Flood monitoring and deduction method based on unmanned aerial vehicle time sequence radar remote sensing

Abstract

The invention discloses a flood monitoring and deduction method based on unmanned aerial vehicle time sequence radar remote sensing, which comprises the following steps: S1, collecting multi-source flood disaster high-spatial-resolution time sequence radar remote sensing data through unmanned aerial vehicle radar remote sensing, and processing the data into a remote sensing data set with consistent space-time reference; s2, preprocessing each time sequence radar remote sensing image in the remote sensing data set; s3, segmenting a submerged area and a non-submerged area of the preprocessed time sequence radar remote sensing image, removing fine water in the submerged area, and extracting a radar remote sensing high-resolution submerged range; and S4, quantitatively extracting flood evolution characteristic elements, and performing flood disaster potential influence area deduction on the current submerging range in combination with the digital elevation model and hydrological connectivity. According to the method, the problem that the large-range flood evolution process is difficult to dynamically monitor is solved, and support is provided for rapid emergency and risk avoiding transfer of watershed flood by forward deduction of the potential influence area of flood disasters.

Description

Technical Field

[0001] The present invention belongs to the technical field of flood disaster monitoring, and in particular relates to a flood monitoring and deduction method based on unmanned aerial vehicle (UAV) time-series radar remote sensing. Background Art

[0002] Floods are the most frequent, widespread, and costly natural disaster in my country. They can flood large areas of homes, damage infrastructure, and destroy crops, resulting in significant losses to human life, property, and socioeconomic development.

[0003] Drone remote sensing, with its advantages of wide coverage, rapidity, comprehensiveness, and accuracy, is a crucial tool for monitoring flood development. However, optical remote sensing by drones relies on visible and near-infrared wavelengths, making it susceptible to weather conditions. Floods are often accompanied by cloudy and rainy weather, making it difficult for optical remote sensing sensors to obtain clear images of the surface. Furthermore, optical imagery has limitations in water body identification, particularly when distinguishing shallow water areas, mixed pixel areas, and areas covered by vegetation.

[0004] Radar remote sensing has unique advantages. It utilizes microwaves for detection, can penetrate clouds, and is unrestricted by weather conditions, enabling all-day, all-weather observations. Drone radar remote sensing, with its high spatial and temporal resolution, can identify surface flooding and inundation characteristics in detail.

[0005] With the continuous development of drone network observation technology and radar sensors, multi-source collaborative drone radar remote sensing data can provide high-frequency re-entry, high spatial resolution, and high-quality remote sensing imagery, providing important support for rapid monitoring and simulation of flood disaster emergencies. However, due to the high noise, complex backscatter characteristics, and inundated water area morphology of current drone radar remote sensing images, there is currently a lack of effective algorithms for multi-time series drone radar remote sensing flood inundation monitoring. Accurate and timely flood monitoring and simulation are crucial for disaster warning, emergency response, and post-disaster recovery. Furthermore, drone remote sensing currently focuses on emergency monitoring, lacking precise flood simulation.

[0006] In summary, the development of a flood monitoring and deduction method based on UAV time-series radar remote sensing can effectively utilize the high temporal and spatial resolution advantages of multi-time-series radar remote sensing, overcome the shortcomings of traditional methods, and respond to the urgent needs of emergency risk avoidance in flood disasters. Summary of the Invention

[0007] In response to the above-mentioned deficiencies in the existing technology, the flood monitoring and deduction method based on UAV time-series radar remote sensing provided by the present invention solves the problem that the existing related methods are difficult to dynamically monitor the evolution process of large-scale floods and lack accurate deduction.

[0008] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a flood monitoring and deduction method based on UAV time-series radar remote sensing, comprising the following steps: S1. Collect high spatial resolution time series radar remote sensing data of multi-source flood disasters through UAV radar remote sensing, and process it into a remote sensing dataset with consistent temporal and spatial references; S2, preprocessing each time series radar remote sensing image in the remote sensing data set; S3, segmenting the pre-processed time series radar remote sensing image into submerged and non-submerged areas, removing small water bodies in the submerged areas, and extracting the high-resolution submerged range of radar remote sensing; S4. Quantitatively extract characteristic elements of flood evolution, and use the digital elevation model and improved Manning formula to deduce the potential impact area of ​​flood disasters within the current inundation range.

