser data, the radar cruise data is stitched together to form real-time radar data, and the aerial photography cruise data is stitched together to form real-time aerial photography data.

[0019] Optionally, the step of performing point cloud difference analysis on the real-time laser data and the historical laser data to obtain a variable point cloud data set includes:

[0020] The historical laser data is registered using the real-time laser data to obtain registered laser data.

[0021] The registered laser data and the real-time laser data are compared in elevation to obtain elevation difference data;

[0022] Real-time point cloud data is filtered from the real-time laser data based on the elevation difference data;

[0023] Historical point cloud data are filtered from the registered laser data based on the elevation difference data;

[0024] The real-time point cloud data and the historical point cloud data are aggregated into a variable point cloud data group.

[0025] Optionally, the step of using the real-time laser data to perform point cloud registration on the historical laser data to obtain registered laser data includes:

[0026] Point cloud matching is performed on the real-time laser data and the historical laser data to obtain a matching point cloud group;

[0027] Real-time matching point cloud and historical matching point cloud are extracted from the matching point cloud group respectively;

[0028] The covariance centroid point cloud is calculated using the following centroid matching algorithm based on the real-time matching point cloud and the historical matching point cloud.

[0029]

[0030] Wherein, H refers to the covariance centroid point cloud, i refers to the point cloud index, N refers to the total number of points in the real-time matched point cloud, and the total number of points in the real-time matched point cloud is equal to the total number of points in the historical matched point cloud, p i This refers to the point cloud with point cloud index i in the real-time matched point cloud, q i It refers to the point cloud with point cloud index i in the historical matching point cloud, where T is the transpose symbol;

[0031] Singular value decomposition is performed on the covariance centroid point cloud to obtain the rotation matrix;

[0032] The translation vector is calculated based on the rotation matrix, the real-time matching point cloud, and the historical matching point cloud.

[0033] The historical laser data is registered using the translation vector and the rotation matrix to obtain the registered laser data.

[0034] Optionally, the step of performing point cloud repair and 3D modeling on the changed point cloud data group to obtain a changed model group includes:

[0035] Point cloud filtering is performed on the variable point cloud data group to obtain a noise-reduced point cloud data group;

[0036] Generate a point cloud feature data group corresponding to the denoised point cloud data group, and use the following point cloud distance algorithm to segment the point cloud feature data group to obtain a segmented point cloud feature cluster group:

[0037]

[0038] Where D refers to the point cloud feature distance of the point cloud distance algorithm, α and β are preset adversarial coefficients, · is the dot product symbol, * is the cross product symbol, p refers to point cloud feature p in the point cloud feature data set, q refers to point cloud feature q in the point cloud feature data set, T is the transpose symbol, cov() is the covariance symbol, cov(p,p) refers to the covariance of point cloud feature p, cov(p,q) refers to the covariance between point cloud feature p and point cloud feature q, cov(q,p) refers to the covariance between point cloud feature q and point cloud feature p, and cov(q,q) refers to the covariance of point cloud feature q.

[0039] Based on the segmented point cloud cluster feature group, the noise-reduced point cloud data group is triangulated into a segmented triangular model group.

[0040] Extract the set of triangle vertices and the set of triangle faces from the segmented triangle model set;

[0041] Based on the set of triangle vertices and the set of triangle faces, the segmented triangle model group is reconstructed to obtain a variable model group.

[0042] Optionally, the step of performing parallax analysis on the real-time aerial data and the historical aerial data to obtain a group of changing map tiles includes:

[0043] The historical aerial photography data is image registered using the real-time aerial photography data to obtain registered aerial photography data.

[0044] Grayscale pixel matching is performed on the real-time aerial photography data and the registered aerial photography data to obtain the parallax pixel region;

[0045] Real-time changing patches are selected from the real-time aerial photography data using the parallax pixel region, and historical changing patches are selected from the registered aerial photography data using the parallax pixel region.

[0046] The real-time changing tiles and the historical changing tiles are combined into a changing tile group.

[0047] Optionally, the step of identifying model loss in the variable model group to obtain the laser disaster data group includes:

[0048] The variable model group is divided into a real-time model group and a historical model group;

[0049] Each real-time model in the real-time model group is selected as a target real-time model, and model features are extracted from the target real-time model to obtain the target real-time object features;

[0050] The historical model corresponding to the target real-time model is selected from the historical model group as the target historical model, and the model features are extracted from the target historical model to obtain the target historical object features;

[0051] Based on the characteristics of the target historical object, semantic recognition is performed on the target historical model to obtain the target model semantics;

[0052] The target model loss is calculated based on the target historical object features and the target real-time object features;

[0053] The target model semantics and the target model loss are aggregated into laser disaster data, and all the laser disaster data are aggregated into a laser disaster data group.

[0054] Optionally, the step of performing phase loss identification on the interferometric phase map group to obtain the radar disaster data group includes:

[0055] The interferometric phase map group is split into a real-time phase map and a historical phase map;

[0056] The historical phase map is segmented into blocks to obtain historical phase map block groups;

[0057] The real-time phase map is divided into real-time phase map groups based on the historical phase map group;

[0058] Semantic recognition is performed on the historical phase map block group to obtain the phase semantic group;

[0059] A phase loss group is calculated based on the real-time phase block group and the historical phase block group;

[0060] A radar disaster data group is generated based on the phase semantic group and the phase loss group.

