Method, device, equipment and storage medium for identifying earthquake-damaged buildings

Abstract

The present invention provides a method for identifying earthquake-damaged buildings, comprising: estimating the coherence of the pre-seismic time series and the co-seismic map using a homogeneous point selection method based on the pre-seismic time series intensity map, thereby obtaining a pre-seismic time series and co-seismic prediction coherence map; obtaining a pre-seismic time series and co-seismic enhancement coherence map based on coherence map histogram matching; subtracting the pre-seismic prediction and enhancement coherence maps to obtain a pre-seismic time series coherence difference map; obtaining a reliability map and a background effect removal threshold through the time series prediction and coherence difference map; subtracting the pre-seismic prediction and the co-seismic enhancement map to obtain a co-seismic coherence difference map; processing the co-seismic coherence difference map based on a threshold to obtain a result map; and intersecting the result map, the reliability map, and the building distribution map to obtain a final proxy map for building damage. This solution solves the problem in the prior art that when using a common rule window to estimate coherence, coherence estimation bias is easily generated, and subtracting the coherence map generates additional bias, thereby affecting identification accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of SAR image recognition, and in particular relates to a method, device, equipment and storage medium for identifying earthquake-damaged buildings. Background Art

[0002] Earthquakes are one of the world's deadliest and most frequent natural disasters, causing massive damage to buildings. It's crucial to quickly and accurately obtain information on building damage in earthquake-affected areas, which will be very helpful in guiding emergency rescue and material dispatch efforts in disaster areas. Traditional manual field survey methods are inefficient, costly, and dangerous; high-resolution optical satellite imagery is often affected by weather conditions such as clouds and fog, making it difficult to implement post-earthquake emergency responses; SAR (Synthetic Aperture Radar) satellite imagery has all-day, all-weather, and large-scale imaging capabilities, making it an important data source for post-earthquake emergency response. InSAR (Interferometric Synthetic Aperture Radar) coherence is increasingly being used to identify earthquake-damaged buildings because of its extreme sensitivity to building damage.

[0003] Homogeneous point selection, also known as homogeneous sample selection, is an algorithm that uses statistical inference to measure the similarity between neighboring pixels and a central pixel (Jiang M, Ding XL, Hanssen RF, et al. 2015; Paarizzi A, Bricic R. 2011). Under the assumption that similar ground objects have identical backscattering properties and phase scattering centers, similar pixels are aggregated for parameter estimation. This improves the signal-to-noise ratio while maintaining image resolution, making it particularly suitable for fine deformation monitoring in complex scenes.

[0004] In a paper published in the Information Science Edition of the Journal of Wuhan University in 2023, Liu Zhenjiang and others conducted a study on the exploration of the seismic faults of the 2023 Herat earthquake sequence and the assessment of building damage under the constraints of InSAR data. The scheme is based on the ascending and descending SAR images of the European Space Agency's Sentinel-1, and uses InSAR technology to obtain the co-seismic deformation field of the 2023 Herat earthquake sequence. With the ascending and descending InSAR observations as constraints, the seismic fault geometry and fault slip distribution were inverted and determined, and the seismic structure was analyzed and discussed; at the same time, based on the multi-phase InSAR coherence change detection method, the post-earthquake building damage proxy map (BDPM) of this earthquake was analyzed and extracted. The BDPM results show that this earthquake sequence caused relatively serious damage to buildings within 40km of the epicenter, which may be the direct cause of the large-scale casualties caused by this earthquake.

[0005] However, the aforementioned scheme uses a commonly used regular window to estimate coherence, which severely reduces the spatial resolution of coherence and is prone to biased coherence estimation. Furthermore, the scheme uses only ascending orbit data, making the results susceptible to geometric distortion. Furthermore, the scheme does not quantitatively validate the BDPM results, making it impossible to obtain information on identification accuracy. Furthermore, when using InSAR coherence to identify earthquake-damaged buildings, the high-resolution SAR imagery commonly used suffers from issues such as limited spatial coverage, limited archived data, a lack of global coverage with a fixed revisit period, and reduced accuracy due to geometric distortion in SAR side-view imaging. Summary of the Invention

