A new method for extracting earthquake damage information of buildings based on frequency domain texture features

Abstract

The invention discloses a novel frequency domain texture feature extraction method for earthquake damage information of buildings. The improved Yamaguchi four-component decomposition (IYFD) method is used to polarize the fully polarized synthetic aperture radar (PolSAR) data, extract the secondary scattering component and the volume scattering component, and classify the to-be-identified ground objects in the PolSAR image corresponding to the secondary scattering component as parallel intact buildings. In order to improve the disadvantage of over-classification of collapsed buildings caused by the use of the traditional IYFD method, the invention proposes a LARTF parameter that can describe the texture feature of the Fourier transform of the frequency domain amplitude spectrum. The proposed LARTF parameter can better distinguish between collapsed buildings and azimuth buildings, which are similar ground objects with dominant volume scattering. The LARTF parameter can utilize the frequency domain characteristics to well show the texture difference of the spatial distribution of collapsed buildings and azimuth buildings. Therefore, the LARTF parameter is used to sort out a large number of azimuth buildings mixed in the collapsed buildings identified by the traditional IYFD method to improve the accuracy of earthquake damage identification of buildings.

Description

Technical Field

[0001] The patent of this invention relates to the technical field of building earthquake damage information extraction, specifically a new method for extracting building earthquake damage information based on frequency domain texture features. Background Art

[0002] In the past decade, earthquakes have occurred frequently around the world, and earthquake disasters with great destructive power have caused huge losses to human life and property. Post-earthquake casualties are mainly caused by the collapse of buildings. When searching and rescuing people buried in the rubble, the sooner the information of building collapse is obtained, the more hope there is for survival. Therefore, it is the primary task to quickly and accurately count the distribution of collapsed buildings in the earthquake zone after the earthquake, and it is also the guideline for deploying other earthquake emergency work. The main means of obtaining traditional earthquake damage information is on-site field surveys. Although this method has high credibility in obtaining information, it often has shortcomings such as large workload and low efficiency when carried out in a large area. Remote sensing technology can quickly monitor disasters over a large area and has become an important means of post-earthquake disaster assessment. Earthquakes often occur at night, or are often accompanied by bad weather such as clouds, rain and snow, so optical remote sensing cannot play a role. Radar waves have strong penetration ability and are not affected by sunlight, so information about the disaster area can be obtained at any time. Therefore, after an earthquake occurs, it is more secure to use radar remote sensing to monitor the disaster. In the past, it was often after an earthquake that multi-phase SAR data before and after the earthquake was collected, and the difference between the post-earthquake and pre-earthquake periods was compared to determine the disaster situation. However, collecting pre-earthquake SAR data is time-consuming and laborious. The present invention only uses post-earthquake single-phase SAR data to identify earthquake damage to buildings in the disaster area, which saves time and effort, and can also avoid multi-phase data registration operations. The amount of information contained in the fully polarized radar is far richer than the radar remote sensing data in single-polarization and dual-polarization modes, because PolSAR contains four polarization channels: HH, HV, VH and VV, where H represents horizontal polarization and V represents vertical polarization. Therefore, when only post-earthquake single-scene SAR data can be used to assess earthquake damage to buildings, choosing PolSAR data as the data source will increase the accuracy of earthquake damage identification and make the earthquake damage assessment results more reliable.

[0003] In post-earthquake SAR images, the walls of many collapsed buildings collapsed, causing the dihedral structure formed by the wall and the ground to be destroyed. The scattering mechanism is no longer dominated by secondary scattering, but by volume scattering, which makes the scattering intensity of collapsed buildings weaker than that of intact buildings. However, there is also a special phenomenon in SAR images. The arrangement direction of some intact buildings is not parallel to the flight direction of the radar. This type of building is called azimuth building. This type of building has a strong depolarization effect, and its dominant scattering mechanism is volume scattering, and the scattering intensity is weak. Therefore, azimuth buildings and collapsed buildings have similar scattering mechanisms and scattering intensities, and the mutual confusion phenomenon in SAR images is relatively serious, which often causes collapsed buildings to be over-evaluated. In response to this problem, the present invention proposes a novel frequency domain texture feature based on Fourier transform, namely the Fourier amplitude spectrum sector texture feature index (LARTF), and uses the LARTF parameter to design a building earthquake damage identification scheme. Experimental results show that the LARTF feature parameters proposed by the present invention can significantly improve the accuracy of earthquake damage identification. The following will specifically introduce our research design and experiments.