[0061] To address the aforementioned problems, the present invention also provides a disaster area loss analysis device based on unmanned aerial vehicles (UAVs), the device comprising:

[0062] The data acquisition module is used to use drones to cruise and photograph the disaster area under test, obtain real-time laser data, real-time radar data and real-time aerial photography data, and extract historical laser data, historical radar data and historical aerial photography data from historical cruise data.

[0063] The 3D modeling module is used to perform point cloud difference analysis on the real-time laser data and the historical laser data to obtain a variable point cloud data group, and to perform point cloud repair and 3D modeling on the variable point cloud data group to obtain a variable model group.

[0064] The differential interferometry module is used to perform differential interferometric analysis on the real-time radar data and the historical radar data to obtain an interferometric phase map group, and to perform parallax analysis on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometric analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: performing radar registration on the historical radar data using the real-time radar data to obtain registration and matching data; and calculating the decomposition deformation between the real-time radar data and the registered radar data using the following phase decomposition algorithm:

[0065]

[0066] Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map.

[0067] The semantic recognition module is used to identify model loss in the variable model group to obtain laser disaster data group, identify phase loss in the interferometric phase map group to obtain radar disaster data group, and identify image loss in the variable map block group to obtain aerial disaster data group.

[0068] The disaster analysis module is used to perform a comprehensive loss analysis on the disaster area to be tested based on the laser disaster data group, the radar disaster data group, and the aerial disaster data group, and to obtain the damage results of the disaster area.

[0069] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the aforementioned UAV-based disaster area loss analysis method.

[0070] This invention utilizes unmanned aerial vehicles (UAVs) to cruise and photograph disaster areas, obtaining real-time laser data, real-time radar data, and real-time aerial photography data. UAVs can be used to collect real-time 3D information, 2D image information, and radar information of the disaster area, facilitating subsequent disaster analysis. By performing point cloud difference analysis on the real-time laser data and historical laser data, a variable point cloud data group is obtained. This allows for direct filtering of point cloud data that has changed over time, reducing the amount of data required for disaster analysis and improving its efficiency. By performing point cloud repair and 3D modeling on the variable point cloud data group, a variable model group is obtained, enabling the repair of point cloud data and facilitating more accurate disaster analysis. By performing differential interferometry analysis on the real-time radar data and historical radar data, an interferometric phase map group is obtained. By performing parallax analysis on the real-time aerial photography data and historical aerial photography data, a variable patch group is obtained. This allows for direct filtering of radar images showing differences before and after the disaster, as well as filtering of aerial images showing differences before and after the disaster, reducing the order of magnitude of data analysis and improving the efficiency of disaster analysis.

[0071] By identifying model loss in the changing model group, a laser-damaged data group is obtained; by identifying phase loss in the interferometric phase map group, a radar-damaged data group is obtained; and by identifying image loss in the changing patch group, an aerial-damaged data group is obtained. Triple semantic recognition can be achieved based on model data, radar data, and image data, improving the accuracy of surface information extraction in disaster areas. Furthermore, the magnitude of the disaster impact is determined by the differences between model data, radar data, and image data before and after the disaster, improving the accuracy of disaster area loss analysis. By performing a comprehensive loss analysis on the disaster area based on the laser-damaged data group, the radar-damaged data group, and the aerial-damaged data group, the damage results of the disaster area are obtained. This method combines three-dimensional model data, radar data, and image data for triple disaster analysis, improving the accuracy of disaster area loss analysis. Therefore, the UAV-based disaster area loss analysis method, device, and computer-readable storage medium proposed in this invention can solve the problem of low efficiency in loss analysis. Attached Figure Description

[0072] Figure 1 This is a flowchart illustrating a method for disaster area loss analysis based on unmanned aerial vehicles (UAVs) according to an embodiment of the present invention.

[0073] Figure 2 This is a schematic diagram of the process for extracting variable point cloud data groups according to an embodiment of the present invention;

[0074] Figure 3 This is a schematic diagram of the process for generating a group of variable blocks according to an embodiment of the present invention;

[0075] Figure 4 This is a device architecture diagram of a disaster area loss analysis device based on a drone, provided as an embodiment of the present invention.

[0076] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0077] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0078] This application provides a method for disaster area loss analysis based on unmanned aerial vehicles (UAVs). The execution entity of the UAV-based disaster area loss analysis method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the UAV-based disaster area loss analysis method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0079] Reference Figure 1 The diagram shown is a flowchart illustrating a disaster area loss analysis method based on unmanned aerial vehicles (UAVs) according to an embodiment of the present invention. In this embodiment, the UAV-based disaster area loss analysis method includes:

[0080] S1. Use drones to cruise and photograph the disaster area under test to obtain real-time laser data, real-time radar data and real-time aerial photography data. Extract historical laser data, historical radar data and historical aerial photography data from historical cruise data.

[0081] In this embodiment of the invention, the real-time laser data is point cloud data obtained by using the laser scanning device of the UAV to cruise and photograph the disaster area to be measured. The real-time laser data can be data obtained by LiDAR technology or three-dimensional laser scanning technology. The real-time radar data is radar echo signal obtained by using the radar scanning device of the UAV to cruise and photograph the disaster area to be measured. The historical aerial photography data is two-dimensional image data obtained by using a high-resolution camera to cruise and photograph the disaster area to be measured.

[0082] In this embodiment of the invention, the step of using a drone to cruise and photograph the disaster area under test to obtain real-time laser data, real-time radar data, and real-time aerial photography data includes:

[0083] The drone is used to conduct a comprehensive patrol of the disaster area under test according to a preset patrol route.

