A flood disaster monitoring method based on threshold interval and neighborhood analysis

Abstract

The present invention relates to the field of flood disaster monitoring technology, and in particular to a flood disaster monitoring method based on threshold interval and neighborhood analysis. The method comprises the following steps: obtaining polarization radar images of the flood-stricken area and preprocessing them to obtain radar backscatter coefficient band data; secondly, plotting a radar backscatter coefficient histogram and obtaining the peak backscatter coefficient values ​​of land and water bodies through histogram bimodal identification; then, calculating the threshold segmentation interval based on the peak points, segmenting the image into water bodies, land, and mixed water and land pixels, and obtaining preliminary water body extraction results; finally, eliminating pseudo-water bodies in the water body, and performing neighborhood water and land attribute discrimination on the mixed water and land pixels, ultimately obtaining flood submerged water body extraction results. The present invention improves the monitoring accuracy of submerged water bodies.

Description

Technical Field

[0001] The present invention relates to the technical field of flood disaster monitoring, and in particular to a flood disaster monitoring method based on threshold interval and neighborhood analysis. Background Art

[0002] Remote sensing technology has great potential in extracting flood disaster inundation information from large-scale and complex surfaces. However, flood disasters are often accompanied by extreme weather, which causes optical remote sensing using visible light and near-infrared bands to be greatly affected by clouds and rain, making it impossible to obtain effective surface information.

[0003] Remote sensing technology is used as a routine means of monitoring floods and waterlogging. Threshold segmentation methods use manual experience, the OTSU (maximum inter-class variance) method, or the KI (Kittler-Illingworth) method to select one or more thresholds to delineate between water and non-water bodies. These methods typically only consider the magnitude of the backscatter coefficient in radar imagery. For complex surface areas, such as mountain shadows and densely populated urban areas, they fail to account for the noise impact of low-backscatter non-water features such as shadows, roads, and airports on water extraction. This fails to fully utilize the spatial distribution information contained in radar remote sensing imagery, resulting in poor application effectiveness in complex scenarios. Currently, among high-precision flood monitoring methods, machine learning relies on the construction of large sample sets, resulting in high computational complexity and poor real-time performance, making it difficult to meet the rapid response requirements of flood monitoring. Region growing algorithms are sensitive to noise and computationally intensive, making them ineffective for complex image processing. Methods such as change detection and time series analysis require large amounts of image data and are significantly affected by coherent speckle noise. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a flood disaster monitoring method based on threshold interval and neighborhood analysis to solve at least one of the above technical problems.

[0005] To achieve the above objectives, a flood disaster monitoring method based on threshold interval and neighborhood analysis includes the following steps:

[0006] Step S1: obtaining polarimetric radar images of the flood-stricken area; preprocessing the polarimetric radar images of the flood-stricken area to obtain radar backscatter coefficient band data;

[0007] Step S2: Draw a histogram of the radar backscatter coefficient band data to obtain a radar backscatter coefficient histogram; perform histogram bimodal identification on the polarization radar image of the flood-stricken area based on the radar backscatter coefficient histogram to obtain the backscatter coefficient value of the peak point on the land and the backscatter coefficient value of the peak point on the water body;

[0008] Step S3: Calculating a threshold segmentation interval for the polarimetric radar image of the flood-stricken area based on the backscatter coefficient values ​​of the peak land points and the backscatter coefficient values ​​of the peak water points to obtain a radar image threshold segmentation interval; segmenting the polarimetric radar image of the flood-stricken area into water, land, and mixed water and land pixels based on the radar image threshold segmentation interval to obtain a preliminary water extraction result, wherein the preliminary water extraction result includes water, land, and mixed water and land pixels;

[0009] Step S4: remove pseudo water bodies from the water bodies in the preliminary water body extraction results, and perform neighborhood water and land attribute discrimination on the water-land mixed pixels in the preliminary water body extraction results, thereby obtaining the flood-inundated water body extraction results.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: collecting polarimetric radar images of the target flood-stricken area to obtain polarimetric radar images of the flood-stricken area;

[0012] Step S12: performing multi-view processing and filtering on the polarimetric radar image of the flood-stricken area to obtain a multi-view filtered image of the flood-stricken area;

[0013] Step S13: geocoding the multi-view filtered image of the flood-stricken area to obtain a geocoded image of the flood-stricken area;

[0014] Step S14: performing radiometric calibration on the geocoded image of the flood-stricken area to obtain a polarimetric radar image of the flood-stricken area;

[0015] Step S15: performing coherent polarization decomposition correction on the polarimetric radar image of the flood-stricken area to obtain radar backscatter coefficient band data.

