An unsupervised flood detection method and system taking into account spatiotemporal polarization information
By constructing the spatiotemporal polarized SAR data matrix and improving the distance measurement function of the K-means algorithm, the problem of low accuracy in unsupervised change detection is solved, and more accurate flood detection is achieved, reducing errors and improving detection accuracy.
Patent Information
- Application Number
- CN202410573056.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-10
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-05-10
AI Technical Summary
The existing unsupervised change detection methods are not very accurate in flood detection, and are easily affected by other weakly scattered land objects to produce error accumulation. Moreover, the K-means algorithm is not very applicable to SAR images, resulting in the inaccurate detection results of flood change.
Unsupervised flood detection method that takes into account space-time polarization information is adopted. By obtaining the polarization characteristics of time-series SAR images, a spatio-temporal polarization SAR data matrix is constructed, and a flood monitoring model based on K-means is input for pixel point clustering, using cross entropy Improved distance metric function to improve detection accuracy.
It improves the accuracy of flood change detection, reduces the cumulative error caused by separate clustering of different phases, can better fit the water feature distribution, accurately extract the edge information of flood areas, and enhances the detection accuracy.
Smart Images

Figure CN118537788B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of remote sensing image detection, and in particular relates to an unsupervised flood detection method and system taking into account temporal and spatial polarization information. Background Art
[0002] Existing Synthetic Aperture Radar (SAR) image change detection can be divided into supervised change detection methods and unsupervised change detection methods. Among them, supervised change detection methods are greatly affected by sample quality, and sample labeling requires a lot of time and manpower. In addition, SAR's unique imaging mechanism also brings inconvenience to image interpretation. Unsupervised change detection methods do not require the support of prior information. They learn input features by themselves for clustering. They have the characteristics of simple design and fast processing speed, and are more suitable for rapid response in flood detection. However, most of the current unsupervised change detection methods are used to process single-phase images separately, which are easily affected by other weakly scattering objects and cause error accumulation.
[0003] K-means is the most commonly used unsupervised method, but it still has some limitations when applied to flood detection in SAR images and is not very applicable to SAR images.
[0004] In summary, both supervised change detection and unsupervised change detection have certain limitations, which leads to the final obtained flood change detection results being inaccurate and unable to provide timely flood disaster warnings. Summary of the invention
[0005] In view of this, the present invention proposes an unsupervised flood detection method and system taking into account spatiotemporal polarization information, so as to solve the problem of low accuracy of flood detection based on SAR images.
[0006] In a first aspect, the present invention discloses an unsupervised flood detection method taking into account spatiotemporal polarization information, the method comprising:
[0007] Acquire time-series SAR images of the target area, extract polarization features of the time-series SAR images, and construct a spatiotemporal polarization SAR data matrix;
[0008] A K-means-based flood monitoring model was established, and the spatiotemporal polarimetric SAR data matrix was input into the K-means-based flood monitoring model to cluster the pixels of the time-series SAR images.
[0009] Binarization is performed on the pixel clustering results to obtain flood change detection results in the target area. Based on the above technical solution, preferably, the extraction of polarization features of time-series SAR images and the construction of spatiotemporal polarization SAR data matrix specifically include:
[0010] Extract the polarization characteristics of time-series SAR images and construct the polarization feature vector of each pixel;
[0011] According to the polarization eigenvector, the spatial polarization SAR data matrix of any single-phase SAR image is determined, and the spatial polarization SAR data matrix of the multi-phase SAR image is obtained;
[0012] The spatiotemporal polarimetric SAR data matrix is constructed based on the spatial polarimetric SAR data matrix of multi-temporal SAR images.
[0013] On the basis of the above technical solution, preferably, the expression of the spatiotemporal polarization SAR data matrix is:
[0014]
[0015] Among them, X mnft is the spatiotemporal polarimetric SAR data matrix, is the spatial polarization SAR data matrix of the t-th phase; x mnft represents the f-th polarization eigenvalue at the (m,n) pixel position of the SAR image in the t-th phase; m = 1, 2, ..., M, n = 1, 2, ..., N, f = 1, 2, ..., F, t = 1, 2, ..., T; M, N are the total number of rows and columns of the SAR image, F is the total number of polarization features, and T is the total number of phases;
[0016] The spatiotemporal polarization characteristic matrix of any pixel point (m, n) of the time-series SAR image is expressed as:
[0017]
[0018] in, Represents the polarization eigenvector of the pixel (m,n) in the tth phase, is the spatiotemporal polarization feature matrix of pixel point (m,n).