[0006] In order to solve the problems in the prior art of using a commonly used regular window to estimate coherence, which seriously reduces the spatial resolution of coherence, easily produces coherence estimation bias, and reduces recognition accuracy due to geometric distortion of SAR side-looking imaging, the present invention provides a method, device, equipment and storage medium for identifying earthquake-damaged buildings.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] A method for identifying earthquake-damaged buildings comprises the following steps:

[0009] The pre-earthquake SAR images and co-seismic SAR images taken by spaceborne SAR in the study area were obtained in ascending and descending orbits, and the pre-earthquake time-series intensity maps, pre-earthquake time-series differential interferograms, and co-seismic differential interferograms were obtained after interferometric processing of the pre-earthquake SAR images and co-seismic SAR images in ascending and descending orbits.

[0010] Using pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels, a coherence estimation method based on homogeneous point selection is used to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits, thereby obtaining pre-seismic time series and co-seismic predicted coherence maps of the ascending and descending orbits; and performing corresponding histogram matching on the coherence maps to obtain pre-seismic time series and co-seismic enhancement coherence maps of the ascending and descending orbits;

[0011] The pre-earthquake predicted coherence map of the ascending and descending orbits is subtracted from the pre-earthquake enhanced coherence map pixel by pixel, respectively, to obtain the pre-earthquake time series coherence difference map of the ascending and descending orbits. The reliability map and background effect removal threshold of the ascending and descending orbits are generated respectively based on the pre-earthquake time series predicted coherence map and pre-earthquake time series coherence difference map of the ascending and descending orbits;

[0012] The pixel-by-pixel difference between the pre-earthquake prediction coherence map of the ascending and descending orbits closest to the earthquake and the coseismic enhancement coherence map is obtained;

[0013] Based on the background effect removal thresholds for ascending and descending orbits, the background effect of the coseismic coherence difference maps of the ascending and descending orbits is removed to obtain the result maps of the ascending and descending orbits. The building damage proxy maps of the ascending and descending orbits are obtained by taking the common intersection of the result maps of the ascending and descending orbits, the reliability map, and the existing building distribution map.

[0014] The building damage proxy maps of the ascending and descending orbits are fused to obtain the final building damage proxy map.

[0015] Furthermore, the step of fusing the ascending and descending building damage proxy maps to obtain a final building damage proxy map includes:

[0016] According to the slope, aspect, elevation and satellite orbit direction of the study area, an orbit that is relatively less affected by geometric distortion is determined as the dominant orbit;

[0017] Obtain the union of the building damage proxy maps of the ascending and descending tracks. If there is a building damage signal at the same identification position on both the ascending and descending tracks, only retain the dominant track signal and output it to the final building damage proxy map.

[0018] Furthermore, the step of estimating the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits using a coherence estimation method based on homogeneous point selection based on pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels comprises:

[0019] Each pixel in the pre-earthquake intensity map covering the study area of ​​the ascending and descending orbits is used as a reference pixel. The non-parametric statistical Baumgartner-Weiβ-Schindler test is used to compare the reference pixels in the pre-earthquake intensity map with the adjacent time series pixels to be judged in a preset size window. For each reference pixel in the window, a statistically homogeneous pixel is selected within the preset size window.

[0020] The homogeneous pixel information is used to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits.

[0021] Furthermore, the regions in the co-seismic coherence difference maps of the ascending orbit and the descending orbit with difference values ​​greater than the background effect removal threshold are retained and output respectively, to obtain the result maps of the ascending orbit and the descending orbit after removing the background effect.

[0022] Furthermore, the step of generating the reliability graphs of the ascending orbit and the descending orbit by using the pre-earthquake time series prediction coherence graphs of the ascending orbit and the descending orbit includes:

[0023] Calculate the pixel-by-pixel mean and standard deviation of the pre-earthquake time series prediction coherence map for ascending and descending orbits respectively;

[0024] The statistical mean of the standard deviation within the building area in the study area is used as the threshold, and the areas with standard deviation less than the threshold are output as reliable areas, thereby obtaining the reliability maps of the ascending and descending orbits.

[0025] Furthermore, the sum of the average value and 1 standard deviation of the pre-earthquake time series coherence difference map of the ascending orbit and the descending orbit is used as the background effect removal threshold of the ascending orbit and the descending orbit, respectively.