[0016] Preferably, step S2 includes the following steps:

[0017] Step S21: extracting pixel values ​​from radar backscatter coefficient band data to obtain a one-dimensional array of backscatter coefficients;

[0018] Step S22: plotting a histogram of the radar backscatter coefficient band data based on the one-dimensional array of backscatter coefficients to obtain a radar backscatter coefficient histogram;

[0019] Step S23: performing histogram double-peak identification on the radar backscatter coefficient histogram to obtain the backscatter coefficient value of the land peak point and the backscatter coefficient value of the water peak point.

[0020] Preferably, step S23 includes the following steps:

[0021] Perform smooth curve extraction on the radar backscatter coefficient histogram to obtain a smooth curve of the backscatter coefficient histogram;

[0022] Based on the backscatter coefficient histogram smoothing curve, the histogram bimodal identification of the multi-view filtered image of the flood-stricken area is performed to obtain the backscatter coefficient values ​​of the peak points of land and the peak points of water. The specific processing process of the histogram bimodal identification of the multi-view filtered image of the flood-stricken area based on the backscatter coefficient histogram smoothing curve is as follows:

[0023] ;

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] in, is the backscatter coefficient histogram smoothing curve, w is the filter window size, p is the polynomial order of the Savitzky-Golay filter, is the backscattering coefficient value with the largest function value in the extreme point, is the backscattering coefficient value with the second largest function value in the extreme point, is the backscattering coefficient value of the peak point on land, is the backscattering coefficient value at the peak point of the water body.

[0030] Preferably, step S3 includes the following steps:

[0031] Step S31: The backscatter coefficient value of the land peak point As the mean of the Gaussian distribution of land, the backscattering coefficient value of the peak point of water body As the Gaussian distribution mean of the water body, the pixel values ​​in the radar backscatter coefficient band data that are less than the Gaussian distribution mean of the water body are recorded as the first candidate pixel set, and the pixel values ​​in the radar backscatter coefficient band data that are greater than the Gaussian distribution mean of the land are recorded as the second candidate pixel set;

[0032] Step S32: Calculate the standard deviation of the first candidate pixel set to obtain the standard deviation of the water body Gaussian distribution , and calculate the standard deviation of the second candidate pixel set to obtain the standard deviation of the land Gaussian distribution ;

[0033] Step S33: Use 4 times the standard deviation as the initial segmentation interval of the polarimetric radar image of the flood-stricken area. The lower limit of the initial segmentation interval is the Gaussian distribution mean of the water body + 4 times the Gaussian distribution standard deviation of the water body. , the upper limit of the initial segmentation interval is the mean of the Gaussian distribution of the land 4 times the standard deviation of the land Gaussian distribution , that is, the initial segmentation interval is , determine the initial segmentation interval as Is the lower limit of the interval greater than the upper limit of the interval? If so, the initial segmentation interval is reduced by 0.5 times the standard deviation and the segmentation interval is recalculated until the lower limit of the segmentation interval is less than the upper limit of the segmentation interval, thereby obtaining the radar image threshold segmentation interval as ,k={4,3.5,3,2.5,2,1.5,1,0};

[0034] Step S34: Segment the polarization radar image of the flood-stricken area into water, land, and water-land mixed pixels according to the radar image threshold segmentation interval to obtain a preliminary water extraction result, wherein the preliminary water extraction result includes water, land, and water-land mixed pixels.