[0009] Furthermore, in step S1, the processing of the time series radar remote sensing data includes: By establishing an affine transformation model that represents the transformation relationship between the coordinates of the ground control points and the image coordinates of the time series radar remote sensing data, the image coordinates of the time series radar remote sensing data are converted into the actual geographic coordinates in the geodetic coordinate system; And convert the digital quantization value of the radar image in the time series radar remote sensing data into the radar backscatter coefficient.

[0010] Furthermore, step S2 includes the following sub-steps: S21, performing cross-sensor radiometric normalization processing on each time series radar remote sensing image; S22. Correct the histogram of the time series radar remote sensing image after radiometric normalization using the percentile interception method; S23. Using a waveform correction method to filter each corrected time series radar remote sensing image.

[0011] Furthermore, the step S21 includes the following sub-steps: S21-1. Find the ground objects with relatively stable radiation characteristics in the radar remote sensing images of different phases and sensors as pseudo-invariant feature points; S21-2. Establishing a radiation conversion model from the source sensor to the target sensor based on the radiation values ​​of the selected pseudo-invariant feature points; S21-3. Apply the established radiation conversion model to the entire source sensor time series radar remote sensing image to make its radiation characteristics consistent with the target sensor time series radar remote sensing image.

[0012] Furthermore, the step S22 is specifically as follows: S22-1. Calculate the cumulative probability distribution of the radiometrically normalized time-series radar remote sensing image histogram; S22-2. Find, based on the calculated cumulative probability distribution, a lower grayscale value corresponding to a preset lower threshold when the cumulative probability distribution value is equal to a preset upper threshold, and an upper grayscale value corresponding to a preset upper threshold when the cumulative probability distribution value is equal to a preset upper threshold; S22-3, truncating the grayscale values ​​less than the lower limit grayscale value to the lower limit grayscale value, truncating the grayscale values ​​greater than the upper limit grayscale value to the upper limit grayscale value, and then recalculating the time series radar remote sensing image histogram to complete the correction.

[0013] Furthermore, in step S23, the filtering method is: Each pixel in each time series radar remote sensing image As the center, take a neighborhood window and perform weighted summation on the pixels in the window. The expression is:

[0014]

[0015] Where, Represents the grayscale value of the pixel in the window after weighted summation, Represents the grayscale value of the pixel, represents the waveform correction filter function, represents the standard deviation of the filter function, represents the waveform correction coefficient, i, j Indicates the horizontal and vertical coordinates of the pixels in the window. n Indicates the total number of pixels within the window.

[0016] Furthermore, in step S3, the method for dividing the flooded area and the non-flooded area is specifically as follows: S31, dividing the preprocessed time series radar remote sensing image into a plurality of local areas, and calculating the polarization ratio of each local area; S32. In each local area, based on the backscattering intensity and the polarization ratio, a local Otsu method is used to determine the optimal segmentation threshold of each local area; S33 , dividing the pixels in each local area into a flooded area and a masked area according to the optimal segmentation threshold.

[0017] Furthermore, in step S31, each local area R ij , polarization ratio for:

[0018] Where, represents the backscattering intensity of HH polarization, represents the backscatter intensity of VV polarization; In step S32, when determining the optimal segmentation threshold of each local area, the objective function of the local Otsu method is to maximize the inter-class variance in the local area. :

[0019] Where, t represents the grayscale threshold, and Respectively indicate that the gray value is less than or equal to t and greater than t The probability that a pixel is in the local area, and Represent the average grayscale values ​​of these two types of pixels respectively; In step S32, Maximum grayscale threshold t As a local optimal threshold .