[0084] The laser scanning equipment of the UAV is used to perform laser scanning on the disaster area to be measured during the coverage patrol to obtain laser patrol data;

[0085] The radar equipment of the UAV is used to perform radar scanning on the disaster area to be measured during the coverage patrol to obtain radar patrol data.

[0086] The camera equipment of the drone is used to take images of the disaster area to be tested during the coverage patrol, and aerial patrol data is obtained.

[0087] Based on the stated cruise route, the laser cruise data is stitched together to form real-time laser data, the radar cruise data is stitched together to form real-time radar data, and the aerial photography cruise data is stitched together to form real-time aerial photography data.

[0088] In this embodiment of the invention, the cruise route refers to the route along which the UAV can cover and scan the disaster area to be tested. The laser scanning device can be a laser emitter or a laser scanner, the radar device can be a microwave radar or a millimeter-wave radar, and the camera device can be a Flight Eye FE 320 or an FCB-EV9500M.

[0089] In detail, stitching the laser cruise data into real-time laser data according to the cruise route means determining the position information corresponding to each laser data in the laser cruise data according to the cruise route, and overlapping and synthesizing the laser cruise data according to the position information to obtain real-time aurora data.

[0090] In this embodiment of the invention, the historical cruise data refers to the data obtained by the UAV during past periods of cruise photography of the disaster area to be tested; the historical laser data refers to real-time laser data within past time periods; the historical radar data refers to real-time radar data within past time periods; and the historical aerial photography data refers to real-time aerial photography data within past time periods.

[0091] In this embodiment of the invention, by using a drone to cruise and photograph the disaster area under test, real-time laser data, real-time radar data, and real-time aerial photography data can be obtained. The drone can be used to collect real-time three-dimensional information, two-dimensional image information, and radar information of the disaster area under test, thereby facilitating subsequent disaster analysis.

[0092] S2. Perform point cloud difference analysis on the real-time laser data and the historical laser data to obtain a variable point cloud data group. Perform point cloud repair and 3D modeling on the variable point cloud data group to obtain a variable model group.

[0093] In this embodiment of the invention, the variable point cloud data group includes real-time point cloud data and historical point cloud data. The real-time point cloud data is the point cloud data of the portion of real-time laser data that has changed when the real-time laser data and the historical laser data are compared and matched. The historical point cloud data is the point cloud data of the portion of historical laser data that has changed when the real-time laser data and the historical laser data are compared and matched.

[0094] In this embodiment of the invention, reference is made to Figure 2 As shown, the point cloud difference analysis performed on the real-time laser data and the historical laser data to obtain a variable point cloud data set includes:

[0095] S21. Use the real-time laser data to perform point cloud registration on the historical laser data to obtain registered laser data;

[0096] S22. Compare the registered laser data and the real-time laser data in terms of elevation to obtain elevation difference data;

[0097] S23. Based on the elevation difference data, filter out real-time point cloud data from the real-time laser data;

[0098] S24. Based on the elevation difference data, historical point cloud data is selected from the registration laser data;

[0099] S25. The real-time point cloud data and the historical point cloud data are aggregated into a variable point cloud data group.

[0100] Specifically, the point cloud registration refers to calibrating the point cloud data of the real-time laser data and the point cloud data of the historical laser data into the same coordinate system, thereby eliminating the mismatch problem of point cloud data caused by path error.

[0101] Specifically, the step of using the real-time laser data to perform point cloud registration on the historical laser data to obtain registered laser data includes:

[0102] Point cloud matching is performed on the real-time laser data and the historical laser data to obtain a matching point cloud group;

[0103] Real-time matching point cloud and historical matching point cloud are extracted from the matching point cloud group respectively;

[0104] The covariance centroid point cloud is calculated using the following centroid matching algorithm based on the real-time matching point cloud and the historical matching point cloud.

[0105]

[0106] Wherein, H refers to the covariance centroid point cloud, i refers to the point cloud index, N refers to the total number of points in the real-time matched point cloud, and the total number of points in the real-time matched point cloud is equal to the total number of points in the historical matched point cloud, p i This refers to the point cloud with point cloud index i in the real-time matched point cloud, q i It refers to the point cloud with point cloud index i in the historical matching point cloud, where T is the transpose symbol;

[0107] Singular value decomposition is performed on the covariance centroid point cloud to obtain the rotation matrix;

[0108] The translation vector is calculated based on the rotation matrix, the real-time matching point cloud, and the historical matching point cloud.

[0109] The historical laser data is registered using the translation vector and the rotation matrix to obtain the registered laser data.

[0110] In detail, the point cloud matching of the real-time laser data and the historical laser data to obtain a matching point cloud group refers to matching the point cloud data at the same position in the real-time laser data and the historical laser data, and aggregating the corresponding point clouds into a matching point cloud group. The matching point cloud group can be obtained by using a feature matching algorithm to perform point cloud matching on the real-time laser data and the historical laser data.

[0111] Specifically, the real-time matching point cloud refers to a portion of the point cloud belonging to the real-time laser data in the matching point cloud group, and the historical matching point cloud refers to a portion of the point cloud belonging to the historical laser data in the matching point cloud group, wherein the real-time matching point cloud and the historical matching point cloud correspond one-to-one.

[0112] In detail, by using the centroid matching algorithm to calculate the covariance centroid point cloud based on the real-time matched point cloud and the historical matched point cloud, data distribution information can be provided, the main direction of the point cloud data can be calculated, and thus the point cloud data can be aligned in the best way.