[0026] Furthermore, the coseismic coherence difference maps of the ascending and descending orbits are obtained by subtracting the pixel values ​​of the pre-earthquake prediction coherence maps closest to the earthquake from the coseismic enhancement coherence maps of the ascending and descending orbits.

[0027] A device for identifying earthquake-damaged buildings, comprising:

[0028] The image acquisition and preprocessing module is used to obtain the pre-earthquake SAR images and co-seismic SAR images of the study area taken by the spaceborne SAR in the ascending and descending orbits, and to perform interferometric processing on the pre-earthquake SAR images and co-seismic SAR images in the ascending and descending orbits, respectively, to obtain the pre-earthquake time-series intensity maps, pre-earthquake time-series differential interferograms, and co-seismic differential interferograms in the ascending and descending orbits;

[0029] A first processing module is configured to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits using a coherence estimation method based on homogeneous point selection, using pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels, thereby obtaining pre-seismic time series and co-seismic estimated coherence maps of the ascending and descending orbits; and further perform corresponding histogram matching based on the coherence maps to obtain pre-seismic time series and co-seismic enhancement coherence maps of the ascending and descending orbits;

[0030] The second processing module is configured to perform pixel-by-pixel subtraction between the pre-earthquake predicted coherence maps of the ascending and descending orbits and the pre-earthquake enhanced coherence maps, respectively, to obtain pre-earthquake time series coherence difference maps of the ascending and descending orbits, and then generate reliability maps and background effect removal thresholds for the ascending and descending orbits respectively based on the pre-earthquake time series predicted coherence maps and pre-earthquake time series coherence difference maps of the ascending and descending orbits;

[0031] The third processing module is used to obtain the coseismic coherence difference map of the ascending and descending orbits by performing pixel-by-pixel difference between the pre-earthquake prediction coherence map of the ascending and descending orbits closest to the earthquake and the coseismic enhancement coherence map;

[0032] The fourth processing module is configured to remove the background effect from the coseismic coherence difference maps of the ascending and descending orbits according to the background effect removal thresholds of the ascending and descending orbits, respectively, to obtain the result maps of the ascending and descending orbits; and obtain the building damage proxy maps of the ascending and descending orbits by taking the common intersection of the result maps of the ascending and descending orbits, the reliability map, and the existing building distribution map;

[0033] The fifth processing module is used to fuse the ascending and descending building damage proxy maps to obtain the final building damage proxy map.

[0034] A computer device includes a memory and a processor, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory to implement the earthquake-damaged building identification method as described above.

[0035] A computer-readable storage medium is used to store computer-executable instructions. When the computer-executable instructions are executed by a processor, they are used to implement the earthquake-damaged building identification method as described above.

[0036] The earthquake-damaged building identification method provided by the present invention has the following beneficial effects:

[0037] This scheme obtains the pre-seismic time series intensity map, pre-seismic time series differential interference map and co-seismic differential interference map of the ascending and descending orbits; uses the pixels in the pre-seismic time series intensity map of the ascending and descending orbits as reference pixels, and estimates the coherence of the pre-seismic time series and co-seismic differential interference map of the ascending and descending orbits respectively based on the coherence estimation method of homogeneous point selection; after obtaining the pre-seismic time series and co-seismic estimated coherence map of the ascending and descending orbits, corresponding histogram matching is performed to obtain the pre-seismic time series and co-seismic enhanced coherence map of the ascending and descending orbits; the coherence estimation method of homogeneous pixel point selection is used to maintain the spatial resolution of the coherence and reduce the influence of the coherence estimation deviation; the histogram matching processing is used to weaken the adjacent baseline inconsistency. The fusion of these two techniques effectively improves the accuracy of earthquake-damaged building identification by eliminating the errors caused by subtracting consistent coherence maps. The coseismic enhancement coherence map is then subtracted from the pre-earthquake estimated coherence map closest to the earthquake to generate a coseismic coherence difference map. Furthermore, based on the background effect removal thresholds for ascending and descending orbits, background-removed result maps for ascending and descending orbits are generated. Proxy maps of building damage for ascending and descending orbits are then derived based on the intersection of the result maps, the reliability map, and the building distribution map. Finally, the proxy maps for ascending and descending orbits are fused to produce the final proxy map. This fusion approach addresses the low accuracy of SAR side-looking imaging due to geometric distortion. This approach comprehensively addresses the issues of commonly used regular window coherence estimation, which severely reduces spatial resolution and easily leads to coherence estimation bias, as well as the additional bias caused by subtracting coherence maps with inconsistent spatiotemporal baselines, which can affect identification accuracy. Furthermore, this approach eliminates the need to rely on prior information about damaged buildings after the earthquake, enabling rapid and accurate identification of earthquake-damaged buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] To more clearly illustrate the embodiments of the present invention and its design, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0039] Figure 1 Detailed steps of the identification method according to an embodiment of the present invention are shown in FIG.