[0019] On the basis of the above technical solution, preferably, the step of inputting the spatiotemporal polarization SAR data matrix into the K-means-based flood monitoring model to perform pixel clustering of the time series SAR image specifically includes:
[0020] Preset the number of K clusters;
[0021] The spatiotemporal polarization feature matrices of K pixels are randomly selected from the spatiotemporal polarimetric SAR data matrix as the initial clustering centers.
[0022] The distance between the spatiotemporal polarization feature matrix of each pixel and the K cluster centers is calculated in sequence, and the pixel is divided into the cluster corresponding to the minimum distance;
[0023] Calculate the association matrix between each pixel and the cluster center, and update K cluster centers according to the association matrix;
[0024] Repeat the above process of calculating distance and updating cluster centers until the iteration end condition is reached.
[0025] On the basis of the above technical solution, preferably, the K-means-based flood monitoring model uses cross entropy to calculate the distance between the spatiotemporal polarization feature matrix and the cluster center, and the distance calculation formula is:
[0026]
[0027] Among them, dmnk represents the spatiotemporal polarization feature matrix of pixel point (m,n) and the kth cluster center The distance between is the pixel point corresponding to the kth cluster center (m k ,n k ) is the polarization eigenvector in the tth phase, k = 1, 2, …, K; tr[·] is the trace of the computation matrix;
[0028] and They are respectively the spatiotemporal polarization feature matrix of the pixel point (m, n) after normalized mapping and the kth cluster center.
[0029] On the basis of the above technical solution, preferably, the formula for the normalized mapping process is:
[0030] And meet
[0031] And meet
[0032] in, It represents the fth polarization eigenvalue at the pixel point (mk,nk) corresponding to the kth cluster center in the tth time phase.
[0033] On the basis of the above technical solution, preferably, the expression of the correlation matrix is:
[0034] Q mnk =[q mnk ] M×N×K ;
[0035] Among them, Q mnk is the association matrix between the pixel point (m,n) and the kth cluster center,
[0036] On the basis of the above technical solution, preferably, the expression for updating the K cluster centers according to the association matrix is:
[0037]
[0038] is the updated k-th cluster center, k = 1, 2, ..., K; is the spatiotemporal polarization feature matrix of pixel point (m,n).
[0039] In a second aspect, the present invention discloses an unsupervised flood detection system taking into account spatiotemporal polarization information, the system comprising:
[0040] Data processing module: used to obtain time-series SAR images of the target area, extract the polarization characteristics of the time-series SAR images, and construct the space-time polarization SAR data matrix;
[0041] Pixel clustering module: used to establish a K-means-based flood monitoring model, input the spatiotemporal polarimetric SAR data matrix into the K-means-based flood monitoring model to cluster the pixels of the time series SAR images;
[0042] Change detection module: used to perform binarization processing on the pixel clustering results to obtain flood change detection results in the target area. In a third aspect, the present invention discloses an electronic device, including: at least one processor, at least one memory, a communication interface and a bus;
[0043] Wherein, the processor, memory, and communication interface communicate with each other via the bus;
[0044] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement the method according to the first aspect of the present invention.
[0045] Compared with the prior art, the present invention has the following beneficial effects:
[0046] 1) The present invention integrates the time and polarization information of multi-phase SAR images, constructs a spatiotemporal polarization SAR data matrix, and inputs the K-means-based flood monitoring model to cluster the pixels of the time series SAR images, thereby realizing the detection of flood change areas, making up for the defect of insufficient information utilization of SAR images in traditional unsupervised change detection, and improving the detection accuracy.
[0047] 2) Before clustering, the present invention establishes a spatial polarization SAR data matrix by constructing a polarization feature vector for each pixel point. On this basis, a spatiotemporal polarization SAR data matrix is established in combination with multi-phase information. Based on the spatiotemporal polarization SAR data matrix, the spatiotemporal polarization feature matrix of each pixel point is conveniently extracted, which facilitates subsequent integrated data clustering processing and reduces the cumulative error caused by separate clustering of different phases.