[0113] Specifically, the Singular Value Decomposition (SVD) is a linear algebraic method that decomposes a matrix into a specific form. The step of performing SVD on the covariance centroid point cloud to obtain a rotation matrix includes performing SVD on the covariance centroid point cloud to obtain an orthogonal matrix and a diagonal matrix, and then calculating the rotation matrix based on the orthogonal matrix.

[0114] In detail, calculating the translation vector based on the rotation matrix, the real-time matched point cloud, and the historical matched point cloud means subtracting the product of the rotation matrix and the historical matched point cloud from the real-time matched point cloud to obtain the translation vector. Registering the historical laser data using the translation vector and the rotation matrix to obtain registered laser data means calculating the point cloud transformation matrix based on the translation vector and the rotation matrix, and multiplying the historical laser data by the point cloud transformation matrix to obtain registered aurora data.

[0115] In this embodiment of the invention, the step of performing point cloud repair and 3D modeling on the changed point cloud data group to obtain a changed model group includes:

[0116] Point cloud filtering is performed on the variable point cloud data group to obtain a noise-reduced point cloud data group;

[0117] Generate a point cloud feature data group corresponding to the denoised point cloud data group, and use the following point cloud distance algorithm to segment the point cloud feature data group to obtain a segmented point cloud feature cluster group:

[0118]

[0119] Where D refers to the point cloud feature distance of the point cloud distance algorithm, α and β are preset adversarial coefficients, · is the dot product symbol, * is the cross product symbol, p refers to point cloud feature p in the point cloud feature data set, q refers to point cloud feature q in the point cloud feature data set, T is the transpose symbol, cov() is the covariance symbol, cov(p,p) refers to the covariance of point cloud feature p, cov(p,q) refers to the covariance between point cloud feature p and point cloud feature q, cov(q,p) refers to the covariance between point cloud feature q and point cloud feature p, and cov(q,q) refers to the covariance of point cloud feature q.

[0120] Based on the segmented point cloud cluster feature group, the noise-reduced point cloud data group is triangulated into a segmented triangular model group.

[0121] Extract the set of triangle vertices and the set of triangle faces from the segmented triangle model set;

[0122] Based on the set of triangle vertices and the set of triangle faces, the segmented triangle model group is reconstructed to obtain a variable model group.

[0123] In detail, generating the point cloud feature data group corresponding to the denoised point cloud data group refers to calculating the voxel features and curvature features of each point cloud in the denoised point cloud data group one by one, combining the voxel features and curvature features into point cloud features, and combining all the point cloud features into a point cloud feature data group.

[0124] In this embodiment of the invention, the k-means clustering algorithm or the DBSCAN clustering algorithm can be used to segment the point cloud feature data group according to the point cloud distance algorithm to obtain segmented point cloud feature clusters. By using the point cloud distance algorithm to segment the point cloud feature data group to obtain segmented point cloud feature clusters, feature clustering can be achieved by combining the distribution of point cloud feature data, thereby improving the accuracy of point cloud segmentation.

[0125] In detail, the variable point cloud data set can be filtered using outlier filtering algorithms, voxel filtering algorithms, or Gaussian filtering algorithms to obtain a denoised point cloud data set. The denoised point cloud data set can then be triangulated into a segmented triangular model set using Delaunay triangulation, Marching Cubes algorithm, or Ball Pivoting algorithm.

[0126] In this embodiment of the invention, by performing point cloud difference analysis on the real-time laser data and the historical laser data, a variable point cloud data group is obtained. The point cloud data that has changed before and after can be directly selected, which reduces the amount of data for disaster analysis and improves the efficiency of disaster analysis. By performing point cloud repair and 3D modeling on the variable point cloud data group, a variable model group is obtained, which can realize the repair of point cloud data, thereby facilitating more accurate disaster analysis.

[0127] S3. Perform differential interferometry analysis on the real-time radar data and the historical radar data to obtain an interferometric phase map group; perform parallax analysis on the real-time aerial photography data and the historical aerial photography data to obtain a variable map block group.

[0128] In this embodiment of the invention, the interferometric phase map group includes a real-time phase map and a historical phase map. The real-time phase map is the radar phase map corresponding to the part of the real-time radar data that changed before and after the disaster, and the historical phase map is the radar phase map corresponding to the part of the historical radar data that changed before and after the disaster.

[0129] In this embodiment of the invention, the step of performing differential interferometric analysis on the real-time radar data and the historical radar data to obtain an interferometric phase map group includes:

[0130] The historical radar data is registered using the real-time radar data to obtain registration and matching data.

[0131] The decomposition deformation between the real-time radar data and the registered radar data is calculated using the following phase decomposition algorithm:

[0132]

[0133] Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registration radar data, and θ is the wavelength of the radar wave in the real-time radar data, and the wavelength of the radar wave in the real-time radar data is equal to the wavelength of the radar wave in the registration radar data.

[0134] The decomposed deformation is subjected to threshold screening to obtain the standard decomposed deformation;

[0135] The real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation.

[0136] Historical phase maps are extracted from the registered radar data based on the standard decomposition deformation.

[0137] The real-time phase map and the historical phase map are combined into an interferometric phase map.

