Landslide area detection method and system based on radar remote sensing data

Abstract

The invention provides a landslide area detection method and system based on radar remote sensing data. The method comprises the following steps: extracting a backscattering coefficient and a coherence coefficient when a detection radar detects a target scanning area under each time phase from the radar remote sensing data; determining the earth surface deformation sensitivity of each pixel point in the target scanning area based on a scattering characteristic matrix constructed by the backscattering coefficient under each time phase in combination with the spatial-temporal change characteristics of the coherence coefficient; according to the singular value decomposition result of the scattering characteristic matrix and the earth surface deformation sensitivity of each pixel point, dividing a deformation sensitive area and a stability subarea; according to the scattering entropy of the pixel points in the deformation sensitive area, hidden landslide points of the target scanning area are identified from the stability subarea, and then a landslide risk probability graph is generated through all the hidden landslide points in combination with the deformation sensitive area. According to the technical scheme provided by the invention, the hidden landslide point in the scanning area can be effectively identified when the weak deformation signal is submerged.

Description

Technical Field

[0001] This application relates to the field of radio detection technology. More specifically, this application relates to a landslide area detection method and system based on radar remote sensing data. Background Art

[0002] With the rapid development of modern technology, radio plays a key role in many fields. Among them, radar radio detection technology is particularly important. Radar detects the position, speed, shape and other information of target objects by emitting radio waves and receiving their reflected signals. However, with the increasingly complex electromagnetic environment, the interference and clutter increase, which puts higher requirements on the accuracy and reliability of radar radio detection, prompting continuous innovation in related technologies to improve their performance and adaptability.

[0003] In existing radio detection, radio detection is mainly based on the propagation characteristics of electromagnetic waves for detection. Radar radio detection is a typical example. Radar radio detection emits electromagnetic waves of a specific frequency. When these electromagnetic waves encounter a target object, they will be reflected. The radar receives the reflected electromagnetic wave signals and determines the position, speed and distance information of the target according to the propagation time and frequency change of the signals, so as to realize the detection of the target. However, in the landslide area detection based on radar remote sensing data, the surface deformation in the initial stage of the landslide is weak and has non-linear characteristics. The single deformation parameter relied on by traditional methods is easily affected by environmental noise (such as vegetation coverage, humidity change), resulting in the weak deformation signal being submerged, making it difficult to distinguish the real deformation area from the pseudo-change area, resulting in the inability to effectively identify the hidden landslide points in the scanned area, and then reducing the effectiveness of landslide area detection. Therefore, how to effectively identify the hidden landslide points in the scanned area when the weak deformation signal is submerged, and then improve the effectiveness of landslide area detection has become a difficult problem faced by the industry. Summary of the Invention

[0004] This application provides a landslide area detection method and system based on radar remote sensing data, which can effectively identify the hidden landslide points in the scanned area when the weak deformation signal is submerged, and then improve the effectiveness of landslide area detection.

[0005] In the first aspect, this application provides a landslide area detection method based on radar remote sensing data, including the following steps:

[0006] Obtain the radar remote sensing data of the target scanned area at multiple time phases;

[0007] Extract the backscattering coefficient and coherence coefficient when the detection radar detects the target scanned area in the corresponding radar remote sensing data at each time phase;

[0008] Construct a scattering characteristic matrix of the radar wave by the surface medium in the target scanning area based on the backscattering coefficients at each time phase, and determine the surface deformation sensitivity of each pixel point in the target scanning area according to the temporal and spatial variation characteristics of the coherence coefficient combined with the scattering characteristic matrix;

[0009] Perform singular value decomposition on the scattering characteristic matrix, and divide the deformation sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point;

[0010] Determine the scattering entropy of the pixel points in the deformation sensitive area, identify the hidden landslide points in the target scanning area from the stability partition according to the scattering entropy, and then generate a landslide risk probability map by combining all the hidden landslide points with the deformation sensitive area.

[0011] In some embodiments, extracting the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at each time phase from the corresponding radar remote sensing data specifically includes:

[0012] Preprocess the radar remote sensing data at each time phase to generate a single-look complex radar image of the target scanning area at each time phase;

[0013] Select a time phase as the selected time phase, and extract the backscattering coefficient when the detection radar detects the target scanning area at the selected time phase from the single-look complex radar image corresponding to the selected time phase according to the radiometric calibration function;

[0014] Register the single-look complex radar image corresponding to the selected time phase with the single-look complex radar image corresponding to the adjacent time phase to generate a pair of single-look complex radar images corresponding to the selected time phase;

[0015] Determine the coherence coefficient when the detection radar detects the target scanning area at the selected time phase according to the pair of single-look complex radar images;

[0016] Continue to determine the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at the remaining time phases.

[0017] In some embodiments, constructing a scattering characteristic matrix of the radar wave by the surface medium in the target scanning area based on the backscattering coefficients at each time phase specifically includes:

[0018] Obtain the backscattering coefficients at each time phase, and there is a backscattering coefficient corresponding to different pixel points in the target scanning area;

[0019] Arrange the backscattering coefficients at each time phase in a time series to generate a backscattering coefficient vector at each pixel point in the target scanning area;

[0020] The backscattering coefficient vectors at all pixel points are combined into a two-dimensional matrix according to the spatial dimension, and the scattering characteristic matrix of the surface medium in the target scanning area to the radar wave is obtained.

