River flow remote sensing monitoring method and device

Abstract

The present invention provides a method and device for remote sensing monitoring of river flow, the method comprising: reconstructing missing images of the water surface based on water body occurrence rate and / or water surface isobaths to obtain reconstructed time-series binary water body images; obtaining a time-series river width data set based on the reconstructed time-series binary water body images; simulating underwater terrain based on elevation cycle iteration to obtain a relationship curve between river width and river flow; obtaining a time-series river flow data set based on the time-series river width data set and the relationship curve between river width and river flow. The present invention reconstructs the water surface to obtain river water surface distribution data without manual interpretation, simulates the underwater terrain to obtain river cross-section terrain data without actual measurement of the terrain, and constructs an automated river flow monitoring solution.

Description

Technical Field

[0001] The invention relates to the technical field of river runoff monitoring, and in particular to a river flow remote sensing monitoring method and device. Background Art

[0002] Satellite remote sensing technology is widely used in river flow monitoring and is highly anticipated due to its advantages such as timeliness, efficiency, regular revisits, and wide observation range.

[0003] At present, there are many studies on estimating runoff based on remote sensing. For example, microwave remote sensing, low-altitude remote sensing and underwater topographic mapping technology are used to obtain river hydraulic parameters, and the flow rate is estimated by combining the Manning formula to monitor flow. However, in this scheme, the calculation of river width requires manual measurement, and the underwater topography needs to be measured, so the degree of automation is low.

[0004] Therefore, building an automated river flow monitoring solution is a technical problem that needs to be solved urgently. Summary of the invention

[0005] The invention provides a river flow remote sensing monitoring method and device, which are used to solve the defect of low automation degree of river flow monitoring scheme in the prior art.

[0006] The present invention provides a river flow remote sensing monitoring method, comprising:

[0007] Reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image;

[0008] Based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained;

[0009] The underwater terrain is simulated based on elevation cycle iteration to obtain the relationship curve between river width and river flow;

[0010] Based on the time-series river width dataset and the relationship curve between the river width and the river flow, a time-series river flow dataset is obtained.

[0011] Optionally, reconstructing the water surface of the water surface missing image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image includes:

[0012] Masking a water body occurrence rate image using the water surface missing image to obtain a water body occurrence rate histogram of the water surface missing image;

[0013] Determine a water body occurrence rate threshold value according to a water body occurrence rate histogram of the water surface missing image;

[0014] Water body reconstruction processing is performed on pixels in the water surface missing image that are greater than the water body occurrence rate threshold to obtain a reconstructed time-series binary water body image.

[0015] Optionally, reconstructing the water surface of the water surface missing image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image includes:

[0016] Determine the ratio of a first water area in the water surface missing image to a second water area in the water surface missing image; the first water area is the water area between the water body boundary and the lower depth contour of the adjacent depth contour in the non-missing area of ​​the water surface missing image; the second water area is the water area between the minimum neighboring depth contour areas in the non-missing area of ​​the water surface missing image;

[0017] Determine the position of boundary points between adjacent depth contours according to the ratio;

[0018] The boundary point positions are connected to obtain the reconstructed time-series binary water body image.

[0019] Optionally, reconstructing the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image, including:

[0020] Reconstruct the missing water surface image based on the water body occurrence rate to obtain a reconstructed first time-series binary water body image;

[0021] The reconstructed first time-series binary water body image is corrected based on the water surface isobath to obtain a reconstructed second time-series binary water body image.

[0022] Optionally, the underwater terrain is simulated based on the elevation cycle iteration to obtain a relationship curve between the river width and the river flow, including:

[0023] Determine the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels; the adjacent pixels are pixels adjacent to the pixel to be simulated;

[0024] Based on the elevation of the pixel to be simulated, the underwater terrain is obtained;

[0025] Based on the underwater topography, river width and hydraulic gradient, the relationship curve between river width and river flow is obtained.

[0026] Optionally, before determining the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels, the following steps are included:

[0027] Determine the deepest point of the river;

[0028] Taking the deepest point of the river channel as the boundary, the underwater terrain on the corresponding side is simulated according to the elevation and slope information of the above-water pixels.

[0029] Optionally, before determining the deepest point of the river, include:

[0030] Determine the main direction of the river;

[0031] Rotating the river channel edge image according to the main direction of the river channel until the cross section of the river channel edge image is horizontal;

[0032] Averaging the elevation values ​​in the river channel edge image along a direction perpendicular to the cross section line to obtain a river channel cross section;

[0033] The water surface of the river cross section is masked to obtain the underwater terrain area to be simulated.

[0034] Optionally, determine the main direction of the river channel, including:

[0035] transforming the river channel edge image from the spatial domain to the frequency domain;

[0036] In the frequency domain, the origin of the river channel edge image is used as the starting point to divide the sectors and determine the energy of each sector;

[0037] The direction corresponding to the sector with the largest energy is determined as the main direction of the river channel.

[0038] Optionally, the water body incidence image is obtained by the following steps:

[0039] Water body extraction is performed on multi-source remote sensing water body images over many years to obtain time series binary water body images;

[0040] The time series binary water body images are superimposed to obtain the water body occurrence rate image.

[0041] Optionally, based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained, including:

[0042] Performing distance transformation, image gradient and morphological skeletonization processing on the reconstructed time-series binary water body image to obtain a river channel centerline image;

[0043] The time series river width dataset is obtained based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary.