[0020] Furthermore, the step S4 includes the following sub-steps: S41. Based on the normal water area, quantitatively calculate the flood evolution characteristics within the inundation area, including inundation area, inundation speed, and water head position; S42. Based on the surface roughness and digital elevation model within the inundation area, an improved Manning formula based on a simplified hydrodynamic model is constructed to calculate the flood flow velocity; S43, based on the calculation of flood flow velocity, starting from the boundary pixel of the current inundation range, and expanding in a certain step time according to the flow direction and terrain slope; S44. For each expanded pixel, when its elevation value is less than the average elevation value of the surrounding areas according to the inundation range, the pixel is designated as a potential flood inundation area, thereby realizing the deduction of the potential impact area of ​​flood disasters.

[0021] Furthermore, in step S41, the flooded area express:

[0022] The flooding speed express:

[0023] The water head position is: the position coordinates of the boundary pixel at the forefront of the flooded range ; Where, express The flooding range of time, Indicates flooding range The actual area corresponding to a single pixel in Indicates flooding range The pixels inside express The extent of flooding at the moment; The step S42 includes the following sub-steps: S42-1. Use time-series radar remote sensing images to obtain the surface roughness of the flooded area, which reflects the surface resistance characteristics, and quantify it into the roughness coefficient required by the hydrodynamic model; S42-2. Based on the digital elevation model of the inundation range and the surface roughness coefficient, an improved Manning formula is constructed to simulate the surface water flow velocity and direction, and then the flood water flow velocity is calculated: Among them, the improved Manning formula is:

[0024] Where, Indicates the flood water flow speed, represents the roughness coefficient, Indicates the local water depth. Indicates the local terrain slope; among them, the flood flow speed and flow direction The initial slope direction is determined by analyzing the elevation difference between the central pixel and its adjacent pixels.

[0025] The beneficial effects of the present invention are: (1) This invention constructs a flood monitoring and prediction method based on UAV time-series radar remote sensing, which breaks through the problem of difficulty in dynamic monitoring of large-scale flood evolution process; (2) The present invention constructs a radar remote sensing image preprocessing method that combines histogram statistical matching and waveform correction with a Gaussian filtering method to achieve radar value threshold optimization and image noise reduction; (3) The present invention adopts the variance shift Otsu method and the neighborhood window opening and closing method to achieve high-resolution submergence range extraction of radar remote sensing; (4) The present invention quantitatively extracts characteristic elements of flood evolution and combines digital elevation models and hydrological connectivity to achieve forward deduction of potential impact areas of flood disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of the flood monitoring and deduction method based on UAV time-series radar remote sensing provided by the present invention. DETAILED DESCRIPTION

[0027] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0028] The embodiment of the present invention provides a flood monitoring and deduction method based on UAV time series radar remote sensing, such as Figure 1 As shown, the following steps are included: S1. Collect high spatial resolution time series radar remote sensing data of multi-source flood disasters through UAV radar remote sensing, and process it into a remote sensing dataset with consistent temporal and spatial references; S2, preprocessing each time series radar remote sensing image in the remote sensing data set; S3, segmenting the pre-processed time series radar remote sensing image into submerged and non-submerged areas, removing small water bodies in the submerged areas, and extracting the high-resolution submerged range of radar remote sensing; S4. Quantitatively extract characteristic elements of flood evolution, and use the digital elevation model and improved Manning formula to deduce the potential impact area of ​​flood disasters within the current inundation range.

[0029] In step S1 of the embodiment of the present invention, the time-series radar remote sensing data is the time-series radar remote sensing data obtained through the UAV network observation and the carried radar multi-band payload. In addition, the digital elevation model (DEM) data and vegetation coverage (NDVI) data of the area to be monitored are obtained through the carried lidar sensor.

[0030] Specifically, a drone network observation system is formed through the original drone ground base stations and drones deployed for flood emergency. Each drone is equipped with Ka-band, C-band or X-band radar payloads and lidar sensors to conduct high-frequency, multi-range data collection in flood-affected areas, and realize the rapid acquisition of time-series radar remote sensing data in flood-affected areas to ensure that the temporal and spatial evolution process of flood disasters can be fully reflected.