[0040] Figure 2 This is a time-space baseline diagram of the ascending and descending orbit interference of the method embodiment of the present invention: Figure 2 (a) is the time-space baseline diagram of orbit-raising interferometry. Figure 2 (b) is the time-space baseline diagram of descending orbit interferometry;

[0041] Figure 3 A co-seismic coherence difference diagram of an embodiment of the method of the present invention;

[0042] Figure 4 A reliability diagram of an embodiment of the method of the present invention;

[0043] Figure 5 This is a diagram showing the background removal effect results of an embodiment of the method of the present invention;

[0044] Figure 6 A building damage proxy map (BDPM) is provided for an embodiment of the method of the present invention. DETAILED DESCRIPTION

[0045] In order to enable those skilled in the art to better understand the technical solution of the present invention and to be able to implement it, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention.

[0046] Method Example

[0047] The present invention provides a method for identifying earthquake-damaged buildings. Figure 1 As shown, the specific steps include:

[0048] Step 1: Coherence map estimation and histogram matching.

[0049] Firstly, the ascending and descending SAR images of the study area taken by spaceborne SAR were obtained one year before the earthquake and the first scene after the earthquake (co-seismic SAR images) and processed by single-link interferometry to generate the ascending and descending orbit pre-earthquake time series intensity maps, un-unwrapped pre-earthquake time series differential interferograms, and co-seismic differential interferograms, respectively.

[0050] Each pixel in the pre-earthquake intensity map covering the study area, from both ascending and descending orbits, is used as a reference pixel. Using the nonparametric Baumgartner-Weiβ-Schindler (BWS) test, the reference pixel in the pre-earthquake intensity map is compared with adjacent time-series pixels to be identified within a window of a predetermined size. For each reference pixel within the window, statistically homogeneous pixels are selected within the window of a predetermined size. The values ​​of these selected homogeneous pixels are statistically considered samples of the same population and are used in subsequent coherence estimation. The coherence of the pre-seismic time series and co-seismic differential interferograms for ascending and descending orbits is estimated based on the selected homogeneous point distribution information, pre-seismic time series differential interferograms, and co-seismic differential interferograms, respectively. This results in the pre-seismic time series and co-seismic predicted coherence maps for ascending and descending orbits. Finally, histogram matching is performed on the pre-seismic time series and co-seismic predicted coherence maps for ascending and descending orbits (matching relationship: Pair 32 to Pair 31, Pair 31 to Pair 30, ..., Pair 2 to Pair 1, where Pair X is the number of the interferometric pair (coherence) used, using ascending SAR image data as an example), resulting in the pre-seismic time series and co-seismic enhanced coherence maps for ascending and descending orbits.

[0051] Step 2: Obtain the pre-earthquake coherence difference map, reliability map, and background effect removal threshold.