[0048] 3) In view of the problem that the distance metric function of the traditional clustering algorithm is not well targeted to the changes of different land objects, the present invention expands the traditional K-means clustering method and improves its distance metric function through cross entropy to make it suitable for spatiotemporal polarization SAR data. It can better fit the distribution of water body characteristics, more accurately extract the edge information of the flood area, and improve the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0050] Figure 1 A flow chart of an unsupervised flood detection method taking into account spatiotemporal polarization information provided by the present invention;
[0051] Figure 2 Another schematic diagram of the process of an unsupervised flood detection method taking into account spatiotemporal polarization information provided by the present invention;
[0052] Figure 3 Examples of flood detection results for two target areas provided by the present invention. DETAILED DESCRIPTION
[0053] The following will be combined with 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 the embodiments. 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.
[0054] See also Figure 1 The present invention discloses an unsupervised flood detection method taking into account temporal and spatial polarization information, the method comprising:
[0055] S1. Acquire the time-series SAR images of the target area, extract the polarization characteristics of the time-series SAR images, and construct the spatiotemporal polarization SAR data matrix.
[0056] Step S1 specifically includes the following sub-steps:
[0057] S11. Time series SAR image preprocessing and polarization feature extraction.
[0058] Use remote sensing image processing software such as ENVI and PolSARpro to preprocess the SAR images of each phase respectively; select the same reference image to align the time series SAR images, and extract polarization features such as coherence matrix and covariance matrix.
[0059] S12. Extract the polarization features of the time-series SAR image and construct a polarization feature vector for each pixel.
[0060] For any single-phase SAR image, determine the polarization eigenvector of any pixel (m,n):
[0061]
[0062] in, represents the polarization feature vector of the pixel (m, n) in the tth phase, M is the total number of rows in an image, N is the total number of columns, F is the total number of polarization features extracted, and T is the total number of phases in the image sequence; x mnft It represents the f-th polarization eigenvalue at the (m,n) pixel position in the t-th time phase, m=1,2,…,M, n=1,2,…,N, f=1,2,…,F, t=1,2,…,T.
[0063] S13. Determine the spatial polarization SAR data matrix of any single-phase SAR image according to the polarization eigenvector, and obtain the spatial polarization SAR data matrix of the multi-phase SAR image.
[0064] The spatial polarimetric SAR data of any single-phase synthetic aperture radar SAR image can be expressed as a three-dimensional matrix:
[0065]
[0066] in, Represents the spatial polarimetric SAR data matrix of the synthetic aperture radar SAR image at the tth phase.
[0067] The spatial polarimetric SAR data matrix of each single-temporal SAR image is integrated to obtain the spatial polarimetric SAR data matrix of the multi-temporal SAR image.
[0068] S14. Construct a spatiotemporal polarimetric SAR data matrix based on the spatial polarimetric SAR data matrix of the multi-temporal SAR images.
[0069] The spatial polarimetric SAR data matrix of multi-phase SAR images is fused to obtain a four-dimensional spatial-temporal polarimetric SAR data matrix, which is expressed as:
[0070]
[0071] Among them, X mnft is the spatiotemporal polarimetric SAR data matrix, is the spatial polarization SAR data matrix of the t-th phase; x mnft represents the f-th polarization eigenvalue at the (m,n) pixel position of the SAR image in the t-th phase; m = 1, 2, ..., M, n = 1, 2, ..., N, f = 1, 2, ..., F, t = 1, 2, ..., T; M, N are the total number of rows and columns of the SAR image, F is the total number of polarization features, and T is the total number of phases;
[0072] Based on the spatiotemporal polarization SAR data matrix, the spatiotemporal polarization characteristic matrix of each pixel can be determined. The spatiotemporal polarization characteristic matrix of any pixel (m, n) of the time series SAR image can be expressed as:
[0073]
[0074] in, Represents the polarization eigenvector of the pixel (m,n) in the tth phase, is the spatiotemporal polarization feature matrix of pixel point (m,n).