[0004] Invention patent content

[0005] The purpose of the present invention is to provide a new method for extracting earthquake damage information of buildings based on frequency domain texture features. To achieve the above purpose, the present invention provides the following technical solution: a new method for extracting earthquake damage information of buildings based on frequency domain texture features, the method comprising the following steps:

[0006] S1 uses the improved Yamaguchi four-component decomposition (IYFD) method to perform polarization decomposition of PolSAR data and extract secondary scattering components and volume scattering components;

[0007] S2 classifies the unidentified ground objects in the PolSAR image corresponding to the secondary scattering component as parallel intact buildings;

[0008] S3 calculates and extracts the SPAN image of PolSAR data;

[0009] S4 sets a suitable calculation window, traverses the entire SPAN image according to the LARTF parameter calculation method shown in the formula, calculates the LARTF value, and finally obtains the LARTF image of the entire PolSAR data;

[0010] S5 selects a suitable threshold for LARTF and divides the volume scattering components obtained by polarization decomposition into collapsed buildings and azimuth buildings according to the classification rules shown in the formula;

[0011] S6 merges the oriented buildings and parallel intact buildings into non-collapsed buildings and directly defines the collapsed building identification results as the collapsed building category.

[0012] Preferably, in S1, the IYFD method is generated by improving the Yamaguchi four-component decomposition model in order to improve the recognition accuracy of the azimuth building.

[0013] Preferably, in S2, the volume scattering components produced by the polarization decomposition method are mixed with azimuth buildings and collapsed buildings. Although their scattering mechanisms and scattering intensities are similar, they still have different texture representation phenomena in the PolSAR image. The use of frequency domain texture features can better reflect the scattering intensity texture difference of the two types of buildings. The characteristic parameter LARTF is applied to classify and identify the two types of earthquake zone buildings to improve the over-classification phenomenon of collapsed buildings and suppress the over-correction of azimuth buildings.

[0014] Preferably, in S3, the total power image data of PolSAR (called SPAN image) contains the intensity information of all polarization channels, and the calculation method of SPAN is as follows:

[0015] SPAN=HH 2 +HV 2 +VH 2 +VV 2 .

[0016] Preferably, in S4, the LARTF parameter calculation formula is as follows:

[0017]

[0018] FWI=FFT(WI)

[0019]

[0020] RDFT=mean[std(AFWI_R1),std(AFWI_R2),…,std(AFWI_R m )]

[0021] LARTF = 10 × lg (RDFT + 3 × ADFT)

[0022] Preferably, the classification rule of LARTF in S5 can be expressed by the formula:

[0023] x∈volume_dominated buildings

[0024] if LARTF≤ε

[0025] x∈collapsed buildings.

[0026] if LARTF>ε

[0027] x∈obliquely-oriented standing buildings

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] In order to improve the drawback of over-classification of collapsed buildings caused by the use of traditional IYFD methods, the present invention proposes a texture feature LARTF parameter that can describe the Fourier transform amplitude spectrum. The proposed LARTF parameter can better distinguish between collapsed buildings and azimuth buildings, two similar landforms that are both dominated by volume scattering. The LARTF parameter can utilize frequency domain characteristics to well show the texture difference in spatial distribution of collapsed buildings and azimuth buildings. Therefore, the present invention uses the LARTF parameter to sort out a large number of azimuth buildings mixed in the collapsed buildings identified by the traditional IYFD method to improve the accuracy of building earthquake damage identification, calculates the LARTF image corresponding to the SPAN image of the PolSAR data, and divides the volume scattering component produced by IYFD on the PolSAR data according to the LARTF parameter value, and further divides it into collapsed buildings and azimuth buildings, which greatly improves the problem of over-identification of disasters that directly divides all the volume scattering components produced by IYFD into collapsed buildings. It can also be seen from the experimental results that the building earthquake damage information extraction method proposed in the present invention can better identify the collapsed buildings and non-collapsed buildings in the earthquake zone, and obtain a relatively high building earthquake damage identification accuracy. It successfully stripped 70.18% of the azimuth buildings from the collapsed building identification results produced by the traditional IYFD method, significantly improving the building earthquake damage information extraction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0031] Figure 1 This is a diagram showing the ability of the LARTF parameters of the present invention to distinguish between oriented buildings and collapsed buildings;