[0052] The pre-earthquake estimated and enhanced coherence maps of ascending and descending orbits are used to perform pixel-by-pixel subtraction (the estimated coherence map of the previous interferometer pair minus the enhanced coherence map of the next interferometer pair) to obtain the pre-earthquake time series coherence difference maps of ascending and descending orbits (the difference calculation order is: Pair 1 minus Pair 2, Pair 2 minus Pair 3, ...; where Pair 1 is the estimated coherence map of the previous interferometer pair, and Pair 2 is the enhanced coherence map of the next interferometer pair after histogram matching, taking ascending SAR images as an example). Then, the pre-earthquake time series estimated coherence maps and coherence difference maps of ascending and descending orbits are used to generate the reliability maps and background effect removal thresholds of ascending and descending orbits, respectively. The reliability map is used to screen stable areas, while the background effect removal threshold is used to eliminate background effects (such as minor changes in coherence caused by changes in surface vegetation and atmospheric effects). First, the mean and standard deviation of the estimated coherence maps (without histogram matching) for the year before the earthquake are calculated for ascending and descending orbits, respectively. The statistical mean of the standard deviation within the building area in the study area is used as the threshold, and areas with a standard deviation less than the threshold are output as reliable areas, thereby obtaining the reliability maps for ascending and descending orbits. The sum of the mean and 1 times the standard deviation of the pre-earthquake time series coherence difference maps for ascending and descending orbits is used as the background effect removal threshold for ascending and descending orbits, respectively. To remove background effects.

[0053] Step 3: Generation of Building Damage Proxy Map (BDPM).

[0054] The ascending coseismic coherence difference map (CCD) is obtained by subtracting the pre-earthquake coherence map of the ascending track closest to the earthquake with the coseismic coherence map of the ascending track (the coherence obtained from the first images before and after the earthquake closest to the earthquake). The descending coseismic coherence difference map is obtained by subtracting the pre-earthquake coherence map of the descending track closest to the earthquake with the coseismic coherence map of the descending track. Regions in the CCD maps of the ascending and descending tracks with differences greater than the background effect removal threshold are retained and output, resulting in the background-removed ascending and descending result maps. The intersection of the ascending and descending result maps, the reliability map, and the existing building distribution map is then taken to obtain the ascending and descending building damage proxy maps (BDPMs), respectively. The building distribution map is provided by Open Street Map (OSM) or World Settlement Footprint 2019 (WSF 2019). Finally, the ascending and descending building damage proxy maps are fused to produce the final building damage proxy map (BDPM). When fusing ascending and descending data, an orbit that is less affected by geometric distortion is first determined based on the slope, aspect, elevation, and direction of the study area, as well as the satellite's orbital orientation. This selected orbit is then designated as the dominant orbit. The union of the ascending and descending building damage proxy maps is then obtained. If building damage signals are present at the same identified location on both ascending and descending orbits, only the signal from the dominant orbit (which is more reliable) is retained and output to the final building damage proxy map.

[0055] The identification method of the present invention is further explained below using the Mw 6.2 earthquake in Afghanistan on June 22, 2022 as an example:

[0056] Figure 2 is the space-time baseline diagram of the ascending and descending interferometer pairs. Pair X represents the encoding of the interferometer pair (coherence diagram). Figure 2 (a) and Figure 2 (b) represents the ascending track 71 and the descending track 78 respectively.

[0057] Figure 3 It is the coseismic coherence difference map obtained by the pre-earthquake prediction coherence map and the coseismic enhancement coherence map closest to the earthquake. Figure 3 (a)-(c) and Figure 4 (d)-(f) are the pre-earthquake prediction coherence map, coseismic enhancement coherence map, and coseismic coherence difference map of the ascending and descending orbit images, respectively. Figure 3 (c) and Figure 3 The coherence difference in (f) is obtained by subtracting the coseismic enhancement coherence map from the pre-earthquake predicted coherence map, and its value ranges from -1 to 1.

[0058] Figure 4 Reliability map of the study area. Figure 4 (a)-(c) and Figure 4 Figures (d)-(f) show the average temporal coherence, standard deviation of temporal coherence, and reliability maps for ascending and descending SAR interferometry pairs, respectively, one year before the earthquake. Statistical analysis indicates that the threshold for reliability ranges from 0.1 to 0.3. In this study, a threshold of 0.14 was chosen as the threshold for distinguishing stable from unstable regions within the study area. Regions with a standard deviation less than 0.14 were identified as stable coherence regions, while regions with a standard deviation greater than 0.14 were defined as unstable coherence regions (e.g., vegetation areas).