[0031] The UAV remote sensing image will produce geometric deformation due to the characteristics of the multi-sensor itself, the attitude change of the UAV platform and the influence of the earth curvature factor. Therefore, in this embodiment, it is necessary to process it. In step S1 of this embodiment, the processing of the time series radar remote sensing data includes: By establishing an affine transformation model that represents the transformation relationship between the coordinates of the ground control points and the image coordinates of the time series radar remote sensing data, the image coordinates of the time series radar remote sensing data are converted into the actual geographic coordinates in the geodetic coordinate system; And convert the digital quantization value of the radar image in the time series radar remote sensing data into the radar backscatter coefficient.

[0032] Specifically, in this embodiment, it is assumed that the coordinates of the ground control point are (In the actual geodetic coordinate system), the corresponding coordinates on the image are , the established affine transformation model is expressed as:

[0033] Where, These are the transformation parameters to be determined; these parameters are solved through multiple ground control points to transform the image coordinates into actual geographic coordinates and correct the geometric deformation of the image.

[0034] In this embodiment, when converting the digital quantization value of the radar image into the radar backscatter coefficient, for the multi-time series spaceborne radar sensor, the following calibration formula exists:

[0035] Where, is the calibration constant, defined by the sensor on the drone; A digital quantization value representing an image; It is a parameter related to the radar beam illumination area, etc.

[0036] In this embodiment, the DN value can be converted into a backscatter coefficient with physical meaning through calibration, so as to facilitate comparison between different images.

[0037] Step S2 in the embodiment of the present invention includes the following sub-steps: S21, performing cross-sensor radiometric normalization processing on each time series radar remote sensing image; S22. Correct the histogram of the time series radar remote sensing image after radiometric normalization using the percentile interception method; S23. Using a waveform correction method, filter each corrected time-series radar remote sensing image.

[0038] In step S21 of this embodiment, to address the radiometric differences between multi-source time-series radar remote sensing data, a relative radiometric normalization method based on pseudo-invariant features (PIFs) is used to improve the temporal and spatial consistency of the data. The method includes the following steps: S21-1. Find the ground objects with relatively stable radiation characteristics in the radar remote sensing images of different phases and sensors as pseudo-invariant feature points; S21-2. Establishing a radiation conversion model from the source sensor to the target sensor based on the radiation values ​​of the selected pseudo-invariant feature points; S21-3. Apply the established radiation conversion model to the entire source sensor time series radar remote sensing image to make its radiation characteristics consistent with the target sensor time series radar remote sensing image.

[0039] Specifically, in step S21-1 of this embodiment, within the study area, objects with stable backscattering characteristics in the time series radar remote sensing images of different phases and different sensors are selected as pseudo-invariant features. The selection criteria can comprehensively consider the object type information and the time series variability of the backscattering coefficient. For example, for any pixel exist Backscatter coefficient at time , its time series coefficient of variation It can be expressed as:

[0040] Select Areas below the set threshold are considered as potential PIFs areas and are further screened based on the feature type information.

[0041] In step S21-2 of this embodiment, based on the selected Pseudo-invariant feature points ,in and Represents the source sensor and target sensor in the The backscattering coefficient values ​​at the pseudo-invariant feature points are used to construct a nonlinear regression model. In this embodiment, a quadratic polynomial regression model is used to construct a radiation conversion model, which is expressed as:

[0042] in, 、 、 is the unknown regression coefficient, which is solved by the least square method; is the residual term.

[0043] In step S21-3 of this embodiment, the obtained radiation conversion model is applied to each pixel of the time series radar remote sensing image of the source sensor. , get the corrected backscatter coefficient value , which is expressed as:

[0044] Through the above method, the relative radiometric normalization processing of the source sensor time series radar remote sensing image to the target source sensor time series radar remote sensing image is realized.

[0045] Step S22 of this embodiment is specifically as follows: S22-1. Calculate the cumulative probability distribution of the histogram of the radiometrically normalized time-series radar remote sensing image ; S22-2. According to the calculated cumulative probability distribution, find the lower limit gray value corresponding to the preset lower limit threshold when the cumulative probability distribution value is equal to the preset lower limit threshold. , and the upper grayscale value corresponding to the preset upper threshold when the cumulative probability distribution value is ; For example, the above preset lower limit threshold is expressed as , the preset upper threshold is expressed as ; S22-3. Cut grayscale values ​​less than the lower grayscale limit to the lower grayscale limit. , truncate the grayscale value greater than the upper grayscale value to the upper grayscale value , and then recalculate the histogram of the time series radar remote sensing image to complete the correction.