[0035] Preferably, step S34 includes the following steps:

[0036] Assume that the radar backscatter coefficient band data is I, and the preliminary extraction result of water body is The specific processing process of water body, land and water-land mixed pixel segmentation is as follows:

[0037] .

[0038] Preferably, step S4 includes the following steps:

[0039] Step S41: extracting DEM values ​​and generating histograms for the water bodies in the preliminary water body extraction results to obtain a DEM histogram of water body pixels;

[0040] The DEM histogram of water body pixels is fitted with lognormal distribution using the preset lognormal distribution function, and the DEM value of the cumulative distribution probability curve of the preset lognormal distribution function at 90% is calculated to obtain the water body DEM threshold value. The water body DEM threshold value is recorded as ; Among them, the probability density function and cumulative distribution function formulas of the preset lognormal distribution function are as follows:

[0041] ;

[0042] ;

[0043] in, is the mean value of the water pixel DEM after logarithmic transformation, is the standard deviation of the water pixel DEM after logarithmic transformation, is the cumulative distribution function of the standard normal distribution;

[0044] Step S42: The water body is higher than The water pixels are recorded as high altitude water bodies. The pseudo water bodies in the high-altitude areas in the preliminary water body extraction results are eliminated, and the neighborhood water and land attributes of the water-land mixed pixels in the preliminary water body extraction results are discriminated to obtain the flood-inundated water body extraction results.

[0045] Preferably, the water body is above The water pixels are recorded as high altitude water bodies. The specific process of removing pseudo water bodies in high-altitude areas from the preliminary water body extraction results is as follows:

[0046] With the water body pixel in the high altitude area as the center, a neighborhood window is set with a size of N. The mountain shadow index of the water body pixel in the high altitude area is defined as MSI. When the MSI is greater than 5, the water body pixel in the high altitude area is judged to be a mountain shadow pixel. When the MSI is less than or equal to 5, the water body pixel in the high altitude area is judged to be a real water body. The MSI calculation formula is as follows:

[0047] ;

[0048] ;

[0049] ;

[0050] in, is the number of mixed pixels in the neighborhood window, is the number of water pixels in the neighborhood window, It is the maximum value of the normalized DEM in the neighborhood window.

[0051] Preferably, the specific process of performing neighborhood water and land attribute discrimination on the water-land mixed pixels in the preliminary water body extraction result is as follows:

[0052] Assume that the neighborhood window size of the water-land mixed pixel is S. Based on the water-land mixed pixel, the number of water pixels and the number of land pixels in the neighborhood window S are calculated. According to the attribute discrimination rule of the water-land mixed pixel, the mixed pixel attributes in the water-land mixed pixel are judged. The attribute discrimination rule of the water-land mixed pixel is as follows:

[0053] L(i,j)= −1−W(i,j);

[0054] ;

[0055] Among them, S is the neighborhood window size of the water-land mixed pixel, and the total number of pixels in the neighborhood is −1, L(i,j) is the number of land pixels, W(i,j) is the number of water pixels, i is the row number of the mixed land and water pixel, and j is the column number of the mixed land and water pixel.

[0056] This method automatically sets upper and lower threshold limits based on the Gaussian distribution of land and water mixtures in radar remote sensing images, preliminarily classifying radar images into three categories: water, land, and mixed land and water pixels. By analyzing the distribution and elevation characteristics of water and mixed land pixels in different spatial regions, a mountain shadow index (MSI) is constructed to effectively remove pseudo-water bodies caused by mountain shadows. Neighborhood analysis is used to further identify the land and water properties of mixed land and water pixels, thereby overcoming classification errors caused by low-scattering objects and significantly improving monitoring accuracy. This method features minimal human intervention, simple threshold setting, and strong adaptability. It offers the advantages of rapid response and high-precision monitoring in flood disaster emergency monitoring. It also exhibits strong adaptability and robustness in complex terrain and multi-source radar remote sensing applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Other features, objects and advantages of the present invention will become more apparent from reading the detailed description made with reference to the following drawings:

[0058] Figure 1 A schematic flow chart of the steps of a flood disaster monitoring method based on threshold interval and neighborhood analysis according to an embodiment is shown.