[0032] Figure 2 This is the LARTF calculation principle diagram of the present invention;

[0033] Figure 3 Figure 1 is a process diagram of extracting collapsed buildings and non-collapsed buildings according to the present invention;

[0034] Figure 4 The SPAN image of the urban area of ​​Yushu County, which marks the verification samples of three types of earthquake zone buildings for the present invention;

[0035] Figure 5 Figure 1 is a diagram showing the selection of the LARTF parameter calculation window of the present invention;

[0036] Figure 6 AFWI of the present invention shows Figure 2 Figure 2 shows the selection of the number n of equal parts of the angular domain of a 2π circle;

[0037] Figure 7 For the present invention Figure 2 Figure 1 shows the selection of the number of equal-width rings m for dividing the inscribed circle of AFWI;

[0038] Figure 8 FIG. 1 is a diagram showing the determination of the LARTF threshold of the present invention;

[0039] Fig. 9 The figure is a diagram of the extraction results of three types of buildings in Yushu County after the earthquake of the present invention. DETAILED DESCRIPTION

[0040] The following will combine the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] A novel method for extracting building earthquake damage information based on frequency domain texture features comprises the following steps:

[0042] S1 uses the improved Yamaguchi four-component decomposition (IYFD) method to perform polarization decomposition of PolSAR data and extract secondary scattering components and volume scattering components;

[0043] S2 classifies the unidentified ground objects in the PolSAR image corresponding to the secondary scattering component as parallel intact buildings;

[0044] S3 calculates and extracts the SPAN image of PolSAR data;

[0045] S4 sets a suitable calculation window, traverses the entire SPAN image according to the LARTF parameter calculation method shown in the formula, calculates the LARTF value, and finally obtains the LARTF image of the entire PolSAR data;

[0046] S5 selects a suitable threshold for LARTF and divides the volume scattering components obtained by polarization decomposition into collapsed buildings and azimuth buildings according to the classification rules shown in the formula;

[0047] S6 merges the oriented buildings and parallel intact buildings into non-collapsed buildings and directly defines the collapsed building identification results as the collapsed building category.

[0048] In the post-earthquake PolSAR images, some intact buildings are parallel to the radar flight direction, which we call parallel intact buildings in the present invention. The dominant scattering component of this type of building (dominant scattering component) is a secondary scattering component, which can be directly produced by polarization decomposition. The scattering mechanism of the collapsed buildings in the earthquake zone and the intact buildings as azimuth buildings are dominated by volume scattering. The volume scattering components produced by the polarization decomposition method are mixed with azimuth buildings and collapsed buildings. Although their scattering mechanisms and scattering intensities are similar, they still have different texture characterization phenomena in the PolSAR images. The strength of the Fourier amplitude spectrum of the SAR image can show the speed of change of the scattering intensity. The azimuth building texture is more regular and has more fine textures, while the collapsed building has relatively more coarse textures and more messy textures. The use of frequency domain texture features can better reflect the difference in scattering intensity texture between the two types of buildings. In an embodiment of the present invention, the feature parameter LARTF is applied to classify and identify the two types of earthquake zone buildings to improve the over-classification phenomenon of collapsed buildings and suppress the over-correction of azimuth buildings. From Figure 1 It can be seen that the LARTF value of the azimuth building is greater than that of the collapsed building. LARTF has a good recognition performance for these two types of buildings. The calculation method and process of LARTF are as follows: Figure 2 As shown in the figure, the SPAN image has the richest intensity information. Therefore, the SPAN image is used to calculate the LARTF of the PolSAR data. The calculation method of SPAN is as follows:

[0049] SPAN=HH 2 +HV 2 +VH 2 +VV 2 (1)

[0050] Use the calculation window to traverse the entire SPAN image, calculate the LARTF value of each calculation window, and then obtain the LARTF data of the entire PolSAR image. The following takes the calculation process of the LARTF value in the calculation window as an example to explain the LARTF parameter calculation method;