[0046] In this embodiment, the above-mentioned image histogram correction process reduces the influence of abnormal values ​​at both ends of the image histogram on the subsequent processing of the radar image.

[0047] In step S23 of this embodiment, the filtering method is: Each pixel in each time series radar remote sensing image As the center, take a neighborhood window and perform weighted summation on the pixels in the window. The expression is:

[0048]

[0049] Where, Represents the grayscale value of the pixel in the window after weighted summation, Represents the grayscale value of the pixel, represents the waveform correction filter function, represents the standard deviation of the filter function, represents the waveform correction coefficient, i, j Indicates the horizontal and vertical coordinates of the pixels in the window. n Indicates the total number of pixels within the window.

[0050] In this embodiment, the above filtering process removes speckle noise from the time-series radar remote sensing image. By setting and adjusting the waveform correction coefficient, the filtering effect is optimized, and the influence of house shadows and mountain shadows on water area extraction is removed to the greatest extent.

[0051] In step S3 of the embodiment of the present invention, a partitioned adaptive threshold segmentation method combining polarization information is used to improve the accuracy of segmenting flooded areas and non-flooded areas in time series radar remote sensing images, including the following steps: S31, dividing the preprocessed time series radar remote sensing image into a plurality of local areas, and calculating the polarization ratio of each local area; S32. In each local area, based on the backscattering intensity and the polarization ratio, a local Otsu method is used to determine the optimal segmentation threshold of each local area; S33 , dividing the pixels in each local area into a flooded area and a masked area according to the optimal segmentation threshold.

[0052] In step S31 of this embodiment, the pre-processed time series radar remote sensing image is divided into non-overlapping or partially overlapping local area R ij , in each local area R ij Based on the radar multi-polarization information (HH and VV polarization), the polarization ratio is calculated. :

[0053] Where, represents the backscattering intensity of HH polarization, represents the backscatter intensity of VV polarization.

[0054] In step S32 of this embodiment, the backscattering intensity of the local area is combined and polarization ratio , the local Otsu method is used to calculate the optimal segmentation threshold of the region .

[0055] Among them, the objective function of the local Otsu method is to maximize the inter-class variance in the local area :

[0056] Where, t is the grayscale threshold, and Grayscale values ​​are less than or equal to t and greater than t The probability that a pixel is in the local area, and are the average grayscale values ​​of these two types of pixels respectively.

[0057] Based on the inter-class variance determined by the above method, we traverse all possible grayscale values ​​in the local area to find the The largest t is used as the local optimal threshold .

[0058] In step S33 of this embodiment, for the local area R ijEach pixel within , according to its backscattering intensity With the local optimal threshold The comparison results are classified.

[0059] Step S4 of the embodiment of the present invention includes the following sub-steps: S41. Based on the normal water area, quantitatively calculate the flood evolution characteristics within the inundation area, including inundation area, inundation speed, and water head position; S42. Based on the surface roughness and digital elevation model within the inundation area, an improved Manning formula based on a simplified hydrodynamic model is constructed to calculate the flood flow velocity; S43, based on the flood flow velocity, starting from the boundary pixel of the current inundation range, and expanding in accordance with the flow direction and terrain slope, with a certain step time; S44. For each expanded pixel, when its elevation value is less than the average elevation value of the surrounding areas according to the inundation range, the pixel is designated as a potential flood inundation area, thereby realizing the deduction of the potential impact area of ​​flood disasters.

[0060] In step S41 of this embodiment, the flooded area express:

[0061] Submergence speed express:

[0062] The water head position is: the position coordinates of the boundary pixel at the forefront of the flooded range ; Where, express The flooding range of time, Indicates flooding range The actual area corresponding to a single pixel in Indicates flooding range The pixels inside express The flooding range at the moment.