[0021] In some embodiments, determining the surface deformation sensitivity of each pixel point in the target scanning area according to the scattering characteristic matrix in combination with the spatio-temporal variation characteristics of the coherence coefficient specifically includes:

[0022] Based on the backscattering coefficient vector corresponding to each pixel point in the scattering characteristic matrix, determining the scattering fluctuation intensity corresponding to each pixel point;

[0023] Obtaining the coherence coefficient sequence of each pixel point within all time intervals;

[0024] Determining the spatio-temporal variation characteristics corresponding to each pixel point according to the coherence coefficient sequence corresponding to each pixel point;

[0025] For each pixel point, normalizing and weighted fusing the scattering fluctuation intensity corresponding to the pixel point with the spatio-temporal variation characteristics, thereby obtaining the surface deformation sensitivity at each pixel point in the target scanning area.

[0026] In some embodiments, dividing the deformation-sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point specifically includes:

[0027] Extracting the principal singular vector from the singular value decomposition result of the scattering characteristic matrix;

[0028] Determining the principal component projection value of each pixel point according to the principal singular vector in combination with the backscattering coefficient vector of each pixel point in the scattering characteristic matrix;

[0029] Determining the characteristic response map of the surface deformation in the target scanning area through the principal component projection value corresponding to each pixel point and the surface deformation sensitivity;

[0030] Based on the characteristic response map, dividing the deformation-sensitive area and the stability partition in the target scanning area.

[0031] In some embodiments, determining the scattering entropy of the pixel points in the deformation-sensitive area specifically includes:

[0032] Extracting the backscattering coefficient sequences of the pixel points in the deformation-sensitive area;

[0033] Selecting a pixel point as the selected pixel point, and determining the scattering entropy of the selected pixel point according to the backscattering coefficient sequence of the selected pixel point;

[0034] Continuing to determine the scattering entropy of the remaining pixel points in the deformation-sensitive area.

[0035] In some embodiments, radar remote sensing data of a target scanning area at multiple time phases is obtained through a detection radar.

[0036] In a second aspect, the present application provides a landslide area detection system based on radar remote sensing data for performing a landslide area detection method based on radar remote sensing data, including:

[0037] An acquisition module for acquiring radar remote sensing data of a target scanning area at multiple time phases;

[0038] A processing module for extracting the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at each time phase in the corresponding radar remote sensing data;

[0039] The processing module is further configured to construct a scattering characteristic matrix of the radar wave by the surface medium in the target scanning area based on the backscattering coefficient at each time phase, and determine the surface deformation sensitivity of each pixel point in the target scanning area according to the scattering characteristic matrix in combination with the spatio-temporal variation characteristics of the coherence coefficient;

[0040] The processing module is further configured to perform singular value decomposition on the scattering characteristic matrix, and divide the deformation sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point;

[0041] An execution module for determining the scattering entropy of the pixel points in the deformation sensitive area, identifying the hidden landslide points in the target scanning area from the stability partition according to the scattering entropy, and then generating a landslide risk probability map by combining all the hidden landslide points with the deformation sensitive area.

[0042] In a third aspect, the present application provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to obtain the code and execute the above-mentioned landslide area detection method based on radar remote sensing data.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned landslide area detection method based on radar remote sensing data is implemented.

[0044] The technical solutions provided by the disclosed embodiments of the present application have the following beneficial effects:

[0045] In the landslide area detection method and system based on radar remote sensing data provided by the present application, first, radar remote sensing data of a target scanning area at multiple time phases is obtained; second, the backscattering coefficient and coherence coefficient when a detection radar detects the target scanning area at each time phase are extracted from the corresponding radar remote sensing data; further, a scattering characteristic matrix of the surface medium in the target scanning area for radar waves is constructed based on the backscattering coefficients at each time phase, and the surface deformation sensitivity of each pixel point in the target scanning area is determined according to the scattering characteristic matrix in combination with the spatio-temporal variation characteristics of the coherence coefficient; then, the scattering characteristic matrix is subjected to singular value decomposition, and the deformation sensitive area and stability partition in the target scanning area are divided according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point; finally, the scattering entropy of the pixel points in the deformation sensitive area is determined, and the latent landslide points in the target scanning area are identified from the stability partition according to the scattering entropy, and then a landslide risk probability map is generated by combining all the latent landslide points with the deformation sensitive area.

[0046] Thus, the present application can effectively identify the latent landslide points in the scanning area when the weak deformation signal is submerged, thereby improving the effectiveness of landslide area detection; first, the backscattering coefficient and coherence coefficient when a detection radar detects the target scanning area at each time phase are extracted from the corresponding radar remote sensing data, which can effectively identify the surface deformation or landslide signs and reflect the stability and change degree of the surface scattering characteristics; second, a scattering characteristic matrix of the surface medium in the target scanning area for radar waves is constructed based on the backscattering coefficients at each time phase, and the surface deformation sensitivity of each pixel point in the target scanning area is determined according to the scattering characteristic matrix in combination with the spatio-temporal variation characteristics of the coherence coefficient to measure the degree index of the sensitivity of each pixel position in the target scanning area to surface deformation response, thereby effectively identifying those surface areas that show obvious anomalies or instability in the radar wave reflection characteristics, thus avoiding the problem of deformation signals being submerged caused by environmental noise interference; further, the scattering characteristic matrix is subjected to singular value decomposition, and the deformation sensitive area and stability partition in the target scanning area are divided according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point to pre-delineate the area with potential geological disaster risks and provide data support for subsequent landslide risk assessment, thereby realizing the forward movement and precision of disaster prevention and control work; then, the latent landslide points in the target scanning area are identified from the stability partition according to the scattering entropy of the pixel points in the deformation sensitive area to obtain the latent landslide risk area and avoid missing the monitoring of the landslide risk area with weak surface disturbance; finally, a landslide risk probability map is generated by combining all the latent landslide points with the deformation sensitive area; in summary, the technical solution provided by the present application can effectively identify the latent landslide points in the scanning area when the weak deformation signal is submerged, thereby improving the effectiveness of landslide area detection. Description of the Drawings

[0047] Figure 1 is an exemplary flowchart of a landslide area detection method based on radar remote sensing data shown in some embodiments of the present application;

[0048] Figure 2 is an exemplary flowchart of determining a scattering characteristic matrix shown in some embodiments of the present application;

[0049] Figure 3 is an exemplary flowchart of determining a deformation-sensitive area and a stability partition shown in some embodiments of the present application;

[0050] Figure 4 is a schematic structural diagram of a landslide area detection system based on radar remote sensing data shown in some embodiments of the present application;

[0051] Figure 5 is a schematic structural diagram of a computer device for implementing a landslide area detection method based on radar remote sensing data shown in some embodiments of the present application. Detailed implementation manners

[0052] To better understand the technical solutions of the present application, the technical solutions of the present application will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners.