[0051] WI is the SPAN image in the calculation window. First, perform a two-dimensional fast Fourier transform (FFT) on WI to obtain FWI. Second, calculate the amplitude spectrum of FWI to obtain AFWI. The third step is as follows: Figure 2 As shown in the figure, take the center point of the AFWI image as the coordinate origin, divide the circle 2π into n equal parts around the coordinate origin, the vertex angle value of the coordinate origin of each equal part is 2π / n radians, calculate the sum of the AFWI values ​​of all points in the i-th equal part, and get AFWI_A i, calculate the sum of the AFWI values ​​of each equal part in turn, and get n AFWI_A i Values ​​{AFWI_A1,AFWI_A2,AFWI_A3,…,AFWI_A n}, the fourth step is to calculate the ADFT value of AFWI according to the fourth line of formula (2), the fifth step is as follows Figure 2 As shown in the figure, the center point of the AFWI image is taken as the coordinate origin, and the inscribed circle of the AFWI is divided into m rings with equal ring width. The width of each ring is RW. The sum of the AFWI values ​​of all points in the jth ring is calculated to obtain AFWI_R j , calculate the sum of the AFWI values ​​of all points in each ring in turn, and get m AFWI_R j Values ​​{AFWI_R1,AFWI_R2,AFWI_R3,…,AFWI_R m}, Step 6, calculate the RDFT value of AFWI according to the fifth line of formula (2), and finally, calculate the LARTF parameter value of WI according to the last line of formula (2);

[0052] The calculation process of LARTF can be expressed as:

[0053]

[0054] FWI=FFT(WI)

[0055]

[0056] RDFT=mean[std(AFWI_R1),std(AFWI_R2),...,std(AFWI_R m )]

[0057] LARTF = 10 × lg (RDFT + 3 × ADFT)

[0058] In the formula, real and imag represent the real part and imaginary part respectively, mean, std and lg represent the calculation of mean, variance and logarithm with base 10 respectively;

[0059] like Figure 3 As shown in Figure 1, the process of extracting building damage information mainly includes the following six steps: first, the IYFD method is used to perform polarization decomposition on the PolSAR data to extract the secondary scattering component and the volume scattering component; second, the objects to be identified in the PolSAR image corresponding to the secondary scattering component are classified as parallel intact buildings; third, the SPAN image of the PolSAR data is calculated and extracted; fourth, an appropriate calculation window is set according to formula (2) and Figure 2The LARTF parameter calculation method shown in the figure traverses the entire SPAN image, calculates the LARTF value, and finally obtains the LARTF image of the entire PolSAR data. Fifth, select a suitable threshold for LARTF, and divide the volume scattering components obtained by polarization decomposition into collapsed buildings and azimuth buildings according to the classification rule shown in formula (3). Figure 1 It can be seen that the LARTF value of the oriented building is larger than that of the collapsed building. Therefore, the classification rule of LARTF can be expressed as follows:

[0060] x∈volume_dominated buildings

[0061] if LARTF≤ε

[0062] x∈collapsed buildings(3)

[0063] if LARTF>ε

[0064] x∈obliquely-oriented standing buildings

[0065] The ε in the formula represents the threshold for dividing collapsed buildings and azimuth buildings using the LARTF parameter. Finally, the azimuth buildings and parallel intact buildings are merged into non-collapsed buildings, and the recognition results of collapsed buildings are directly defined as collapsed building categories.

[0066] Example:

[0067] The present invention uses the 4.14Ms7.1 Yushu earthquake that occurred in Yushu County, Qinghai Province, China on April 14, 2010 as a research earthquake case. The longitude and latitude of the epicenter of this earthquake are located at (33.1°N, 96.6°E). Yushu County has a high altitude, a dry and cold climate, and there is little vegetation in the urban area. The vegetation is sparsely distributed and mostly low-lying, which has little impact on the identification of buildings in the earthquake zone. Therefore, the vegetation land type will be ignored in this experiment. The boundary shp data of Yushu County is used to mask out the mountains around the urban area, and only the building earthquake damage information extraction experiment is carried out for the urban area;

[0068] The experimental data are airborne high-resolution PolSAR images acquired one day after the earthquake using China's airborne SAR mapping system, with a spatial resolution of 1 meter. Figure 4 The SPAN image of PolSAR data used to calculate LARTF parameters is given. The image size is 8192×4384 pixels. In this experiment, 25,000 experimental samples were selected from collapsed buildings, azimuth buildings, and parallel intact buildings. Figure 4 They are marked in red, blue and green respectively.