[0138] In detail, by using the phase decomposition algorithm to calculate the decomposition deformation between the real-time radar data and the registered radar data, the difference between radar images can be determined based on the phase difference, thereby determining the deformation of the ground surface. The threshold filtering of the decomposition deformation to obtain the standard decomposition deformation refers to selecting moments in the decomposition deformation that are greater than a preset deformation threshold and aggregating them into the standard decomposition deformation. The extraction of the real-time phase map from the real-time radar data based on the standard decomposition deformation refers to extracting the corresponding timestamp from the standard decomposition deformation and extracting the real-time phase map from the real-time radar data based on the timestamp.

[0139] For details, refer to Figure 3 As shown, the parallax analysis of the real-time aerial data and the historical aerial data to obtain the variable tile group includes:

[0140] S31. Use the real-time aerial photography data to perform image registration on the historical aerial photography data to obtain registered aerial photography data;

[0141] S32. Perform grayscale pixel matching on the real-time aerial photography data and the registered aerial photography data to obtain the parallax pixel region;

[0142] S33. Using the parallax pixel region, filter out real-time changing map patches from the real-time aerial photography data, and use the parallax pixel region to filter out historical changing map patches from the registered aerial photography data;

[0143] S34. Combine the real-time changing map blocks and the historical changing map blocks into a changing map block group.

[0144] Specifically, the grayscale pixel matching of the real-time aerial photography data and the registered aerial photography data to obtain the disparity pixel region refers to pixel-by-pixel grayscale comparison, and filtering out the mismatched pixel regions as the disparity pixel regions.

[0145] In this embodiment of the invention, differential interferometric analysis is performed on the real-time radar data and the historical radar data to obtain an interferometric phase map group, and parallax analysis is performed on the real-time aerial photography data and the historical aerial photography data to obtain a variable map patch group. This allows for the direct filtering of radar images that differ between pre-disaster and post-disaster conditions, as well as the filtering of aerial images that differ between pre-disaster and post-disaster conditions, thereby reducing the order of magnitude of data analysis and improving the efficiency of disaster analysis.

[0146] S4. Perform model loss identification on the variable model group to obtain laser disaster data group; perform phase loss identification on the interferometric phase map group to obtain radar disaster data group; and perform image loss identification on the variable map block group to obtain aerial disaster data group.

[0147] In this embodiment of the invention, the laser disaster data group refers to the data group that marks the disaster situation of each terrain in the change model group, the radar disaster data group refers to the data group that marks the disaster situation of each terrain in the interferometric phase map group, and the aerial disaster data group refers to the data group that marks the disaster situation of each terrain in the change map group.

[0148] In this embodiment of the invention, the step of identifying model loss in the variable model group to obtain the laser disaster data group includes:

[0149] The variable model group is divided into a real-time model group and a historical model group;

[0150] Each real-time model in the real-time model group is selected as a target real-time model, and model features are extracted from the target real-time model to obtain the target real-time object features;

[0151] The historical model corresponding to the target real-time model is selected from the historical model group as the target historical model, and the model features are extracted from the target historical model to obtain the target historical object features;

[0152] Based on the characteristics of the target historical object, semantic recognition is performed on the target historical model to obtain the target model semantics;

[0153] The target model loss is calculated based on the target historical object features and the target real-time object features;

[0154] The target model semantics and the target model loss are aggregated into laser disaster data, and all the laser disaster data are aggregated into a laser disaster data group.

[0155] Specifically, the real-time model group is a combination of models composed of the real-time point cloud data in the variable model group, and the historical model group is a combination of models composed of the historical point cloud data in the variable model group.

[0156] In detail, the model feature extraction refers to extracting features such as color, shape, and texture of the model and compiling them into object features. Features such as color, shape, and texture of the model can be extracted using feature extraction algorithms such as Principal Component Analysis (PCA), Fast Point Feature Histograms (FPFH), and Signature of Histograms of Orientations (SHOT).

[0157] Specifically, the target model loss can be calculated using Euclidean distance, cosine distance, or Manhattan distance algorithms based on the historical object features and the real-time object features of the target.

[0158] Specifically, a trained Support Vector Machine (SVM), Convolutional Neural Network (CNN), or K-Nearest Neighbors (KNN) neural network or machine learning algorithm can be used to perform semantic recognition on the target historical model based on the characteristics of the target historical object, thereby obtaining the semantics of the target model.

[0159] Specifically, the step of performing phase loss identification on the interferometric phase map group to obtain the radar damage data group includes:

[0160] The interferometric phase map group is split into a real-time phase map and a historical phase map;

[0161] The historical phase map is segmented into blocks to obtain historical phase map block groups;

[0162] The real-time phase map is divided into real-time phase map groups based on the historical phase map group;

[0163] Semantic recognition is performed on the historical phase map block group to obtain the phase semantic group;

[0164] A phase loss group is calculated based on the real-time phase block group and the historical phase block group;

[0165] A radar disaster data group is generated based on the phase semantic group and the phase loss group.

[0166] Specifically, the historical phase map can be segmented using contour-based tile segmentation or region growing methods to obtain a historical phase map tile group. The phase loss group can be calculated based on the real-time phase map tile group and the historical phase map tile group using Pearson correlation coefficient or Jaccard algorithm.

[0167] In detail, the method of performing image loss recognition on the changed patch group to obtain the aerial disaster data group is the same as the method of performing phase loss recognition on the interferometric phase patch group to obtain the radar disaster data group in step S4 above, and will not be repeated here.