[0053] Refer to Figure 1 , this figure is an exemplary flowchart of a landslide area detection method based on radar remote sensing data shown in some embodiments of the present application. The landslide area detection method 100 based on radar remote sensing data mainly includes the following steps:

[0054] In step S101, radar remote sensing data of a target scanning area at multiple time phases is acquired.

[0055] In specific implementation, radar remote sensing data of a target scanning area at multiple time phases is acquired through a detection radar. The multiple time phases refer to multiple time points. The radar remote sensing data represents the observation data obtained after the detection radar scans the target scanning area. The radar remote sensing data is radar remote sensing image data. Among them, the detection radar is a synthetic aperture radar.

[0056] It should be noted that in the present application, the target scanning area represents a landslide area where landslide risks may occur.

[0057] In step S102, the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at each time phase are extracted from the corresponding radar remote sensing data.

[0058] In some embodiments, the extraction of the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at each phase in the corresponding radar remote sensing data can be achieved by the following steps:

[0059] Preprocess the radar remote sensing data for each phase to generate a single-look complex radar image of the target scanning area at each phase;

[0060] Select a phase as the selected phase, and extract the backscattering coefficient when the detection radar detects the target scanning area at the selected phase from the single-look complex radar image corresponding to the selected phase according to the radiometric calibration function;

[0061] Register the single-look complex radar image corresponding to the selected phase with the single-look complex radar image corresponding to the adjacent phase to generate a pair of single-look complex radar images corresponding to the selected phase;

[0062] Determine the coherence coefficient when the detection radar detects the target scanning area at the selected phase according to the pair of single-look complex radar images;

[0063] Continue to determine the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at the remaining phases.

[0064] It should be noted that in this application, the backscattering coefficient is used to measure the backscattering ability of the ground surface to radar waves. The backscattering coefficients obtained at the same location at different times will vary due to factors such as soil moisture, vegetation status, and structural changes. Therefore, the determination of the backscattering coefficient can be used to identify surface deformation or landslide signs; in this application, the coherence coefficient is an index that measures the phase consistency of radar images obtained at two different times at the same pixel position. The coherence coefficient reflects the stability and change degree of the surface scattering characteristics and is an important characteristic parameter for identifying dynamic changes such as surface deformation and landslide activities.

[0065] In specific implementation, first, the radar remote sensing data at each time phase is preprocessed by using the existing radiometric calibration technology. The radar remote sensing data is a radar remote sensing image, and the preprocessed radar remote sensing image is used as the single-look complex radar image of the target scanning area at the corresponding time phase. Secondly, a time phase is selected as the selected time phase, and the gray values at each pixel point in the single-look complex radar image corresponding to the selected time phase are used as input variables and input into the radiometric calibration function. The output result of the radiometric calibration function is used as the backscattering coefficient at the pixel point. Furthermore, the backscattering coefficient when the detection radar detects the target scanning area at the selected time phase is obtained. The radiometric calibration function refers to a mathematical relationship that converts the original digital value of the radar remote sensing image into a radar backscattering coefficient with physical significance. Further, the single-look complex radar image corresponding to the selected time phase is registered with the single-look complex radar image corresponding to the adjacent time phase to generate a pair of single-look complex radar images corresponding to the selected time phase. Then, according to the pair of single-look complex radar images, the coherence coefficient when the detection radar detects the target scanning area at the selected time phase is determined, that is: for the pixel points at the same position in the pair of single-look complex radar images, with the pixel point as the center and within the range of a preset sliding window, the cross-correlation coefficient of the two single-look complex radar images within the sliding window is used as the coherence coefficient of the pixel point. Furthermore, the coherence coefficient when the detection radar detects the target scanning area at the selected time is obtained. Finally, according to the determination method of "determining the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at the selected time phase", the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at the remaining time phases are determined in the same way, which will not be elaborated here.

[0066] It should be noted that in this embodiment, the single-look complex radar image refers to an image that retains the complex information (amplitude and phase) of each pixel point in the radar echo data after single-view processing, and it is the most basic and most refined image format in the synthetic aperture radar processing flow; the pair of single-look complex radar images in this embodiment represents the combination of two adjacent single-look complex radar images.

[0067] In step S103, a scattering characteristic matrix of the radar wave by the surface medium in the target scanning area is constructed based on the backscattering coefficients at each time phase. According to the scattering characteristic matrix and the spatio-temporal variation characteristics of the coherence coefficient, the surface deformation sensitivity of each pixel point in the target scanning area is determined.