[0069] Experimental Results

[0070] according to Figure 3 The building earthquake damage identification process shown in the figure extracts collapsed buildings and non-collapsed buildings in the PolSAR data used for the experiment of the present invention. Figure 2 The method flow shown in the figure calculates the classification parameter LARTF. In order to simplify the experimental calculation, the calculation window of LARTF is set to a square. Figure 5 It can be seen that when the calculation window of size 57×57 is used to calculate the LARTF parameter value according to formula (2), the overall classification accuracy of the three types of buildings, namely collapsed buildings, oriented buildings and parallel intact buildings, is the highest. Figure 6 As shown, in Figure 2 In the ADFT parameter calculation process shown in the figure, when the number of equal divisions n of the 2π circumference of the image block AFWI is 36, that is, when the vertex angle value of each equal angular domain is π / 18 radians, the overall classification accuracy of the three types of buildings reaches the highest. Figure 7 As shown, according to Figure 2 When calculating RDFT, the ring width RW is set to 5, and the classification results of the three buildings achieve the highest overall classification accuracy. At this time, the number of equal-width rings divided by the inscribed circle of AFWI is m = WS / (2×RW), where WS represents the side length of the calculation window. Figure 1 It can be seen that there is a clear dividing line between the LARTF value of the collapsed building and the LARTF value of the oriented building at -8.22. Figure 8 It can also be further confirmed in that when LARTF is equal to -8.22, the overall classification accuracy of the three types of buildings is the highest. Therefore, when using the LARTF parameter to classify collapsed buildings and oriented buildings according to formula (3), the classification threshold ε is set to -8.22.

[0071] The results of extracting building damage information in this experimental earthquake case are as follows: Fig. 9 As shown, Fig. 9 This is the experimental result of masking the mountains around the urban area of ​​Yushu County, showing only the earthquake damage to urban buildings. Figure 4 The verification samples of three types of buildings, namely, collapsed buildings, azimuth buildings and parallel intact buildings, are Fig. 9 The accuracy of the earthquake damage identification results of the buildings shown in the figure is evaluated, and the accuracy evaluation confusion matrix of the experimental results is listed in Table 1. It can be seen from Table 1 that the overall correct identification accuracy of the three types of buildings in the earthquake zone using the method proposed by the present invention reaches 81.3%.

[0072] Table 1. Confusion matrix of accuracy evaluation of three building recognition results

[0073]

[0074] OA, CB, OB, and PB represent the overall identification accuracy, collapsed buildings, oriented buildings, and parallel intact buildings, respectively;

[0075] Experimental analysis

[0076] From Table 1, it can be concluded that the correct recognition rate of collapsed buildings obtained by extracting building earthquake damage information using the method proposed in the present invention is 81.06%; the comprehensive correct recognition rate of azimuth buildings and parallel intact buildings is 81.42%. That is to say, the earthquake damage recognition accuracy of collapsed buildings and non-collapsed buildings is above 80%. This recognition result is relatively balanced for collapsed buildings and non-collapsed buildings. It neither causes an overestimation of collapsed buildings that leads to an underestimation of non-collapsed buildings, nor an underestimation of the number of collapsed buildings that leads to an overestimation of the number of non-collapsed buildings. Therefore, it can be said that the recognition result produced by the method proposed in the present invention is safe and reliable for post-earthquake building earthquake damage assessment, because excessive recognition of collapsed buildings will lead to a waste of rescue resources, and insufficient recognition of collapsed buildings will delay the timely rescue of buried people, which is very dangerous.

[0077] From the perspective of the number of correctly identified buildings, 70.18% of the buildings were successfully separated from the volume scattering components. If these buildings were misclassified as collapsed buildings, it would mislead a lot of rescue manpower and materials. The traditional IYFD method classifies all the corresponding objects of the volume scattering components as collapsed buildings. Although this method can achieve a 99.92% recognition rate for collapsed buildings, the correct recognition rate for non-collapsed buildings is only 46.33%. Therefore, when using the traditional IYFD method to identify building damage, the inaccurate classification of the volume scattering components will result in an uncontrolled sacrifice of the correct recognition rate of non-collapsed buildings in exchange for the correct recognition rate of collapsed buildings.