[0075] Before performing unsupervised flood detection, the present invention establishes a spatial polarimetric SAR data matrix by constructing a polarization feature vector of each pixel point. On this basis, a spatiotemporal polarimetric SAR data matrix is established in combination with multi-phase information. The spatiotemporal polarimetric feature matrix of each pixel point is conveniently extracted based on the spatiotemporal polarimetric SAR data matrix, which facilitates subsequent integrated data clustering processing and reduces the cumulative error caused by separate clustering of different phases.
[0076] S2. Establish a flood monitoring model based on K-means, and input the spatiotemporal polarization SAR data matrix into the flood monitoring model based on K-means to cluster the pixels of the time series SAR images.
[0077] Figure 2 FIG. 1 is another flow chart of an unsupervised flood detection method taking into account spatiotemporal polarization information provided by the invention, see Figure 2 , step S2 specifically includes the following sub-steps:
[0078] S21. Establish a flood monitoring model based on K-means.
[0079] The K-means-based flood monitoring model of the present invention uses cross entropy to optimize the distance metric function.
[0080] S22, presetting the number of K clusters, and randomly selecting the spatiotemporal polarization feature matrices of K pixels from the spatiotemporal polarization SAR data matrix as initial clustering centers.
[0081] Assume (m k ,n k ) is the pixel point selected when the k-th cluster center is initialized, and the spatiotemporal polarization feature matrix of the k-th cluster center is for:
[0082]
[0083] Where k = 1, 2, ..., K; is the pixel point corresponding to the kth cluster center (m k ,n k ) is the polarization eigenvector in the tth phase.
[0084] S23, calculating the distance between the spatiotemporal polarization feature matrix of each pixel point and the K cluster centers in turn, sorting the distance values, and dividing the pixel point into the cluster corresponding to the minimum distance.
[0085] The present invention uses cross entropy to optimize the distance metric function of the K-means algorithm. When calculating the cross entropy, it is necessary to first convert the spatiotemporal polarization feature matrix of each pixel point and the cluster centers of each pixel point into The eigenvalues of are normalized and mapped to the range of [0,1].
[0086] The spatiotemporal polarization feature matrix of pixel point (m,n) The formula for normalizing the eigenvalues is:
[0087] And meet
[0088] in, is the spatiotemporal polarization feature matrix of the pixel point (m, n) after normalized mapping.
[0089] For K cluster centers The formula for normalizing the eigenvalues is:
[0090] And meet
[0091] in, is the kth cluster center after normalized mapping processing, Indicates that in the tth phase, the pixel point (m k ,n k )’s f-th polarization eigenvalue.
[0092] After normalized mapping, the cross entropy can be calculated. The calculation formula of the cross entropy between the spatiotemporal polarization feature matrix of the pixel point (m, n) and the K cluster centers is:
[0093]
[0094] Among them, d mnk Represents the spatiotemporal polarization feature matrix of pixel point (m,n) and the kth cluster center The distance between them is the cross entropy; tr[·] is the trace of the calculation matrix, that is, the sum of the elements on the diagonal of the matrix.
[0095] The minimum cross entropy from the spatiotemporal polarization feature matrix of each pixel to the K cluster centers is obtained, and the pixel is classified into the cluster corresponding to the minimum cross entropy. After calculating the cross entropy from all pixels to the K cluster centers and reclassifying all pixels, a round of clustering is completed.
[0096] The present invention expands the traditional K-means clustering method and improves its distance metric function through cross entropy to solve the problem that the distance metric function of the traditional clustering algorithm is not well targeted to the changes of different ground objects, making it suitable for spatiotemporal polarization SAR data, and can better achieve the fitting of water body feature distribution, more accurately extract the edge information of the flood area, and improve the detection accuracy.
[0097] S24, calculating the association matrix between each pixel point and the cluster center, and updating K cluster centers according to the association matrix.
[0098] The expression of the correlation matrix is:
[0099] Q mnk =[q mnk ] M×N×K ;
[0100] Among them, Q mnk is the association matrix between the pixel point (m,n) and the kth cluster center,
[0101] Update K cluster centers according to the association matrix, the expression is:
[0102]
[0103] in, is the updated kth cluster center, i.e., pixel point (m k ,n k)’s space-time polarization characteristic matrix, k = 1, 2, …, K; is the spatiotemporal polarization feature matrix of pixel point (m,n).