[0068] In some embodiments, refer to Figure 2 As shown, this figure is an exemplary flowchart for determining the scattering characteristic matrix according to some embodiments of the present application. In this embodiment, the construction of the scattering characteristic matrix of the radar wave by the surface medium in the target scanning area based on the backscattering coefficients at each time phase can be implemented by the following steps:

[0069] In step S1031, the backscattering coefficients at each time phase are obtained, and there is a backscattering coefficient corresponding to different pixel points in the target scanning area;

[0070] In step S1032, the backscattering coefficients at each time phase are arranged in a time series to generate a backscattering coefficient vector at each pixel point in the target scanning area;

[0071] In step S1033, the backscattering coefficient vectors at all pixel points are combined into a two-dimensional matrix according to the spatial dimension to obtain a scattering characteristic matrix of the surface medium to the radar wave in the target scanning area.

[0072] In specific implementation, first, the backscattering coefficients at each time phase are obtained, and there is a backscattering coefficient corresponding to different pixel points in the target scanning area; then, the backscattering coefficients corresponding to the same pixel point position at each time phase are arranged in a time series to generate a backscattering coefficient vector at each pixel point in the target scanning area; finally, the backscattering coefficient vectors at all pixel points are combined into a two-dimensional matrix according to the spatial dimension, where the row vector represents the spatial position of the pixel point, the column vector represents the backscattering coefficients at different time phases, and the combined two-dimensional matrix is used as the scattering characteristic matrix of the surface medium to the radar wave in the target scanning area.

[0073] It should be noted that in this embodiment, the backscattering coefficient vector represents a combination of multiple backscattering coefficients arranged in time order; in this application, the scattering characteristic matrix represents a matrix used to measure the response change law of the surface medium to the radar wave in the time dimension. Specifically, the scattering characteristic matrix refers to a two-dimensional numerical matrix formed by organizing the backscattering coefficients of all pixel points in the target scanning area at multiple time phases according to the spatial position and time series. The scattering characteristic matrix can be used to analyze the dynamic evolution characteristics of the surface physical state and is an important basic data structure for subsequent surface deformation sensitivity identification and deformation area division.

[0074] In some embodiments, determining the surface deformation sensitivity of each pixel point in the target scanning area according to the scattering characteristic matrix in combination with the spatio-temporal change characteristics of the coherence coefficient can be implemented by the following steps:

[0075] Based on the backscattering coefficient vector corresponding to each pixel point in the scattering characteristic matrix, determine the scattering fluctuation intensity corresponding to each pixel point;

[0076] Obtain the coherence coefficient sequence of each pixel point within all time intervals;

[0077] According to the coherence coefficient sequence corresponding to each pixel point, determine the spatio-temporal change characteristic corresponding to each pixel point;

[0078] For each pixel, the scattering fluctuation intensity corresponding to the pixel is normalized and weighted with the spatio-temporal change characteristics, so as to obtain the surface deformation sensitivity at each pixel in the target scanning area.

[0079] It should be noted that in this application, the surface deformation sensitivity represents an index for measuring the degree of sensitivity of each pixel position in the target scanning area to surface deformation. The higher the surface deformation sensitivity, the greater the scattering intensity fluctuation and the less stable the coherence at the pixel at different observation time phases, indicating that the corresponding ground object or surface area may have active changes or deformation trends, and has a higher surface deformation risk or sensitivity. Therefore, by determining the surface deformation sensitivity, it is possible to effectively identify those surface areas that show obvious anomalies or instabilities in the radar wave reflection characteristics, and these areas are often highly correlated with geological bodies (such as landslide bodies) with actual active or potential instability risks.

[0080] When specifically implemented, first, based on the backscattering coefficient vector corresponding to each pixel in the scattering characteristic matrix, the scattering fluctuation intensity corresponding to each pixel is determined, that is: for each pixel, the variance of the backscattering coefficients of the backscattering coefficient vector corresponding to the pixel is used as the scattering fluctuation intensity corresponding to the pixel, so as to obtain the scattering fluctuation intensity corresponding to each pixel; secondly, a coherence coefficient sequence of each pixel within all time intervals is obtained, and the coherence coefficient sequence represents a combination of coherence coefficients including multiple time intervals; then, according to the coherence coefficient sequence corresponding to each pixel, the spatio-temporal change characteristics corresponding to each pixel are determined, that is: for each pixel, the minimum-maximum ratio of the coherence coefficients of the coherence coefficient sequence corresponding to the pixel is used as the spatio-temporal change characteristic corresponding to the pixel, so as to obtain the spatio-temporal change characteristics corresponding to each pixel; finally, for each pixel, the scattering fluctuation intensity corresponding to the pixel is normalized and weighted with the spatio-temporal change characteristics, so as to obtain the surface deformation sensitivity at each pixel in the target scanning area, that is: for each pixel, the scattering fluctuation intensity and the spatio-temporal change characteristics of the pixel are normalized to between 0 and 1 through min-max normalization, and then the normalized scattering fluctuation intensity and spatio-temporal change characteristics are weighted and fused by assigning weights, and the weighted fusion result is used as the surface deformation sensitivity at the pixel, so as to obtain the surface deformation sensitivity at each pixel in the target scanning area, where the weights of the scattering fluctuation intensity and the spatio-temporal change characteristics can be set according to actual needs. For example, in this embodiment, the weights of the scattering fluctuation intensity and the spatio-temporal change characteristics can be set to 0.6 and 0.4 respectively, and no limitation is made here.

[0081] It should be noted that, in this embodiment, the scattering fluctuation intensity represents the degree to which the backscattering coefficient of the pixel position in the target scanning area changes over time. The scattering fluctuation intensity is used to measure the strength of the change in the radar wave reflection characteristics of the pixel position under different observation phases. By determining the scattering fluctuation intensity, the surface deformation and landform changes in the target scanning area can be identified; in this embodiment, the spatiotemporal variation characteristics represent the degree to which the coherence coefficient of the pixel position in the target scanning area changes over time. The spatiotemporal variation characteristics reflect the fluctuation characteristics of the coherence of the pixel position in the entire time series, which are often used to judge the surface stability or identify potential deformation areas. The larger the spatiotemporal variation characteristics, the more drastic the coherence change, indicating that there are greater scatterer structure changes, landform disturbances or deformation activities on the surface. In landslide detection, the spatiotemporal resolution and accuracy of deformation identification can be effectively enhanced by determining the spatiotemporal variation characteristics.