[0168] In this embodiment of the invention, by performing model loss identification on the variable model group, a laser disaster data group is obtained; by performing phase loss identification on the interferometric phase map group, a radar disaster data group is obtained; and by performing image loss identification on the variable map group, an aerial disaster data group is obtained. Triple semantic recognition can be achieved based on model data, radar data, and image data, which improves the accuracy of surface information extraction in disaster areas. Furthermore, by determining the magnitude of the disaster impact through the differences between model data, radar data, and image data before and after the disaster, the accuracy of disaster area loss analysis is improved.

[0169] S5. Based on the laser disaster data set, the radar disaster data set, and the aerial disaster data set, a comprehensive loss analysis is performed on the disaster area to be tested to obtain the damage results of the disaster area.

[0170] In this embodiment of the invention, the step of performing a comprehensive loss analysis on the disaster area to be tested based on the laser disaster data group, the radar disaster data group, and the aerial disaster data group to obtain the disaster area damage result refers to matching all target model semantics in the laser disaster data group, all phase semantics in the radar disaster data group, and all image semantics in the aerial disaster data group, and using the weighted average of the calculated model loss, phase loss, and image loss of the corresponding semantics as the standard disaster loss, and aggregating all semantics and the corresponding standard disaster loss into the disaster area damage result.

[0171] In this embodiment of the invention, by performing a comprehensive loss analysis on the disaster area to be tested based on the laser disaster data set, the radar disaster data set, and the aerial disaster data set, the damage results of the disaster area are obtained. This method can combine three-dimensional model data, radar data, and image data to perform triple disaster analysis, thereby improving the accuracy of disaster area loss analysis.

[0172] This invention utilizes unmanned aerial vehicles (UAVs) to cruise and photograph disaster areas, obtaining real-time laser data, real-time radar data, and real-time aerial photography data. UAVs can be used to collect real-time 3D information, 2D image information, and radar information of the disaster area, facilitating subsequent disaster analysis. By performing point cloud difference analysis on the real-time laser data and historical laser data, a variable point cloud data group is obtained. This allows for direct filtering of point cloud data that has changed over time, reducing the amount of data required for disaster analysis and improving its efficiency. By performing point cloud repair and 3D modeling on the variable point cloud data group, a variable model group is obtained, enabling the repair of point cloud data and facilitating more accurate disaster analysis. By performing differential interferometry analysis on the real-time radar data and historical radar data, an interferometric phase map group is obtained. By performing parallax analysis on the real-time aerial photography data and historical aerial photography data, a variable patch group is obtained. This allows for direct filtering of radar images showing differences before and after the disaster, as well as filtering of aerial images showing differences before and after the disaster, reducing the order of magnitude of data analysis and improving the efficiency of disaster analysis.

[0173] By identifying model loss in the changing model group, a laser-damaged data group is obtained; by identifying phase loss in the interferometric phase map group, a radar-damaged data group is obtained; and by identifying image loss in the changing patch group, an aerial-damaged data group is obtained. Triple semantic recognition can be achieved based on model data, radar data, and image data, improving the accuracy of surface information extraction in disaster areas. Furthermore, the magnitude of the disaster impact is determined by the differences between model data, radar data, and image data before and after the disaster, improving the accuracy of disaster area loss analysis. By performing a comprehensive loss analysis on the disaster area based on the laser-damaged data group, the radar-damaged data group, and the aerial-damaged data group, the damage results of the disaster area are obtained. This method combines three-dimensional model data, radar data, and image data for triple disaster analysis, improving the accuracy of disaster area loss analysis. Therefore, the UAV-based disaster area loss analysis method proposed in this invention can solve the problem of low efficiency in loss analysis.

[0174] like Figure 4 The diagram shown is a device architecture diagram of a disaster area loss analysis device based on a drone provided in an embodiment of the present invention.

[0175] The UAV-based disaster loss analysis device 100 of this invention can be installed in an electronic device. Depending on the functions implemented, the UAV-based disaster loss analysis device 100 may include a data acquisition module 101, a 3D modeling module 102, a differential interferometry module 103, a semantic recognition module 104, and a disaster analysis module 105. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0176] In this embodiment, the functions of each module / unit are as follows:

[0177] The data acquisition module 101 is used to use the UAV to cruise and photograph the disaster area under test, obtain real-time laser data, real-time radar data and real-time aerial photography data, and extract historical laser data, historical radar data and historical aerial photography data from historical cruise data.

[0178] The three-dimensional modeling module 102 is used to perform point cloud difference analysis on the real-time laser data and the historical laser data to obtain a variable point cloud data group, and to perform point cloud repair and three-dimensional modeling on the variable point cloud data group to obtain a variable model group.

[0179] The differential interferometry module 103 is used to perform differential interferometry analysis on the real-time radar data and the historical radar data to obtain an interferometric phase map group, and to perform parallax analysis on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometry analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: using the real-time radar data to perform radar registration on the historical radar data to obtain registration and matching data; and using the following phase decomposition algorithm to calculate the decomposition deformation between the real-time radar data and the registered radar data:

[0180]

[0181] Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map.

[0182] The semantic recognition module 104 is used to perform model loss recognition on the variable model group to obtain laser disaster data group, perform phase loss recognition on the interferometric phase map group to obtain radar disaster data group, and perform image loss recognition on the variable map block group to obtain aerial disaster data group.

[0183] The disaster analysis module 105 is used to perform a comprehensive loss analysis on the disaster area to be tested based on the laser disaster data group, the radar disaster data group and the aerial disaster data group, and obtain the damage results of the disaster area.

[0184] In detail, the modules in the UAV-based disaster loss analysis device 100 described in this embodiment of the invention employ the same methods as described above during use. Figures 1 to 3 The method used is the same as the one described in the article on disaster area loss analysis based on drones, and can produce the same technical effect, so it will not be repeated here.