[0059] Figure 2 A schematic diagram of backscatter coefficient images of the GF-3 HH and HV polarization modes according to an embodiment is shown.

[0060] Figure 3 A schematic diagram of backscatter coefficients at peak points of the histogram of water and land distribution in HH and HV band images according to an embodiment is shown.

[0061] Figure 4 A schematic diagram of the HH band threshold segmentation interval of the GF-3 signal is shown in an embodiment.

[0062] Figure 5 A schematic diagram of the threshold segmentation interval of the GF-3 HV band according to an embodiment is shown.

[0063] Figure 6 A schematic diagram of preliminary water extraction results according to an embodiment is shown.

[0064] Figure 7 A schematic diagram of a DEM data distribution histogram of water body pixels according to an embodiment is shown.

[0065] Figure 8 A schematic diagram showing the division results of a small number of high-altitude area water bodies according to an embodiment is shown.

[0066] Figure 9 A schematic diagram of the effect of removing mountain shadows from a preliminary extraction result according to an embodiment is shown.

[0067] Figure 10 A schematic diagram of flood-inundated water body extraction results according to an embodiment is shown. DETAILED DESCRIPTION

[0068] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0069] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0070] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0071] To achieve this, please refer to Figures 1 to 3 The present invention further describes the technology of the present invention by taking the domestic dual-polarization GF-3 radar image as an example. The specific implementation method includes the following steps:

[0072] Preferably, step S1 includes the following steps:

[0073] Step S11: collecting polarimetric radar images of the target flood-stricken area to obtain polarimetric radar images of the flood-stricken area;

[0074] Step S12: performing multi-view processing and filtering on the polarimetric radar image of the flood-stricken area to obtain a multi-view filtered image of the flood-stricken area;

[0075] Step S13: geocoding the multi-view filtered image of the flood-stricken area to obtain a geocoded image of the flood-stricken area;

[0076] Step S14: performing radiometric calibration on the geocoded image of the flood-stricken area to obtain a polarimetric radar image of the flood-stricken area;

[0077] Step S15: performing coherent polarization decomposition correction on the polarimetric radar image of the flood-stricken area to obtain radar backscatter coefficient band data.

[0078] In this embodiment, HH (Horizontal-Horizontal Polarization) and HV (Horizontal-Vertical Polarization) dual-polarization GF-3 radar remote sensing images of the target flood-stricken area are obtained, and the obtained radar remote sensing images are pre-processed by multi-viewing, geocoding, radiometric calibration, filtering and cropping to obtain backscatter coefficient images in HH and HV polarization modes. The image size is 3600 2700, see Figure 2 .

[0079] Preferably, step S2 includes the following steps:

[0080] Step S21: extracting pixel values ​​from radar backscatter coefficient band data to obtain a one-dimensional array of backscatter coefficients;

[0081] Step S22: plotting a histogram of the radar backscatter coefficient band data based on the one-dimensional array of backscatter coefficients to obtain a radar backscatter coefficient histogram;

[0082] Step S23: performing histogram double-peak identification on the radar backscatter coefficient histogram to obtain the backscatter coefficient value of the land peak point and the backscatter coefficient value of the water peak point.

[0083] In this embodiment, the grayscale histograms of the backscatter coefficient images in the HH and HV polarization modes are statistically analyzed, and the Savitzky-Golay filter is used to obtain the histogram smoothing curve to obtain the backscatter coefficient values ​​of the peak points of land and water. Figure 3 .

[0084] Preferably, step S23 includes the following steps:

[0085] Perform smooth curve extraction on the radar backscatter coefficient histogram to obtain a smooth curve of the backscatter coefficient histogram;

[0086] Based on the backscatter coefficient histogram smoothing curve, the histogram bimodal identification of the multi-view filtered image of the flood-stricken area is performed to obtain the backscatter coefficient values ​​of the peak points of land and the peak points of water. The specific processing process of the histogram bimodal identification of the multi-view filtered image of the flood-stricken area based on the backscatter coefficient histogram smoothing curve is as follows:

[0087] ;

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] ;

[0093] in, is the backscatter coefficient histogram smoothing curve, w is the filter window size, p is the polynomial order of the Savitzky-Golay filter, is the backscattering coefficient value with the largest function value in the extreme point, is the backscattering coefficient value with the second largest function value in the extreme point, is the backscattering coefficient value of the peak point on land, is the backscattering coefficient value at the peak point of the water body.