[0082] In step S104, singular value decomposition is performed on the scattering characteristic matrix, and deformation sensitive areas and stability partitions in the target scanning area are divided according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point.

[0083] In some embodiments, performing singular value decomposition on the scattering characteristic matrix may be implemented by the following steps:

[0084] Decomposing the scattering characteristic matrix into a product form of a left singular vector matrix, a singular value diagonal matrix and a right singular vector matrix;

[0085] All singular values are arranged in descending order, and the left singular vector corresponding to the maximum singular value is extracted as the main singular vector, thereby completing the singular value decomposition of the scattering characteristic matrix.

[0086] In the specific implementation, first, the scattering characteristic matrix is decomposed into the product form of a left singular vector matrix, a singular value diagonal matrix and a right singular vector matrix by using the existing singular value decomposition method, which will not be repeated here; then, all the singular values are arranged in descending order, and the singular vector corresponding to the maximum singular value is extracted, thereby completing the singular value decomposition of the scattering characteristic matrix.

[0087] It should be noted that in this embodiment, the main singular vector is the vector corresponding to the maximum singular value in the left singular vector matrix. The left singular vector corresponding to the maximum singular value represents the most significant spatial pattern in the entire data set. The left singular vector corresponds to the spatial dimension and is used to reveal the response of the surface area (pixel) to the main change trend, while the right singular vector corresponds to the change trend in the time dimension and is more used to analyze the commonalities of time series. It cannot be directly used to divide the landslide area in space. Therefore, the left singular vector can be used to identify deformation sensitive areas.

[0088] In some embodiments, refer to Figure 3 As shown, the figure is an exemplary flowchart for determining the deformation-sensitive area and the stability partition according to some embodiments of the present application. In this embodiment, the deformation-sensitive area and the stability partition in the target scanning area can be divided according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point, which can be implemented by the following steps:

[0089] In step S1041, extract the principal singular vector from the singular value decomposition result of the scattering characteristic matrix;

[0090] In step S1042, determine the principal component projection value of each pixel point according to the principal singular vector and the backscattering coefficient vector of each pixel point in the scattering characteristic matrix;

[0091] In step S1043, determine the characteristic response map of the surface deformation in the target scanning area through the principal component projection value corresponding to each pixel point and the surface deformation sensitivity;

[0092] In step S1044, divide the deformation-sensitive area and the stability partition in the target scanning area based on the characteristic response map.

[0093] In the present application, the deformation-sensitive area represents a spatial area in the target scanning area with a relatively high surface deformation sensitivity. Pixel points in such areas usually show abnormal change trends in terms of backscattering characteristics, and at the same time, there are obvious fluctuations in their coherence coefficients in time series, indicating that these surface positions are more sensitive to external disturbances (such as rainfall, groundwater changes, or slope structure loosening), and are prone to subtle deformations or even precursors of landslides. By identifying the deformation-sensitive area, it is possible to delineate in advance the areas with potential geological disaster risks, providing data support for subsequent landslide risk assessment, thereby realizing the forward movement and precision of disaster prevention and control work; in the present application, the stability partition represents an area in the target scanning area with a relatively low surface deformation sensitivity, usually characterized by gentle changes in backscattering characteristics and stable coherence coefficients, indicating that the surface state of these areas is relatively stable and no significant deformations or disturbances have occurred. The role of dividing the stability partition is to provide a reference for landslide risk identification, exclude risk-free areas at the same time, reduce false alarms, and improve the detection accuracy.

[0094] In specific implementation, first, extract the principal singular vectors from the singular value decomposition result of the scattering characteristic matrix; second, determine the principal component projection value of each pixel in the scattering characteristic matrix based on the principal singular vectors and the backscattering coefficient vector of each pixel in the scattering characteristic matrix, that is: for each pixel, perform a convolution operation on the backscattering coefficient vector of the loudness point and the principal singular vectors, and use the convolution result as the principal component projection value of the pixel, so as to obtain the principal component projection value of each pixel; further, determine the characteristic response map of the ground deformation in the target scanning area through the principal component projection value corresponding to each pixel and the ground deformation sensitivity, that is: for each pixel, use the principal component projection value of the pixel as a weight value to weight the corresponding ground deformation sensitivity, and use the weighted result as the characteristic response value of the ground deformation at the pixel in the target scanning area, so as to obtain the characteristic response values of the ground deformation at each pixel in the target scanning area, and arrange all the characteristic response values according to the spatial dimension of the target scanning area to obtain the characteristic response map of the ground deformation in the target scanning area; finally, divide the deformation-sensitive area and the stability partition in the target scanning area based on the characteristic response map, that is: compare the characteristic response value corresponding to each pixel in the characteristic response map with the characteristic response threshold, classify the pixel positions with characteristic response values less than the characteristic response threshold as the stability area in the target scanning area, and classify the pixel positions with characteristic response values greater than or equal to the characteristic response threshold as the stability area in the target scanning area, where the characteristic response threshold can be set according to actual needs and will not be elaborated here.

[0095] It should be noted that in this embodiment, the principal component projection value represents the projection result of the pixel in the target scanning area on the principal singular vector, and the principal component projection value reflects the response intensity of the pixel on the main change trend of the radar scattering characteristics; in this embodiment, the characteristic response map represents the distribution characteristics of the potential deformation of the ground surface in the target scanning area on the spatio-temporal scale, and is an important intermediate result for identifying the deformation-sensitive area and the stability partition.