[0059] Figure 5 This is the result of removing the background effect. Figure 5 (a)-(b) and Figure 5 (d)-(e) are the average and standard deviation of the time series coherence difference map of SAR data one year before the ascending and descending earthquakes respectively. The result of removing the background effect ( Figure 5 (c) and Figure 5 (f)) is obtained by retaining and outputting the coseismic coherence difference map with the value greater than the background effect removal threshold. Part of what is obtained.

[0060] Figure 6 This is a building damage index map. Figure 6 (b) and (c) correspond to Figure 6 The white dashed boxes (b) and (c) in (a) represent the locations of shelters and damaged buildings, respectively, as manually interpreted by the United Nations Institute for Training and Research (UNITAR) using high-resolution optical imagery. Higher color values ​​indicate greater damage (from green to yellow to red). The red dashed boxes represent areas visually interpreted by UNITAR. Figure 6 (d) Figure 6 (e), Figure 6 (f), Figure 6 (g) is an optical image of the building before and after damage provided by UNITAR. Figure 6 (a) Figure 6 (b) and Figure 6The high-resolution optical image base map in (c) is from Google Earth. To verify the accuracy of the BDPM results, the BDPM and UNITAR results were quantitatively compared. The method proposed in this study had a correct recognition rate of 80.1%, a miss rate of 19.9%, and an erroneous recognition rate of only 17.2%. In addition, the present invention compared the ascending and descending results with the fusion results. The results showed that the correct recognition rate of the ascending orbit was 76.2%, while the correct recognition rate of the descending orbit was only 49%; after the ascending and descending orbits were fused, the correct recognition rate increased to 80.1%. At the same time, the miss rate of the ascending orbit was 23.8%, and the miss rate of the descending orbit was 51%; after the two were fused, the miss rate dropped to 19.9%.

[0061] This paper proposes a method for identifying earthquake-damaged buildings that leverages multi-temporal Sentinel-1 coherence and combines coherence estimation and histogram matching techniques for homogeneous point SAR pixel selection. The former fusion technique maintains the spatial resolution of coherence and reduces the impact of coherence estimation bias, while the latter mitigates the impact of coherence map differences caused by inconsistent adjacent baselines. The combination of these two techniques improves identification accuracy.

[0062] The present invention combines ascending and descending SAR images to solve the accuracy problem caused by geometric distortion of SAR side-view imaging.

[0063] The present invention uses Sentinel-1 SAR data from one year before the earthquake and the first scene after the earthquake (co-seismic), and can quickly and accurately identify earthquake-damaged buildings without relying on prior information about damaged buildings after the earthquake. This method is applicable to all places with stable coherence around the world and has high universality.

[0064] Device embodiment

[0065] The present invention provides a device for identifying earthquake-damaged buildings, comprising:

[0066] The image acquisition and preprocessing module is used to obtain the pre-seismic SAR images and co-seismic SAR images of the ascending and descending orbits taken by spaceborne SAR over the study area, and to perform interferometric processing on the pre-seismic SAR images and co-seismic SAR images of the ascending and descending orbits, respectively, to obtain the pre-seismic time-series intensity maps, pre-seismic time-series differential interferograms, and co-seismic differential interferograms of the ascending and descending orbits.

[0067] The first processing module is configured to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits using pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels, using a coherence estimation method based on homogeneous point selection, to obtain the pre-seismic time series and co-seismic estimated coherence maps of the ascending and descending orbits; and then perform corresponding histogram matching based on the coherence maps to obtain the pre-seismic time series and co-seismic enhancement coherence maps of the ascending and descending orbits.

[0068] The second processing module is used to perform pixel-by-pixel subtraction between the pre-earthquake predicted coherence maps of the ascending and descending orbits and the pre-earthquake enhanced coherence maps, respectively, to obtain pre-earthquake time series coherence difference maps of the ascending and descending orbits. Furthermore, the reliability maps and background effect removal thresholds of the ascending and descending orbits are generated respectively based on the pre-earthquake time series predicted coherence maps and pre-earthquake time series coherence difference maps of the ascending and descending orbits.

[0069] The third processing module is used to obtain the coseismic coherence difference map of the ascending track and the descending track by subtracting the pixel value of the pre-earthquake prediction coherence map closest to the earthquake from the coseismic enhancement coherence map of the ascending track and the descending track respectively.