[0063] In this embodiment, step S42 includes the following steps: S42-1. Use time-series radar remote sensing images to obtain the surface roughness of the flooded area, which reflects the surface resistance characteristics, and quantify it into the roughness coefficient required by the hydrodynamic model; S42-2. Based on the digital elevation model of the inundation range and the surface roughness coefficient, an improved Manning formula is constructed to simulate the surface water flow velocity and direction, and then the flood water flow velocity is calculated: Among them, the improved Manning formula is:

[0064] Where, Indicates the flood water flow speed, represents the roughness coefficient, represents the local water depth (which can be initially assumed to be a minimum value or inferred from the existing flooded range observed on the ground), Indicates the local terrain slope; among them, the flood flow speed and flow direction The initial slope direction is determined by analyzing the elevation difference between the central pixel and its adjacent pixels.

[0065] In step S42-1 of this embodiment, the normalized vegetation index and ground radar coherence are extracted using time-series radar remote sensing images to extract the surface roughness information of the study area and quantify it into the Manning roughness coefficient. ; For example, an empirical relationship between vegetation index and Manning's roughness coefficient can be established.

[0066] In step S43 and step S44 of this embodiment, for the boundary pixels of the current flooded range The priority direction of flood expansion is directly determined by the local water velocity vector Specifically, starting from the current flooded area boundary pixel, according to the flooded water flow direction and speed, the adjacent pixel direction is indicated by a certain step time. For each expanded pixel If the pixel elevation is expanded , then the pixel is marked as a potential flooding area, otherwise the pixel is considered to be a non-flooding area, where Calculated based on the average elevation of the surrounding areas of the flooded area.

[0067] Based on the above process, by calculating the time steps of different flood evolutions, the potential flood inundation area is delineated, thereby realizing the forward deduction of flood disaster image areas, and combining the flood disaster-bearing bodies for early transfer and resettlement.

[0068] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

[0069] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A flood monitoring and prediction method based on UAV time series radar remote sensing, characterized in that: The following steps are involved: S1. Collect high spatial resolution time series radar remote sensing data of multi-source flood disasters through UAV radar remote sensing, and process it into a remote sensing dataset with consistent temporal and spatial references; S2, preprocessing each time series radar remote sensing image in the remote sensing data set; S3, segmenting the pre-processed time series radar remote sensing image into submerged and non-submerged areas, removing small water bodies in the submerged areas, and extracting the high-resolution submerged range of radar remote sensing; S4. Quantitatively extract characteristic elements of flood evolution, and use the digital elevation model and improved Manning formula to deduce the potential impact area of ​​flood disasters within the current inundation range.

2. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 1 is characterized in that: In step S1, the processing of the time series radar remote sensing data includes: By establishing an affine transformation model that represents the transformation relationship between the coordinates of the ground control points and the image coordinates of the time series radar remote sensing data, the image coordinates of the time series radar remote sensing data are converted into the actual geographic coordinates in the geodetic coordinate system; And convert the digital quantization value of the radar image in the time series radar remote sensing data into the radar backscatter coefficient.

3. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 1 is characterized in that: The step S2 comprises the following sub-steps: S21, performing cross-sensor radiometric normalization processing on each time series radar remote sensing image; S22. Correct the histogram of the time series radar remote sensing image after radiometric normalization using the percentile interception method; S23. Using a waveform correction method, filter each corrected time-series radar remote sensing image.

4. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 3 is characterized in that: The step S21 includes the following sub-steps: S21-1. Find the ground objects with relatively stable radiation characteristics in the radar remote sensing images of different phases and sensors as pseudo-invariant feature points; S21-2. Establishing a radiation conversion model from the source sensor to the target sensor based on the radiation values ​​of the selected pseudo-invariant feature points; S21-3. Apply the established radiation conversion model to the entire source sensor time series radar remote sensing image to make its radiation characteristics consistent with the target sensor time series radar remote sensing image.

5. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 3 is characterized in that: The step S22 is specifically as follows: S22-1. Calculate the cumulative probability distribution of the radiometrically normalized time-series radar remote sensing image histogram; S22-2. Find, based on the calculated cumulative probability distribution, a lower grayscale value corresponding to a preset lower threshold when the cumulative probability distribution value is equal to a preset upper threshold, and an upper grayscale value corresponding to a preset upper threshold when the cumulative probability distribution value is equal to a preset upper threshold; S22-3, truncating the grayscale values ​​less than the lower limit grayscale value to the lower limit grayscale value, truncating the grayscale values ​​greater than the upper limit grayscale value to the upper limit grayscale value, and then recalculating the time series radar remote sensing image histogram to complete the correction.

6. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 3 is characterized in that: In step S23, the filtering method is: Each pixel in each time series radar remote sensing image As the center, take a neighborhood window and perform weighted summation on the pixels in the window. The expression is: Where, Represents the grayscale value of the pixel in the window after weighted summation, Represents the grayscale value of the pixel, represents the waveform correction filter function, represents the standard deviation of the filter function, represents the waveform correction coefficient, i, j Indicates the horizontal and vertical coordinates of the pixels in the window. n Indicates the total number of pixels within the window.

7. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 1 is characterized in that: In step S3, the method for dividing the flooded area into the non-flooded area is specifically as follows: S31, dividing the preprocessed time series radar remote sensing image into a plurality of local areas, and calculating the polarization ratio of each local area; S32. In each local area, based on the backscattering intensity and the polarization ratio, a local Otsu method is used to determine the optimal segmentation threshold of each local area; S33 , dividing the pixels in each local area into a flooded area and a masked area according to the optimal segmentation threshold.

8. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 7 is characterized in that: In step S31, each local area R ij , polarization ratio for: Where, represents the backscattering intensity of HH polarization, represents the backscatter intensity of VV polarization; In step S32, when determining the optimal segmentation threshold of each local area, the objective function of the local Otsu method is to maximize the inter-class variance in the local area. : Where, t represents the grayscale threshold, and Respectively indicate that the gray value is less than or equal to t and greater than t The probability that a pixel is in the local area, and Represent the average grayscale values ​​of these two types of pixels respectively; In the step S32, Maximum grayscale threshold t As a local optimal threshold .

9. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 1 is characterized in that: The step S4 comprises the following sub-steps: S41. Based on the normal water area, quantitatively calculate the flood evolution characteristics within the inundation area, including inundation area, inundation speed, and water head position; S42. Based on the surface roughness and digital elevation model within the inundation area, an improved Manning formula based on a simplified hydrodynamic model is constructed to calculate the flood flow velocity; S43, based on the flood flow velocity, starting from the boundary pixel of the current inundation range, and expanding in accordance with the flow direction and terrain slope, with a certain step time; S44. For each expanded pixel, when its elevation value is less than the average elevation value of the surrounding areas according to the inundation range, the pixel is designated as a potential flood inundation area, thereby realizing the deduction of the potential impact area of ​​flood disasters.

10. The flood monitoring and deduction method based on UAV time series radar remote sensing according to claim 9 is characterized in that: In step S41, the flooded area express: The flooding speed express: The water head position is: the position coordinates of the boundary pixel at the forefront of the flooded range ; Where, express The flooding range of time, Indicates flooding range The actual area corresponding to a single pixel in Indicates flooding range The pixels inside express The extent of flooding at the moment; The step S42 includes the following sub-steps: S42-1. Use time-series radar remote sensing images to obtain the surface roughness of the flooded area, which reflects the surface resistance characteristics, and quantify it into the roughness coefficient required by the hydrodynamic model; S42-2. Based on the digital elevation model of the inundation range and the surface roughness coefficient, an improved Manning formula is constructed to simulate the surface water flow velocity and direction, and then the flood water flow velocity is calculated: Among them, the improved Manning formula is: Where, Indicates the flood water flow speed, represents the roughness coefficient, Indicates the local water depth. Indicates the local terrain slope; among them, the flood flow speed and flow direction The initial slope direction is determined by analyzing the elevation difference between the central pixel and its adjacent pixels.

Images