[0096] In step S105, determine the scattering entropy of the pixels in the deformation-sensitive area, identify the hidden landslide points in the target scanning area from the stability partition according to the scattering entropy, and then generate a landslide risk probability map by combining all the hidden landslide points and the deformation-sensitive area.

[0097] In some embodiments, the determination of the scattering entropy of the pixels in the deformation-sensitive area can be implemented by the following steps:

[0098] Extract the backscattering coefficient sequences of the pixels in the deformation-sensitive area;

[0099] Select a pixel as the selected pixel, and determine the scattering entropy of the selected pixel according to the backscattering coefficient sequence of the selected pixel;

[0100] Continue to determine the scattering entropy of the remaining pixel points in the deformation-sensitive region.

[0101] In specific implementation, first, extract the backscattering coefficient sequences of each pixel point in the deformation-sensitive region, where the backscattering coefficient sequences include backscattering coefficients at different time phases; second, determine the scattering entropy of the selected pixel point according to the backscattering coefficient sequence of the selected pixel point, that is: extract the probability distribution of different backscattering coefficients in the backscattering coefficient sequence, where the probability distribution is the frequency of occurrence of the backscattering coefficient in the backscattering coefficient sequence, and use each extracted probability distribution as an input variable to input into the entropy model, and the entropy model outputs the scattering entropy of the selected pixel point. The entropy model uses Shannon entropy, which will not be elaborated here; finally, continue to determine the scattering entropy of the remaining pixel points in the deformation-sensitive region by the determination method of "determining the scattering entropy of the selected pixel point according to the backscattering coefficient sequence of the selected pixel point".

[0102] It should be noted that in this application, the scattering entropy represents the fluctuation complexity of the pixel point position in the deformation-sensitive region in terms of the radar wave scattering intensity. The scattering entropy is used to determine whether there are hidden landslide points in the stability partition that seem stable but actually have extremely complex scattering characteristics. Although these points do not have an obvious deformation trend, their scattering characteristics are unstable due to underground structure disturbances, vegetation changes, or hydrological disturbances, and they have potential landslide risks. Landslide precursors are usually accompanied by weak surface disturbances and physical structure changes, which although not causing obvious deformation, are sufficient to trigger radar scattering fluctuations. Therefore, hidden landslide points in the stability partition can be further identified through the scattering entropy.

[0103] In some embodiments, identifying the hidden landslide points of the target scanning area from the stability partition according to the scattering entropy can be implemented by the following steps:

[0104] Determine multiple hidden landslide guiding points in the target scanning area according to the scattering entropy of each pixel point in the deformation-sensitive region;

[0105] Extract the pixel points adjacent to each hidden landslide guiding point from the stability partition as candidate hidden landslide points;

[0106] Conduct neighborhood scattering deviation analysis on each candidate hidden landslide point to obtain the structural instability degree of each candidate hidden landslide point;

[0107] Extract multiple hidden landslide points of the target scanning area from all candidate hidden landslide points according to each structural instability degree.

[0108] It should be noted that in this application, the latent landslide points refer to the position points in the target scanning area that do not show obvious deformation trends. These latent landslide points are usually located in the stability partition. Due to the small surface deformation amplitude or the lack of continuous change in the short term, they cannot be directly identified by the conventional deformation sensitivity. Although these latent landslide points have not developed into large-scale deformations, they are potential precursors of landslide formation. Therefore, by determining the latent landslide points, the detection effectiveness of the landslide area can be effectively improved.

[0109] When specifically implemented, first, multiple latent landslide guiding points in the target scanning area are determined according to the scattering entropy of each pixel point in the deformation sensitive area, that is: the scattering entropy of each pixel point is compared with the scattering entropy threshold, and the pixel points corresponding to the scattering entropy greater than the scattering entropy threshold are extracted as the latent landslide guiding points in the target scanning area, and then multiple latent landslide guiding points in the target scanning area are obtained. The scattering entropy threshold can be specifically set according to actual needs and is not limited here; second, the pixel points adjacent to each latent landslide guiding point are extracted from the stability partition as candidate latent landslide points; then, the neighborhood scattering deviation analysis is performed on each candidate latent landslide point to obtain the structural instability degree of each candidate latent landslide point, that is: for each candidate latent landslide point, the pixel points within the neighborhood range of the candidate latent landslide point are extracted as reference pixel points, and the mean value of the backscattering coefficient deviation between the candidate latent landslide point and all reference pixel points is calculated as the structural instability degree of the candidate latent landslide point, and then the structural instability degree of each candidate latent landslide point is obtained. Among them, the neighborhood range adopts the eight-neighborhood range, which will not be elaborated here; finally, the structural instability degree of each candidate latent landslide point is compared with the instability degree threshold, and the candidate latent landslide points with the structural instability degree greater than the structural instability degree threshold are used as the latent landslide points in the target scanning area, and then multiple latent landslide points in the target scanning area are obtained. The instability degree threshold can be set according to actual needs or set according to expert knowledge and is not limited here.

[0110] It should be noted that in this embodiment, the latent landslide guiding points refer to the auxiliary points used to guide the identification of latent landslide points; the candidate latent landslide points in this embodiment refer to the initially identified latent landslide points; the structural instability degree in this embodiment refers to an index that measures whether the radar scattering behavior of the candidate latent landslide points in the stability partition significantly deviates from the statistical characteristics of its neighboring pixels in its local neighborhood. The structural instability degree reflects the possible instability or disturbance signs of the candidate latent landslide points on the surface or subsurface structure. When monitoring landslides in the target scanning area, the greater the scattering entropy of a point, the stronger the surface instability. When the surface instability is stronger, it will bring about latent landslide risks in the surrounding areas. Therefore, the latent landslide risk areas can be identified through the scattering entropy.