[0070] The fourth processing module is used to remove the background effect from the coseismic coherence difference maps of the ascending and descending orbits according to the background effect removal thresholds of the ascending and descending orbits, respectively, to obtain the result maps of the ascending and descending orbits; and to obtain the building damage proxy maps of the ascending and descending orbits by taking the common intersection of the result maps of the ascending and descending orbits, the reliability maps, and the existing building distribution maps.

[0071] The fifth processing module is used to fuse the ascending and descending building damage proxy maps to obtain the final building damage proxy map.

[0072] Device Example

[0073] The present invention provides a computer device comprising a memory and a processor, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory to implement a method for identifying earthquake-damaged buildings as described above, wherein the method has been described in detail in the method embodiment and will not be repeated here.

[0074] Storage medium embodiment

[0075] The present invention provides a computer-readable storage medium for storing computer-executable instructions. When the computer-executable instructions are executed by a processor, they are used to implement a method for identifying earthquake-damaged buildings as described above. The method has been described in detail in the method embodiment and will not be repeated here.

[0076] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0077] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0078] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0079] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A functional step specified in one or more boxes.

[0080] It should be noted that the specific embodiments described above can enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although this specification and examples have described the present invention in detail, those skilled in the art should understand that the present invention can still be modified or replaced with equivalents; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are included in the scope of protection of the patent for the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A method for identifying earthquake-damaged buildings, characterized in that: The steps include: The pre-earthquake SAR images and co-seismic SAR images taken by spaceborne SAR in the study area were obtained in ascending and descending orbits, and the pre-earthquake time-series intensity maps, pre-earthquake time-series differential interferograms, and co-seismic differential interferograms were obtained after interferometric processing of the pre-earthquake SAR images and co-seismic SAR images in ascending and descending orbits. Using pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels, and using a coherence estimation method based on homogeneous point selection, the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits is estimated to obtain pre-seismic time series and co-seismic predicted coherence maps of the ascending and descending orbits; and performing histogram matching on the pre-seismic time series and co-seismic predicted coherence maps of the ascending and descending orbits to obtain pre-seismic time series and co-seismic enhanced coherence maps of the ascending and descending orbits. The pre-earthquake predicted coherence map of the ascending and descending orbits is subtracted from the pre-earthquake enhanced coherence map pixel by pixel, respectively, to obtain the pre-earthquake time series coherence difference map of the ascending and descending orbits. The reliability map and background effect removal threshold of the ascending and descending orbits are generated respectively based on the pre-earthquake time series predicted coherence map and pre-earthquake time series coherence difference map of the ascending and descending orbits; The pixel-by-pixel difference between the pre-earthquake prediction coherence map of the ascending and descending orbits closest to the earthquake and the coseismic enhancement coherence map is obtained; Based on the background effect removal thresholds for ascending and descending orbits, the background effect of the coseismic coherence difference maps of the ascending and descending orbits is removed to obtain the result maps of the ascending and descending orbits. The building damage proxy maps of the ascending and descending orbits are obtained by taking the common intersection of the result maps of the ascending and descending orbits, the reliability map, and the existing building distribution map. The building damage proxy maps of the ascending and descending orbits are fused to obtain the final building damage proxy map.

2. The earthquake-damaged building identification method according to claim 1, characterized in that: The step of fusing the ascending and descending building damage proxy maps to obtain a final building damage proxy map comprises: According to the slope, aspect, elevation and satellite orbit direction of the study area, an orbit that is relatively less affected by geometric distortion is determined as the dominant orbit; Obtain the union of the building damage proxy maps of the ascending and descending tracks. If there is a building damage signal at the same identification position on both the ascending and descending tracks, only retain the dominant track signal and output it to the final building damage proxy map.

3. The earthquake-damaged building identification method according to claim 1, characterized in that: The step of estimating the coherence of the pre-earthquake time series and co-seismic differential interferograms of the ascending and descending orbits using a coherence estimation method based on homogeneous point selection based on pixels in the pre-earthquake time series intensity maps of the ascending and descending orbits as reference pixels comprises: Each pixel in the pre-earthquake intensity map covering the study area of ​​the ascending and descending orbits is used as a reference pixel. The non-parametric statistical Baumgartner-Weiβ-Schindler test is used to compare the reference pixels in the pre-earthquake intensity map with the adjacent time series pixels to be judged in a preset size window. For each reference pixel in the window, a statistically homogeneous pixel is selected within the preset size window. The homogeneous pixel information is used to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits.