[0078] It should be noted in this example that:

[0079] Regarding the English comments in the figures and formulas:

[0080] SPAN image of PolSAR data SPAN image of PolSAR data

[0081] Volume_dominated buildings Buildings where the dominant scattering mechanism is volume scattering

[0082] Collapsed buildings

[0083] Obliquely-oriented buildings / Obliquely-oriented standing buildings Orientation buildings

[0084] Parallel standing buildings / Parallel buildings

[0085] Threshold line

[0086] Smaple number

[0087] Calculation window of SPAN image

[0088] 2D fast Fourier Transform

[0089] Compute amplitude value

[0090] Window size calculates the window size

[0091] Overall accuracy

[0092] Number of angular zones in the circle

[0093] Ring width

[0094] Threshold value.

[0095] Although the embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A novel frequency domain texture feature-based building earthquake damage information extraction method, characterized in that: The method comprises the following steps: S1 uses the improved Yamaguchi four-component decomposition (IYFD) method to perform polarization decomposition on the fully polarized synthetic aperture radar (PolSAR) data to extract the secondary scattering component and the volume scattering component; S2 classifies the unidentified ground objects in the PolSAR image corresponding to the secondary scattering component as parallel intact buildings; S3 calculates and extracts the SPAN image of PolSAR data; S4 sets the calculation window, traverses the entire SPAN image according to the LARTF parameter calculation method shown in the formula, calculates the LARTF value, and finally obtains the LARTF image of the entire PolSAR data; S5 is the LARTF selection threshold, and the volume scattering components obtained by polarization decomposition are divided into collapsed buildings and azimuth buildings according to the classification rules shown in the formula; S6 merges the azimuth buildings and parallel intact buildings into non-collapsed buildings, and directly defines the collapsed building identification results as the collapsed building category; In the post-earthquake PolSAR images, some intact buildings are arranged in a direction parallel to the radar flight direction, which are called parallel intact buildings; some intact buildings are arranged in a direction not parallel to the radar flight direction, which are called azimuth buildings; In S3, the SPAN image of PolSAR contains the intensity information of all polarization channels. The calculation method of SPAN is as follows: SPAN=HH 2 +HV 2 +VH 2 +VV 2 ; In S4, the LARTF parameter calculation formula is as follows: FWI=FFT(WI) ADFT=std[mean(AFWI_A1),mean(AFWI_A2),...,mean(AFWI_A n )] RDFT=mean[std(AFWI_R1),std(AFWI_R2),...,std(A,WI_R m )] LARTF=10×lg(RDFT+3×ADFT); Among them, real and imag represent the real part and imaginary part respectively; mean, std and lg represent the calculation of the mean, variance and logarithm with base 10 respectively. WI is the SPAN image in the calculation window. First, perform a two-dimensional fast Fourier transform on WI to obtain FWI. Secondly, calculate the amplitude spectrum of FWI to obtain AFWI. Take the center point of the AFWI image as the coordinate origin and divide the circle 2π into n equal parts around the coordinate origin. The vertex angle value of the coordinate origin of each part is 2π / n radians. Calculate the sum of the AFWI values ​​of all points in the i-th part to obtain AFWI_Ai. Calculate the sum of the AFWI values ​​of each part in turn to obtain n AFWI_Ai values ​​{AFWI_A1, AFWI_A2, AFWI_A3, …, AFWI_An}. Taking the center point of the AFWI image as the coordinate origin, divide the inscribed circle of AFWI into m rings with equal ring width, and the ring width of each ring is RW. Calculate the sum of the AFWI values ​​of all points in the jth ring to get AFWI_Rj, and calculate the sum of the AFWI values ​​of all points in each ring in turn to get m AFWI_Rj values ​​{AFWI_R1, AFWI_R2, AFWI_R3, …, AFWI_Rm}.

2. The novel frequency domain texture feature-based building earthquake damage information extraction method according to claim 1 is characterized by: The classification rule of LARTF in S5 can be expressed as follows: Where ε represents the threshold for dividing collapsed buildings and oriented buildings using LARTF parameters; Volume_dominatedbuildings represents buildings whose dominant scattering mechanism is volume scattering, Collapsed buildings represent collapsed buildings, and Obliquely-oriented standing buildings represent oriented buildings.

Images