[0111] In some embodiments, generating a landslide risk probability map by combining all potential landslide points with the deformation-sensitive area can be achieved through the following steps:

[0112] Combine the area composed of all potential landslide points with the deformation-sensitive area to form the potential landslide risk area of the target scanning area;

[0113] Extract the topographic gradient parameters at each position in the potential landslide risk area, including slope, aspect, and elevation change rate;

[0114] Normalize and fuse the topographic gradient parameters at each position in the potential landslide risk area to generate the landslide risk factor at each position;

[0115] Use the Kriging interpolation algorithm to extend the landslide risk factors at each position to the entire target scanning area to generate a landslide risk probability map.

[0116] In specific implementation, first, integrate all potential landslide points identified from the stability partition with the area with higher deformation sensitivity to form a potential landslide risk area, which reflects the spatio-temporal anomaly characteristics of the surface in terms of electromagnetic scattering and coherence; second, extract the topographic gradient parameters at each position from the potential landslide risk area, and the topographic gradient parameters include slope, aspect, and elevation change rate, and these topographic gradient parameters are obtained through the analysis of existing digital elevation model (DEM) data, which will not be elaborated here; then, for each position, normalize each parameter in the topographic gradient parameters and perform weighted fusion according to empirical weights to generate the landslide risk factor at each position; finally, use the existing Kriging interpolation algorithm to smoothly estimate the risk factors at known positions to the entire target scanning area to generate a landslide risk probability map according to spatial autocorrelation, which will not be elaborated here.

[0117] It should be noted that in this embodiment, the potential landslide risk area refers to the spatial area in the target scanning area that shows a relatively high possibility of landslide occurrence; in this embodiment, the topographic gradient parameter refers to a quantitative index that reflects the change trend of the surface morphology and the spatial undulation characteristics and is used for geological disaster monitoring; in this embodiment, the landslide risk factor refers to an index used to characterize the possibility of landslide disasters, and the landslide risk factor can reflect the surface stability of the area, external inducing conditions, and geological structure vulnerability, and is the core parameter for landslide risk modeling and probability map generation; in this application, the landslide risk probability map represents a spatial distribution map of the likelihood of landslide disasters at each position in the target scanning area. By determining this landslide risk probability map, visual warning of landslide risks can be realized, which can assist in scientifically formulating disaster prevention and control and land use strategies, and provide intuitive, quantitative, and dynamically updated decision-making basis for landslide monitoring, regional control, and emergency response.

[0118] In addition, on the other hand of the present application, in some embodiments, the present application provides a landslide area detection system based on radar remote sensing data. Refer to Figure 4 , this figure is a schematic structural diagram of a landslide area detection system based on radar remote sensing data shown in some embodiments of the present application. The landslide area detection system 200 based on radar remote sensing data includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described as follows:

[0119] The acquisition module 201. In the present application, the acquisition module 201 is mainly used to acquire radar remote sensing data of the target scanning area at multiple time phases;

[0120] The processing module 202. In the present application, the processing module 202 is mainly used to extract the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area in the corresponding radar remote sensing data at each time phase;

[0121] The processing module 202 is further used to construct a scattering characteristic matrix of the surface medium in the target scanning area to the radar wave based on the backscattering coefficient at each time phase, and determine the surface deformation sensitivity of each pixel point in the target scanning area according to the scattering characteristic matrix in combination with the spatio-temporal variation characteristics of the coherence coefficient;

[0122] In addition, the processing module 202 is further used to perform singular value decomposition on the scattering characteristic matrix, and divide the deformation sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point;

[0123] The execution module 203. In the present application, the execution module 203 is mainly used to determine the scattering entropy of the pixel points in the deformation sensitive area, identify the latent landslide points of the target scanning area from the stability partition according to the scattering entropy, and then generate a landslide risk probability map by combining all the latent landslide points with the deformation sensitive area.

[0124] In addition, the present application also provides a computer device. The computer device includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-mentioned landslide area detection method based on radar remote sensing data.

[0125] In some embodiments, refer to Figure 5 , this figure is a schematic structural diagram of a computer device for implementing the landslide area detection method based on radar remote sensing data shown in some embodiments of the present application. The landslide area detection method based on radar remote sensing data in the above embodiments can be passed through Figure 5implemented by the computer device shown. The computer device 300 includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0126] The processor 301 can be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more for controlling the execution of the landslide area detection method based on radar remote sensing data in this application.

[0127] The communication bus 302 can be used to transfer information between the above components.

[0128] The memory 303 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM), or other types of dynamic storage devices that can store information and instructions. It can also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this. The memory 303 can exist independently and be connected to the processor 301 through the communication bus 302. The memory 303 can also be integrated with the processor 301.

[0129] Among them, the memory 303 is used to store the program code for executing the solution of this application and is controlled by the processor 301 for execution. The processor 301 is used to execute the program code stored in the memory 303. The program code can include one or more software modules. The determination of the landslide area detection method based on radar remote sensing data in the above embodiments can be implemented by one or more software modules in the processor 301 and the program code in the memory 303.

[0130] A communication interface 304, using any device such as a transceiver, is used to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0131] In a specific implementation, as an example, a computer device may include multiple processors, and each of these processors may be a single-CPU processor or a multi-CPU processor. Here, the processor may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions).

[0132] The above computer device may be a general-purpose computer device or a special-purpose computer device. In a specific implementation, the computer device may be a desktop computer, a laptop computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiments of the present application do not limit the type of the computer device.

[0133] In addition, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned landslide area detection method based on radar remote sensing data is implemented.