4. The earthquake-damaged building identification method according to claim 1, characterized in that: Regions in the co-seismic coherence difference maps of the ascending orbit and the descending orbit where the difference is greater than the background effect removal threshold are retained and output respectively, to obtain result maps of the ascending orbit and the descending orbit after removing the background effect.

5. The earthquake-damaged building identification method according to claim 1, characterized in that: The step of generating the reliability graphs of the ascending orbit and the descending orbit by using the pre-earthquake time series prediction coherence graphs of the ascending orbit and the descending orbit comprises: Calculate the mean and standard deviation of the pre-earthquake time series prediction coherence maps for ascending and descending orbits respectively; The statistical mean of the standard deviation within the building area in the study area is used as the threshold, and the areas with standard deviation less than the threshold are output as reliable areas, thereby obtaining the reliability maps of the ascending and descending orbits.

6. The earthquake-damaged building identification method according to claim 1, characterized in that: The sum of the average value and 1 standard deviation of the pre-earthquake time series coherence difference map of the ascending and descending orbits is used as the background effect removal threshold of the ascending and descending orbits, respectively.

7. The earthquake-damaged building identification method according to claim 1, characterized in that: The coseismic coherence difference maps of the ascending and descending orbits are obtained by subtracting the pixel-by-pixel pre-earthquake prediction coherence maps of the ascending and descending orbits closest to the earthquake from the coseismic enhancement coherence maps.

8. A device for identifying earthquake-damaged buildings, characterized in that: include: The image acquisition and preprocessing module is used to obtain the pre-earthquake SAR images and co-seismic SAR images of the study area taken by the spaceborne SAR in the ascending and descending orbits, and to perform interferometric processing on the pre-earthquake SAR images and co-seismic SAR images in the ascending and descending orbits, respectively, to obtain the pre-earthquake time-series intensity maps, pre-earthquake time-series differential interferograms, and co-seismic differential interferograms in the ascending and descending orbits; A first processing module is configured to estimate the coherence of the pre-seismic time series and co-seismic differential interferograms of the ascending and descending orbits using a coherence estimation method based on homogeneous point selection, using pixels in the pre-seismic time series intensity maps of the ascending and descending orbits as reference pixels, thereby obtaining pre-seismic time series and co-seismic estimated coherence maps of the ascending and descending orbits; and further perform corresponding histogram matching based on the coherence maps to obtain pre-seismic time series and co-seismic enhancement coherence maps of the ascending and descending orbits; The second processing module is configured to perform pixel-by-pixel subtraction between the pre-earthquake predicted coherence maps of the ascending and descending orbits and the pre-earthquake enhanced coherence maps, respectively, to obtain pre-earthquake time series coherence difference maps of the ascending and descending orbits, and then generate reliability maps and background effect removal thresholds for the ascending and descending orbits respectively based on the pre-earthquake time series predicted coherence maps and pre-earthquake time series coherence difference maps of the ascending and descending orbits; The third processing module is used to obtain the coseismic coherence difference map of the ascending and descending orbits by performing pixel-by-pixel difference between the pre-earthquake prediction coherence map of the ascending and descending orbits closest to the earthquake and the coseismic enhancement coherence map; The fourth processing module is configured to remove the background effect from the coseismic coherence difference maps of the ascending and descending orbits according to the background effect removal thresholds of the ascending and descending orbits, respectively, to obtain the result maps of the ascending and descending orbits; and obtain the building damage proxy maps of the ascending and descending orbits by taking the common intersection of the result maps of the ascending and descending orbits, the reliability map, and the existing building distribution map; The fifth processing module is used to fuse the ascending and descending building damage proxy maps to obtain the final building damage proxy map.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory to implement the earthquake-damaged building identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the earthquake-damaged building identification method according to any one of claims 1 to 7.

Images