[0094] In this embodiment, the backscatter coefficients of the peak points of the water and land distribution histograms are used. 、 , as the mean of the Gaussian distribution of water and land, respectively, and calculate the number of images in each band that is greater than The standard deviation of the pixel value is used as the standard deviation of the land Gaussian distribution , calculate the number of images in each band that is smaller than The standard deviation of the pixel value is used as the standard deviation of the water body Gaussian distribution The lower limit of the interval is the mean of the backscatter coefficient of the water body + 4 times the standard deviation, and the upper limit of the interval is the mean of the backscatter coefficient of the land 4 times the standard deviation, and finally the threshold segmentation interval in this example is obtained as follows Figure 3 , the schematic diagram of the segmentation interval is shown in Figure 4 、 Figure 5 .

[0095] Preferably, step S3 includes the following steps:

[0096] Step S31: The backscatter coefficient value of the land peak point As the mean of the Gaussian distribution of land, the backscattering coefficient value of the peak point of water body As the Gaussian distribution mean of the water body, the pixel values ​​in the radar backscatter coefficient band data that are less than the Gaussian distribution mean of the water body are recorded as the first candidate pixel set, and the pixel values ​​in the radar backscatter coefficient band data that are greater than the Gaussian distribution mean of the land are recorded as the second candidate pixel set;

[0097] Step S32: Calculate the standard deviation of the first candidate pixel set to obtain the standard deviation of the water body Gaussian distribution , and calculate the standard deviation of the second candidate pixel set to obtain the standard deviation of the land Gaussian distribution ;

[0098] Step S33: Use 4 times the standard deviation as the initial segmentation interval of the polarimetric radar image of the flood-stricken area. The lower limit of the initial segmentation interval is the Gaussian distribution mean of the water body + 4 times the Gaussian distribution standard deviation of the water body. , the upper limit of the initial segmentation interval is the mean of the Gaussian distribution of the land 4 times the standard deviation of the land Gaussian distribution , that is, the initial segmentation interval is , determine the initial segmentation interval as Is the lower limit of the interval greater than the upper limit of the interval? If so, the initial segmentation interval is reduced by 0.5 times the standard deviation and the segmentation interval is recalculated until the lower limit of the segmentation interval is less than the upper limit of the segmentation interval, thereby obtaining the radar image threshold segmentation interval as ,k={4,3.5,3,2.5,2,1.5,1,0};

[0099] Step S34: Segment the polarization radar image of the flood-stricken area into water, land, and water-land mixed pixels according to the radar image threshold segmentation interval to obtain a preliminary water extraction result, wherein the preliminary water extraction result includes water, land, and water-land mixed pixels.

[0100] In this embodiment, based on the radar image threshold segmentation interval range, the pixels in the backscatter coefficient image that are less than the threshold interval are classified as water bodies and assigned a value of 1, the pixels that are greater than the threshold interval are classified as land and assigned a value of 0, and the pixels within the threshold interval are classified as mixed water and land pixels and assigned a value of 2. The extraction results corresponding to the two polarization bands (HH polarization band and HV polarization band) are obtained, and the two extraction results are merged to obtain the preliminary water body extraction result, see Figure 6 .

[0101] Preferably, step S34 includes the following steps:

[0102] Assume that the radar backscatter coefficient band data is I, and the preliminary extraction result of water body is The specific processing process of water body, land and water-land mixed pixel segmentation is as follows:

[0103] .