[0185] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0186] Drones were used to patrol and photograph the disaster area under test, obtaining real-time laser data, real-time radar data, and real-time aerial photography data. Historical laser data, historical radar data, and historical aerial photography data were extracted from historical patrol data.

[0187] Point cloud difference analysis is performed on the real-time laser data and the historical laser data to obtain a variable point cloud data group. Point cloud repair and 3D modeling are performed on the variable point cloud data group to obtain a variable model group.

[0188] Differential interferometry analysis is performed on the real-time radar data and the historical radar data to obtain an interferometric phase map group. Parallax analysis is performed on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometry analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: using the real-time radar data to perform radar registration on the historical radar data to obtain registration and matching data; and using the following phase decomposition algorithm to calculate the decomposition deformation between the real-time radar data and the registered radar data:

[0189]

[0190] Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map.

[0191] Model loss identification is performed on the variable model group to obtain the laser disaster data group; phase loss identification is performed on the interferometric phase map group to obtain the radar disaster data group; and image loss identification is performed on the variable map block group to obtain the aerial disaster data group.

[0192] A comprehensive loss analysis is performed on the disaster area to be tested based on the laser disaster data set, the radar disaster data set, and the aerial disaster data set to obtain the damage results of the disaster area.

[0193] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0194] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0195] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0196] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0197] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0198] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0199] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application devices that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0200] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices described in the apparatus embodiments may also be implemented by a single unit or device through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0201] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for disaster area loss analysis based on unmanned aerial vehicles (UAVs), characterized in that, The method includes: Drones were used to patrol and photograph the disaster area under test, obtaining real-time laser data, real-time radar data, and real-time aerial photography data. Historical laser data, historical radar data, and historical aerial photography data were extracted from historical patrol data. Point cloud difference analysis is performed on the real-time laser data and the historical laser data to obtain a variable point cloud data group. Point cloud repair and 3D modeling are performed on the variable point cloud data group to obtain a variable model group. Differential interferometry analysis is performed on the real-time radar data and the historical radar data to obtain an interferometric phase map group. Parallax analysis is performed on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometry analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: using the real-time radar data to perform radar registration on the historical radar data to obtain registration and matching data; and using the following phase decomposition algorithm to calculate the decomposition deformation between the real-time radar data and the registered radar data: Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map. Model loss identification is performed on the variable model group to obtain the laser disaster data group; phase loss identification is performed on the interferometric phase map group to obtain the radar disaster data group; and image loss identification is performed on the variable map block group to obtain the aerial disaster data group. A comprehensive loss analysis is performed on the disaster area to be tested based on the laser disaster data set, the radar disaster data set, and the aerial disaster data set to obtain the damage results of the disaster area.

2. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The method of using drones to cruise and photograph the disaster area under test, obtaining real-time laser data, real-time radar data, and real-time aerial photography data, includes: The drone is used to conduct a comprehensive patrol of the disaster area under test according to a preset patrol route. The laser scanning equipment of the UAV is used to perform laser scanning on the disaster area to be measured during the coverage patrol to obtain laser patrol data; The radar equipment of the UAV is used to perform radar scanning on the disaster area to be measured during the coverage patrol to obtain radar patrol data. The camera equipment of the drone is used to take images of the disaster area to be tested during the coverage patrol, and aerial patrol data is obtained. Based on the stated cruise route, the laser cruise data is stitched together to form real-time laser data, the radar cruise data is stitched together to form real-time radar data, and the aerial photography cruise data is stitched together to form real-time aerial photography data.

3. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The point cloud difference analysis of the real-time laser data and the historical laser data yields a variable point cloud data set, including: The historical laser data is registered using the real-time laser data to obtain registered laser data. The registered laser data and the real-time laser data are compared in elevation to obtain elevation difference data; Real-time point cloud data is filtered from the real-time laser data based on the elevation difference data; Historical point cloud data are filtered from the registered laser data based on the elevation difference data; The real-time point cloud data and the historical point cloud data are aggregated into a variable point cloud data group.

4. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 3, characterized in that, The step of using the real-time laser data to perform point cloud registration on the historical laser data to obtain registered laser data includes: Point cloud matching is performed on the real-time laser data and the historical laser data to obtain a matching point cloud group; Real-time matching point cloud and historical matching point cloud are extracted from the matching point cloud group respectively; The covariance centroid point cloud is calculated using the following centroid matching algorithm based on the real-time matching point cloud and the historical matching point cloud. Wherein, H refers to the covariance centroid point cloud, i refers to the point cloud index, N refers to the total number of points in the real-time matched point cloud, and the total number of points in the real-time matched point cloud is equal to the total number of points in the historical matched point cloud, p i This refers to the point cloud with point cloud index i in the real-time matched point cloud, q i It refers to the point cloud with point cloud index i in the historical matching point cloud, where T is the transpose symbol; Singular value decomposition is performed on the covariance centroid point cloud to obtain the rotation matrix; The translation vector is calculated based on the rotation matrix, the real-time matching point cloud, and the historical matching point cloud. The historical laser data is registered using the translation vector and the rotation matrix to obtain the registered laser data.

5. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The point cloud repair and 3D modeling of the changed point cloud data group yields a changed model group, including: Point cloud filtering is performed on the variable point cloud data group to obtain a noise-reduced point cloud data group; Generate a point cloud feature data group corresponding to the denoised point cloud data group, and use the following point cloud distance algorithm to segment the point cloud feature data group to obtain a segmented point cloud feature cluster group: Where D refers to the point cloud feature distance of the point cloud distance algorithm, α and β are preset adversarial coefficients, · is the dot product symbol, * is the cross product symbol, p refers to point cloud feature p in the point cloud feature data set, q refers to point cloud feature q in the point cloud feature data set, T is the transpose symbol, cov() is the covariance symbol, cov(p,p) refers to the covariance of point cloud feature p, cov(p,q) refers to the covariance between point cloud feature p and point cloud feature q, cov(q,p) refers to the covariance between point cloud feature q and point cloud feature p, and cov(q,q) refers to the covariance of point cloud feature q. Based on the segmented point cloud cluster feature group, the noise-reduced point cloud data group is triangulated into a segmented triangular model group. Extract the set of triangle vertices and the set of triangle faces from the segmented triangle model set; Based on the set of triangle vertices and the set of triangle faces, the segmented triangle model group is reconstructed to obtain a variable model group.

6. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The process of performing parallax analysis on the real-time aerial data and the historical aerial data to obtain a group of changing map tiles includes: The historical aerial photography data is image registered using the real-time aerial photography data to obtain registered aerial photography data. Grayscale pixel matching is performed on the real-time aerial photography data and the registered aerial photography data to obtain the parallax pixel region; Real-time changing patches are selected from the real-time aerial photography data using the parallax pixel region, and historical changing patches are selected from the registered aerial photography data using the parallax pixel region. The real-time changing tiles and the historical changing tiles are combined into a changing tile group.

7. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The step of identifying model loss in the variable model group to obtain a laser disaster data group includes: The variable model group is divided into a real-time model group and a historical model group; Each real-time model in the real-time model group is selected as a target real-time model, and model features are extracted from the target real-time model to obtain the target real-time object features; The historical model corresponding to the target real-time model is selected from the historical model group as the target historical model, and the model features are extracted from the target historical model to obtain the target historical object features; Based on the characteristics of the target historical object, semantic recognition is performed on the target historical model to obtain the target model semantics; The target model loss is calculated based on the target historical object features and the target real-time object features; The target model semantics and the target model loss are aggregated into laser disaster data, and all the laser disaster data are aggregated into a laser disaster data group.

8. The method for disaster area loss analysis based on unmanned aerial vehicles as described in claim 1, characterized in that, The step of performing phase loss identification on the interferometric phase map group to obtain the radar disaster data group includes: The interferometric phase map group is split into a real-time phase map and a historical phase map; The historical phase map is segmented into blocks to obtain historical phase map block groups; The real-time phase map is divided into real-time phase map groups based on the historical phase map group; Semantic recognition is performed on the historical phase map block group to obtain the phase semantic group; A phase loss group is calculated based on the real-time phase block group and the historical phase block group; A radar disaster data group is generated based on the phase semantic group and the phase loss group.

9. A disaster area loss analysis device based on unmanned aerial vehicles (UAVs), characterized in that, The device includes: The data acquisition module is used to use drones to cruise and photograph the disaster area under test, obtain real-time laser data, real-time radar data and real-time aerial photography data, and extract historical laser data, historical radar data and historical aerial photography data from historical cruise data. The 3D modeling module is used to perform point cloud difference analysis on the real-time laser data and the historical laser data to obtain a variable point cloud data group, and to perform point cloud repair and 3D modeling on the variable point cloud data group to obtain a variable model group. The differential interferometry module is used to perform differential interferometric analysis on the real-time radar data and the historical radar data to obtain an interferometric phase map group, and to perform parallax analysis on the real-time aerial photography data and the historical aerial photography data to obtain a variable patch group. The step of performing differential interferometric analysis on the real-time radar data and the historical radar data to obtain the interferometric phase map group includes: performing radar registration on the historical radar data using the real-time radar data to obtain registration and matching data; and calculating the decomposition deformation between the real-time radar data and the registered radar data using the following phase decomposition algorithm: Where F(t) refers to the decomposition deformation at time t, t is the time index, ln() is the natural logarithm function, S1(t) refers to the radar signal at time t in the real-time radar data, A1(t) refers to the amplitude at time t in the real-time radar data, j is the imaginary number, and π is pi. S2(t) refers to the center frequency of the radar wave in the real-time radar data, S2(t) refers to the amplitude at time t in the registration radar data, and A2(t) refers to the amplitude at time t in the registration radar data. θ refers to the center frequency of the radar wave in the registered radar data, and θ is the wavelength of the radar wave in the real-time radar data, which is equal to the wavelength of the radar wave in the registered radar data. The decomposition deformation is threshold-filtered to obtain a standard decomposition deformation. A real-time phase map is extracted from the real-time radar data based on the standard decomposition deformation. A historical phase map is extracted from the registered radar data based on the standard decomposition deformation. The real-time phase map and the historical phase map are then combined into an interferometric phase map. The semantic recognition module is used to identify model loss in the variable model group to obtain laser disaster data group, identify phase loss in the interferometric phase map group to obtain radar disaster data group, and identify image loss in the variable map block group to obtain aerial disaster data group. The disaster analysis module is used to perform a comprehensive loss analysis on the disaster area to be tested based on the laser disaster data group, the radar disaster data group, and the aerial disaster data group, and to obtain the damage results of the disaster area.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the UAV-based disaster area loss analysis method as described in any one of claims 1 to 8.

Images