[0134] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concept. Therefore, the appended claims are intended to be interpreted to include the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0135] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. A landslide area detection method based on radar remote sensing data, characterized in that, It includes the following steps: Obtain radar remote sensing data of the target scanning area at multiple time phases; Extract the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at each time phase from the corresponding radar remote sensing data; Construct a scattering characteristic matrix of the surface medium in the target scanning area for radar waves based on the backscattering coefficients at each time phase, and determine the surface deformation sensitivity of each pixel point in the target scanning area according to the spatio-temporal variation characteristics of the coherence coefficient combined with the scattering characteristic matrix; Perform singular value decomposition on the scattering characteristic matrix, and divide the deformation-sensitive area and stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel point; Determine the scattering entropy of the pixel points in the deformation-sensitive area, identify the hidden landslide points in the target scanning area from the stability partition according to the scattering entropy, and then generate a landslide risk probability map by combining all the hidden landslide points with the deformation-sensitive area; 2. The method according to claim 1, wherein Specifically, extracting the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at each time phase from the corresponding radar remote sensing data includes: Preprocess the radar remote sensing data at each time phase to generate a single-look complex radar image of the target scanning area at each time phase; Select one time phase as the selected time phase, and extract the backscattering coefficient when the detection radar detects the target scanning area at the selected time phase from the single-look complex radar image corresponding to the selected time phase according to the radiometric calibration function; Register the single-look complex radar image corresponding to the selected time phase with the single-look complex radar image corresponding to the adjacent time phase to generate a pair of single-look complex radar images corresponding to the selected time phase; Determine the coherence coefficient when the detection radar detects the target scanning area at the selected time phase according to the pair of single-look complex radar images; Continue to determine the backscattering coefficient and coherence coefficient when the detection radar detects the target scanning area at the remaining time phases.

3. The method according to claim 1, characterized in that Specifically, constructing a scattering characteristic matrix of the surface medium in the target scanning area for radar waves based on the backscattering coefficients at each time phase includes: Obtain the backscattering coefficients at each time phase, and there is a backscattering coefficient corresponding to different pixel points in the target scanning area; Arrange the backscattering coefficients at each time phase in a time series to generate a backscattering coefficient vector at each pixel point in the target scanning area; Combine the backscattering coefficient vectors at all pixel points into a two-dimensional matrix according to the spatial dimension to obtain a scattering characteristic matrix of the surface medium in the target scanning area for radar waves.

4. The method according to claim 1, wherein Specifically, determining the surface deformation sensitivity of each pixel point in the target scanning area according to the spatio-temporal variation characteristics of the coherence coefficient combined with the scattering characteristic matrix includes: Determine the scattering fluctuation intensity corresponding to each pixel point based on the backscattering coefficient vector corresponding to each pixel point in the scattering characteristic matrix; Obtain the coherence coefficient sequence of each pixel point within all time intervals; Determine the spatio-temporal variation characteristics corresponding to each pixel point according to the coherence coefficient sequence corresponding to each pixel point; For each pixel, the scattering fluctuation intensity corresponding to the pixel is normalized and weighted with the spatio-temporal variation characteristics, so as to obtain the surface deformation sensitivity at each pixel in the target scanning area.

5. The method according to claim 1, wherein Dividing the deformation-sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel specifically includes: Extracting the main singular vector from the singular value decomposition result of the scattering characteristic matrix; Determining the principal component projection value of each pixel according to the main singular vector and the backscattering coefficient vector of each pixel in the scattering characteristic matrix; Determining the characteristic response map of surface deformation in the target scanning area through the principal component projection value corresponding to each pixel and the surface deformation sensitivity; Dividing the deformation-sensitive area and the stability partition in the target scanning area based on the characteristic response map.

6. The method according to claim 1, wherein Determining the scattering entropy of the pixels in the deformation-sensitive area specifically includes: Extracting the backscattering coefficient sequences of the pixels in the deformation-sensitive area; Selecting a pixel as the selected pixel, and determining the scattering entropy of the selected pixel according to the backscattering coefficient sequence of the selected pixel; Continuing to determine the scattering entropy of the remaining pixels in the deformation-sensitive area.

7. The method according to claim 1, wherein Obtaining radar remote sensing data of the target scanning area at multiple time phases by a detection radar.

8. A landslide area detection system based on radar remote sensing data, which is used to execute the landslide area detection method based on radar remote sensing data according to any one of claims 1 to 7, characterized in that, Including: An acquisition module, configured to acquire radar remote sensing data of the target scanning area at multiple time phases; A processing module, configured to extract the backscattering coefficient and the coherence coefficient when the detection radar detects the target scanning area at each time phase from the corresponding radar remote sensing data; The processing module is further configured to construct a scattering characteristic matrix of the surface medium to the radar wave in the target scanning area based on the backscattering coefficients at each time phase, and determine the surface deformation sensitivity of each pixel in the target scanning area according to the scattering characteristic matrix and the spatio-temporal variation characteristics of the coherence coefficient; The processing module is further configured to perform singular value decomposition on the scattering characteristic matrix, and divide the deformation-sensitive area and the stability partition in the target scanning area according to the singular value decomposition result of the scattering characteristic matrix and the surface deformation sensitivity of each pixel; An execution module, configured to determine the scattering entropy of the pixels in the deformation-sensitive area, identify the hidden landslide points in the target scanning area from the stability partition according to the scattering entropy, and then generate a landslide risk probability map by combining all the hidden landslide points and the deformation-sensitive area.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory stores code, and the processor is configured to obtain the code and execute the landslide area detection method based on radar remote sensing data according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the landslide area detection method based on radar remote sensing data according to any one of claims 1 to 7.

Images