[0104] Preferably, step S4 includes the following steps:

[0105] Step S41: extracting DEM values ​​and generating histograms for the water bodies in the preliminary water body extraction results to obtain a DEM histogram of water body pixels;

[0106] The DEM histogram of water body pixels is fitted with lognormal distribution using the preset lognormal distribution function, and the DEM value of the cumulative distribution probability curve of the preset lognormal distribution function at 90% is calculated to obtain the water body DEM threshold value. The water body DEM threshold value is recorded as ; Among them, the probability density function and cumulative distribution function formulas of the preset lognormal distribution function are as follows:

[0107] ;

[0108] ;

[0109] in, is the mean value of the water pixel DEM after logarithmic transformation, is the standard deviation of the water pixel DEM after logarithmic transformation, is the cumulative distribution function of the standard normal distribution;

[0110] Step S42: The water body is higher than The water pixels are recorded as high altitude water bodies. The pseudo water bodies in the high-altitude areas in the preliminary water body extraction results are eliminated, and the neighborhood water and land attributes of the water-land mixed pixels in the preliminary water body extraction results are discriminated to obtain the flood-inundated water body extraction results.

[0111] In this embodiment, the above image ( Figure 6 ) to remove the pseudo water body error caused by mountain shadows. The log-normal distribution function is used to simulate the DEM data distribution histogram of water body pixels (see Figure 7 ), whose cumulative distribution probability curve is at 90% of the DEM value =30.97m. Water bodies above this value are classified as high-altitude water bodies. The classification results of high-altitude water bodies are shown in Figure 8 .

[0112] Preferably, the water body is above The water pixels are recorded as high altitude water bodies. The specific process of removing pseudo water bodies in high-altitude areas from the preliminary water body extraction results is as follows:

[0113] With the water body pixel in the high altitude area as the center, a neighborhood window is set with a size of N. The mountain shadow index of the water body pixel in the high altitude area is defined as MSI. When the MSI is greater than 5, the water body pixel in the high altitude area is judged to be a mountain shadow pixel. When the MSI is less than or equal to 5, the water body pixel in the high altitude area is judged to be a real water body. The MSI calculation formula is as follows:

[0114] ;

[0115] ;

[0116] ;

[0117] in, is the number of mixed pixels in the neighborhood window, is the number of water pixels in the neighborhood window, It is the maximum value of the normalized DEM in the neighborhood window.

[0118] In this embodiment, the neighborhood window size is set to 41 with the water pixel in the high altitude area as the center. The larger the window setting is, the better it is to process larger mountain shadow patches. The mountain shadow index MSI is calculated based on the DEM, the mixed water and land pixels and the water pixels. The water pixels with an MSI value greater than 5 in the preliminary water extraction results are judged as mountain shadows and assigned a value of 0. When the shadow patches are large, the shadows can be removed by iteratively calculating the MSI multiple times. The mountain shadow removal results are shown in Figure 9 .

[0119] Preferably, the specific process of performing neighborhood water and land attribute discrimination on the water-land mixed pixels in the preliminary water body extraction result is as follows:

[0120] Assume that the neighborhood window size of the water-land mixed pixel is S. Based on the water-land mixed pixel, the number of water pixels and the number of land pixels in the neighborhood window S are calculated. According to the attribute discrimination rule of the water-land mixed pixel, the mixed pixel attributes in the water-land mixed pixel are judged. The attribute discrimination rule of the water-land mixed pixel is as follows:

[0121] L(i,j)= −1−W(i,j);

[0122] ;

[0123] Among them, S is the neighborhood window size of the water-land mixed pixel, and the total number of pixels in the neighborhood is −1, L(i,j) is the number of land pixels, W(i,j) is the number of water pixels, i is the row number of the mixed land and water pixel, and j is the column number of the mixed land and water pixel.

[0124] In this embodiment, the neighborhood window size S of the water-land mixed pixel is set to 3, the number of water pixels and the number of land pixels in the neighborhood window are calculated, and the attributes of the mixed pixels are judged according to the attribute discrimination rules of the water-land mixed pixels to obtain the flood submerged water body extraction result, see Figure 10 .