[0104] S25. If the preset maximum number of iterations is not reached, increase the current number of iterations by 1, return to step S23, continue the iterative operation, reclassify the pixels, and complete the pixel clustering until the preset maximum number of iterations is reached to obtain the cluster to which each pixel belongs.
[0105] S3. Binarize the pixel clustering results to obtain flood change detection results in the target area.
[0106] like Figure 3 What is shown is a comparison diagram of the ground truth map and flood detection results of two target areas selected in the experiment of the present invention, wherein (a1) and (a2) are the ground truth map and flood detection map of the first target area, respectively, and (b1) and (b2) are the ground truth map and flood detection map of the second target area, respectively. By comparing (a1) and (a2) or (b1) and (b2), it can be seen that the present invention can more accurately determine the newly added flood inundation range of the target area, can accurately extract the edge of the changed area, and can identify the small target changed area.
[0107] Table 1 Comparison of detection accuracy indicators of two target areas
[0108]
[0109] Among them, FA represents the false alarm rate, TE represents the true error rate, OA represents the overall accuracy, and KAPPA represents the consistency coefficient.
[0110] The present invention integrates the time and polarization information of multi-phase SAR images, constructs a spatiotemporal polarization SAR data matrix, and inputs a K-means-based flood monitoring model to cluster the pixels of the time series SAR images, thereby realizing the detection of flood change areas, making up for the defect of insufficient information utilization of SAR images in traditional unsupervised change detection, and reducing the cumulative error caused by clustering different phases separately through integrated data processing, thereby improving the detection accuracy; at the same time, the K-means clustering method is expanded to improve the distance metric function through cross entropy, making it suitable for spatiotemporal polarization SAR data, which can better achieve the fitting of water body feature distribution, accurately extract the edge information of the flood area, and improve the detection accuracy. Corresponding to the above method embodiment, the present invention also proposes an unsupervised flood detection system that takes into account spatiotemporal polarization information, and the system includes:
[0111] Data processing module: used to obtain time-series SAR images, extract polarization characteristics of time-series SAR images, and construct time-space polarization SAR data matrix;
[0112] Pixel clustering module: used to establish a K-means-based flood monitoring model, input the spatiotemporal polarimetric SAR data matrix into the K-means-based flood monitoring model to cluster the pixels of the time series SAR images;
[0113] Change detection module: used to binarize the pixel clustering results, obtain the newly added flood area, extract the change category of the newly added flood area, and obtain the flood change detection result.
[0114] The above system embodiments and method embodiments correspond one to one, and for a brief description of the system embodiments, please refer to the method embodiments.
[0115] The present invention also discloses an electronic device, comprising: at least one processor, at least one memory, a communication interface and a bus; wherein the processor, memory, and communication interface communicate with each other through the bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement the aforementioned method of the present invention.
[0116] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to implement all or part of the steps of the method described in the embodiment of the present invention. The storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk, and other media that can store program codes.
[0117] The system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be distributed to multiple network units. A person skilled in the art may select some or all of the modules according to actual needs to achieve the purpose of the solution of this embodiment without creative effort.