[0125] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0126] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A flood disaster monitoring method based on threshold interval and neighborhood analysis, characterized in that: The following steps are involved: Step S1: obtaining polarimetric radar images of the flood-stricken area; preprocessing the polarimetric radar images of the flood-stricken area to obtain radar backscatter coefficient band data; Step S2: plotting a histogram of the radar backscatter coefficient band data to obtain a radar backscatter coefficient histogram; Based on the radar backscatter coefficient histogram, the polarization radar image of the flood-stricken area is subjected to histogram bimodal identification to obtain the backscatter coefficient values ​​of the peak points of land and water bodies; Step S3: Calculate the threshold segmentation interval of the polarimetric radar image of the flood-stricken area based on the backscatter coefficient value of the peak point of land and the backscatter coefficient value of the peak point of water to obtain the radar image threshold segmentation interval; segment the polarimetric radar image of the flood-stricken area into water, land, and mixed water and land pixels based on the radar image threshold segmentation interval to obtain a preliminary water extraction result, wherein the preliminary water extraction result includes water, land, and mixed water and land pixels; wherein step S3 includes the following steps: Step S31: The backscatter coefficient value of the land peak point As the mean of the Gaussian distribution of land, the backscattering coefficient value of the peak point of water body As the Gaussian distribution mean of the water body, the pixel values ​​in the radar backscatter coefficient band data that are less than the Gaussian distribution mean of the water body are recorded as the first candidate pixel set, and the pixel values ​​in the radar backscatter coefficient band data that are greater than the Gaussian distribution mean of the land are recorded as the second candidate pixel set; Step S32: Calculate the standard deviation of the first candidate pixel set to obtain the standard deviation of the water body Gaussian distribution , and calculate the standard deviation of the second candidate pixel set to obtain the standard deviation of the land Gaussian distribution ; Step S33: Use 4 times the standard deviation as the initial segmentation interval of the polarimetric radar image of the flood-stricken area. The lower limit of the initial segmentation interval is the Gaussian distribution mean of the water body + 4 times the Gaussian distribution standard deviation of the water body. The upper limit of the initial segmentation interval is the mean of the Gaussian distribution of land - 4 times the standard deviation of the Gaussian distribution of land , that is, the initial segmentation interval is , determine the initial segmentation interval as Is the lower limit of the interval greater than the upper limit of the interval? If so, the initial segmentation interval is reduced by 0.5 times the standard deviation and the segmentation interval is recalculated until the lower limit of the segmentation interval is less than the upper limit of the segmentation interval, thereby obtaining the radar image threshold segmentation interval as ,k={4,3.5,3,2.5,2,1.5,1,0}; Step S34: Segmenting the polarimetric radar image of the flood-stricken area into water, land, and mixed water and land pixels according to the radar image threshold segmentation interval to obtain a preliminary water extraction result, wherein the preliminary water extraction result includes water, land, and mixed water and land pixels; Step S4: remove pseudo water bodies from the water bodies in the preliminary water body extraction results, and perform neighborhood water and land attribute discrimination on the water and land mixed pixels in the preliminary water body extraction results, thereby obtaining the flood submerged water body extraction results. Step S4 includes the following steps: Step S41: extracting DEM values ​​and generating histograms for the water bodies in the preliminary water body extraction results to obtain a DEM histogram of water body pixels; The DEM histogram of water body pixels is fitted with lognormal distribution using the preset lognormal distribution function, and the DEM value of the cumulative distribution probability curve of the preset lognormal distribution function at 90% is calculated to obtain the water body DEM threshold value. The water body DEM threshold value is recorded as ; Among them, the probability density function and cumulative distribution function formulas of the preset lognormal distribution function are as follows: ； ； in, is the mean value of the water pixel DEM after logarithmic transformation, is the standard deviation of the water pixel DEM after logarithmic transformation, Ф is the cumulative distribution function of the standard normal distribution; Step S42: The water body is higher than The water pixels are recorded as high altitude water bodies. The pseudo water bodies in the high-altitude areas in the preliminary water body extraction results are eliminated, and the neighborhood water and land attributes of the water-land mixed pixels in the preliminary water body extraction results are discriminated to obtain the flood-inundated water body extraction results.

2. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: collecting polarimetric radar images of the target flood-stricken area to obtain polarimetric radar images of the flood-stricken area; Step S12: performing multi-view processing and filtering on the polarimetric radar image of the flood-stricken area to obtain a multi-view filtered image of the flood-stricken area; Step S13: geocoding the multi-view filtered image of the flood-stricken area to obtain a geocoded image of the flood-stricken area; Step S14: performing radiometric calibration on the geocoded image of the flood-stricken area to obtain a polarimetric radar image of the flood-stricken area; Step S15: performing coherent polarization decomposition correction on the polarimetric radar image of the flood-stricken area to obtain radar backscatter coefficient band data.

3. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 1 is characterized in that: Step S2 includes the following steps: Step S21: extracting pixel values ​​from radar backscatter coefficient band data to obtain a one-dimensional array of backscatter coefficients; Step S22: plotting a histogram of the radar backscatter coefficient band data based on the one-dimensional array of backscatter coefficients to obtain a radar backscatter coefficient histogram; Step S23: performing histogram double-peak identification on the radar backscatter coefficient histogram to obtain the backscatter coefficient value of the land peak point and the backscatter coefficient value of the water peak point.

4. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 3 is characterized in that: Step S23 includes the following steps: Perform smooth curve extraction on the radar backscatter coefficient histogram to obtain a smooth curve of the backscatter coefficient histogram; Based on the backscatter coefficient histogram smoothing curve, the histogram bimodal identification of the multi-view filtered image of the flood-stricken area is performed to obtain the backscatter coefficient values ​​of the peak points of land and the peak points of water. The specific processing process of the histogram bimodal identification of the multi-view filtered image of the flood-stricken area based on the backscatter coefficient histogram smoothing curve is as follows: ； ； ； ； ； ； in, is the backscatter coefficient histogram smoothing curve, w is the filter window size, p is the polynomial order of the Savitzky-Golay filter, is the backscattering coefficient value with the largest function value in the extreme point, is the backscattering coefficient value with the second largest function value in the extreme point, is the backscattering coefficient value of the peak point on land, is the backscattering coefficient value at the peak point of the water body.

5. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 1 is characterized in that: Step S34 includes the following steps: Assume that the radar backscatter coefficient band data is I, and the preliminary extraction result of water body is The specific processing process of water body, land and water-land mixed pixel segmentation is as follows: 。 6. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 1 is characterized in that: The water body is higher than The water pixels are recorded as high altitude water bodies. The specific process of removing pseudo water bodies in high-altitude areas from the preliminary water body extraction results is as follows: With the water body pixel in the high altitude area as the center, a neighborhood window is set with a size of N. The mountain shadow index of the water body pixel in the high altitude area is defined as MSI. When the MSI is greater than 5, the water body pixel in the high altitude area is judged to be a mountain shadow pixel. When the MSI is less than or equal to 5, the water body pixel in the high altitude area is judged to be a real water body. The MSI calculation formula is as follows: ； ； ； in, is the number of mixed pixels in the neighborhood window, is the number of water pixels in the neighborhood window, It is the maximum value of the normalized DEM in the neighborhood window.

7. The flood disaster monitoring method based on threshold interval and neighborhood analysis according to claim 1 is characterized in that: The specific process of identifying the neighborhood water and land attributes of the water-land mixed pixels in the preliminary water body extraction results is as follows: Assume that the neighborhood window size of the water-land mixed pixel is S. Based on the water-land mixed pixel, the number of water pixels and the number of land pixels in the neighborhood window S are calculated. According to the attribute discrimination rule of the water-land mixed pixel, the mixed pixel attributes in the water-land mixed pixel are judged. The attribute discrimination rule of the water-land mixed pixel is as follows: ； ； Among them, S is the neighborhood window size of the water-land mixed pixel, and the total number of pixels in the neighborhood is -1, L(i,j) is the number of land pixels, W(i,j) is the number of water pixels, i is the row number of the mixed land and water pixel, and j is the column number of the mixed land and water pixel.

Images