[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. An unsupervised flood detection method that takes into account both temporal and spatial polarization information, characterized in that: The method comprises: Acquire time-series SAR images of the target area, extract polarization features of the time-series SAR images, and construct a spatiotemporal polarization SAR data matrix; A K-means-based flood monitoring model was established, and the spatiotemporal polarimetric SAR data matrix was input into the K-means-based flood monitoring model to cluster the pixels of the time-series SAR images. Binarize the pixel clustering results to obtain the flood change detection results of the target area; The expression of the space-time polarization SAR data matrix is: Among them, X mnft is the spatiotemporal polarimetric SAR data matrix, is the spatial polarization SAR data matrix of the t-th phase; x mnft represents the f-th polarization eigenvalue at the (m,n) pixel position of the SAR image in the t-th phase; m = 1, 2, ..., M, n = 1, 2, ..., N, f = 1, 2, ..., F, t = 1, 2, ..., T; M, N are the total number of rows and columns of the SAR image, F is the total number of polarization features, and T is the total number of phases; The spatiotemporal polarization characteristic matrix of any pixel point (m, n) of the time-series SAR image is expressed as: in, represents the polarization eigenvector of the pixel (m,n) in the tth time phase, χ mnft is the spatiotemporal polarization feature matrix of pixel point (m,n); The step of inputting the spatiotemporal polarimetric SAR data matrix into the K-means-based flood monitoring model to perform pixel clustering of the time series SAR image specifically includes: Preset the number of K clusters; Randomly select the spatiotemporal polarization feature matrix of K pixels from the spatiotemporal polarization SAR data matrix as the initial clustering center; Calculate the distance between the spatiotemporal polarization feature matrix of each pixel and the K cluster centers in turn, and divide the pixel into the cluster corresponding to the minimum distance; Calculate the association matrix between each pixel and the cluster center, and update K cluster centers according to the association matrix; Repeat the above process of calculating distance and updating cluster center until the iteration end condition is reached; The K-means-based flood monitoring model uses cross entropy to calculate the distance between the spatiotemporal polarization feature matrix and the cluster center. The distance calculation formula is: Among them, d mnk Represents the spatiotemporal polarization feature matrix of pixel point (m,n) and the kth cluster center The distance between is the pixel point corresponding to the kth cluster center (m k ,n k ) is the polarization eigenvector in the tth phase, k = 1, 2, …, K; tr[·] is the trace of the computation matrix; and They are respectively the spatiotemporal polarization feature matrix of the pixel point (m, n) after normalized mapping and the kth cluster center.
2. The unsupervised flood detection method taking into account spatiotemporal polarization information according to claim 1, characterized in that: The step of extracting polarization features of time-series SAR images and constructing a spatiotemporal polarization SAR data matrix specifically includes: Extract the polarization characteristics of time-series SAR images and construct the polarization feature vector of each pixel; According to the polarization eigenvector, the spatial polarization SAR data matrix of any single-phase SAR image is determined, and the spatial polarization SAR data matrix of multi-phase SAR image is obtained; The spatiotemporal polarimetric SAR data matrix is constructed based on the spatial polarimetric SAR data matrix of multi-temporal SAR images.
3. The unsupervised flood detection method taking into account spatiotemporal polarization information according to claim 1, characterized in that: The formula for the normalized mapping process is: And meet And meet in, Indicates the pixel point (m k ,n k ) is the f-th polarization eigenvalue on .
4. The unsupervised flood detection method taking into account spatiotemporal polarization information according to claim 1, characterized in that: The expression of the association matrix is: Q mnk =[q mnk ]M×N×K; Among them, Q mnk is the association matrix between the pixel point (m,n) and the kth cluster center, 5. The unsupervised flood detection method taking into account spatiotemporal polarization information according to claim 4, characterized in that: The expression for updating K cluster centers according to the association matrix is: is the updated k-th cluster center, k = 1, 2, ..., K; χ mnft is the spatiotemporal polarization feature matrix of pixel point (m,n).
6. An unsupervised flood detection system taking into account spatiotemporal polarization information using the method of any one of claims 1 to 5, characterized in that: The system comprises: Data processing module: used to obtain time-series SAR images of the target area, extract the polarization characteristics of the time-series SAR images, and construct the space-time polarization SAR data matrix; Pixel clustering module: used to establish a K-means-based flood monitoring model, input the spatiotemporal polarization SAR data matrix into the K-means-based flood monitoring model to cluster the pixels of the time series SAR images; Change detection module: used to binarize the pixel clustering results to obtain flood change detection results in the target area.
7. An electronic device comprising: At least one processor, at least one memory, a communication interface and a bus; wherein the processor, memory and communication interface communicate with each other via the bus; the memory stores program instructions executable by the processor, and the processor calls the program instructions to implement the method according to any one of claims 1 to 5.
Citation Information
Patent Citations
Flood water body extraction method and system based on time sequence spaceborne SAR (Synthetic Aperture Radar) image
CN117274809A
Multi-polarization SAR farmland flood detection method and system considering rainfall influence
CN117274821A
Flood disaster monitoring method for multi-mode time sequence PolSAR image
CN117333468A
Image analysis device, image analysis method, and computer-readable recording medium
US20200096628A1