[0044] Optionally, based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary, acquiring a time series river width dataset includes:

[0045] Convolving the river centerline image with a convolution kernel to obtain a convolution image;

[0046] Determining the normal direction of the centerline of the river channel based on the number of intersections between the river centerline image and the edge of the convolution kernel and the convolution value in the convolution image;

[0047] The time series river width dataset is obtained based on the intersection of the normal line of the river centerline along the normal line direction and the river boundary.

[0048] The present invention also provides a river flow remote sensing monitoring device, comprising:

[0049] A first acquisition module is used to reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath, and obtain a reconstructed time-series binary water body image;

[0050] A second acquisition module is used to acquire a time-series river width dataset based on the reconstructed time-series binary water body image;

[0051] The third acquisition module is used to simulate the underwater terrain based on the elevation cycle iteration to obtain the relationship curve between the river width and the river flow;

[0052] The fourth acquisition module is used to acquire a time-series river flow data set based on the time-series river width data set and the relationship curve between the river width and the river flow.

[0053] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of any of the above-mentioned river flow remote sensing monitoring methods are implemented.

[0054] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned river flow remote sensing monitoring methods are implemented.

[0055] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps of any of the above-mentioned river flow remote sensing monitoring methods are implemented.

[0056] The river flow remote sensing monitoring method and device provided by the present invention reconstruct the water surface to obtain river surface distribution data without manual interpretation, simulate the underwater terrain to obtain river cross-section terrain data without actual measurement of the terrain, and construct an automated river flow monitoring solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or 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 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.

[0058] Figure 1It is a flow chart of a river flow remote sensing monitoring method provided by an embodiment of the present invention;

[0059] Figure 2 is a schematic diagram of determining a water body occurrence rate threshold provided by an embodiment of the present invention;

[0060] Figure 3 is a schematic diagram of determining the hydraulic gradient provided by an embodiment of the present invention;

[0061] Figure 4 is a schematic diagram of determining the river width provided by an embodiment of the present invention;

[0062] Figure 5 is a schematic diagram of river section topographic data preprocessing provided by an embodiment of the present invention;

[0063] Figure 6 is a schematic diagram for determining the deepest point of a river provided by an embodiment of the present invention;

[0064] Figure 7 is a schematic structural diagram of a river flow remote sensing monitoring device provided by an embodiment of the present invention;

[0065] Figure 8 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0067] Figure 1 FIG. 1 is a flow chart of a river flow remote sensing monitoring method provided by an embodiment of the present invention. Figure 1 As shown, the present invention provides a river flow remote sensing monitoring method, the method comprising:

[0068] Step 101 : reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image.

[0069] Specifically, the remote sensing image interpretation work (such as water system mapping, water body boundary and centerline extraction, etc.) in the present invention is completed on Google Earth Engine (GEE) to maximize automation and reduce manual operation and human intervention. GEE stores PB-level remote sensing images with a time span of up to 40 years, has rich spatial analysis tools and development potential, and has an absolute advantage in large-scale, long-time series target extraction.

[0070] The data sources in the present invention mainly include the Thematic Mapper (TM) images of Landsat 5, the Enhanced Thematic Mapper (ETM+) images of Landsat 7, the Operational Land Imager (OLI) images of Landsat 8, the optical images of Sentinel-2, the synthetic aperture radar images of Sentinel-1, and the auxiliary data mainly include the Digital Elevation Model (DEM) with a resolution of 30 meters of the Shuttle Radder Topograaphy Misson (SRTM). In addition, the global water distribution data set of the Joint Research Centre of European Commission (JRC) is used for water extraction results for verification. The use data of the present invention are all from the GEE platform.

[0071] Collect surface water images taken by satellites and extract water bodies from them. Water bodies are extracted from surface water images by calculating water body index, edge detection, morphological expansion operation, threshold segmentation and slope correction.

[0072] For optical images, the Normalized Difference Water Index (NDWI) or the Modified Normalized Difference Water Index (MNDWI) is used to calculate the water index.

[0073] The expression of NDWI is as follows:

[0074]

[0075] Where NDWI is the normalized water index, ρ NIR represents the near-infrared band of the image, ρRED Represents the red light band of the image.

[0076] The expression of MNDWI is as follows:

[0077]

[0078] Where MNDWI is the improved normalized water index, ρ GREEN represents the green light band of the image, ρ SWNI Represents the image shortwave infrared band.

[0079] After calculating the water index, the image is smoothed using Gaussian filtering to remove noise; the image gradient is then calculated to obtain possible water body edges; non-maximum suppression technology is used to eliminate false edge detection; and finally a double threshold method is used to determine potential water body edges.

[0080] The expression of the value of the Gaussian kernel function at the position (i, j) is as follows:

[0081]

[0082] In the formula, H i,j is the value of the Gaussian kernel function at position (i, j), σ is the variance, and k is the dimension of the kernel matrix.

[0083] The expressions of image gradients in the x and y directions are as follows:

[0084]

[0085]

[0086] In the formula, Respectively represent the image gradient in the x and y directions, and I(x, y) represents the intensity value of the image at (x, y).

[0087] A morphological dilation operation is applied to the detected water body edges to compute potential water and land pixels located near water. The resulting distribution should be bimodal so that land and water are clearly distinguished from the distribution.

[0088] The expression of the intensity image after morphological dilation is as follows:

[0089]

[0090] In the formula, I c is the intensity image after morphological expansion, I is the original intensity value image, C is the edge detection result, and S is the expansion operator.

[0091] The intensity image segmentation threshold after morphological expansion is obtained according to the Otsu method. After the image is binarized by the image segmentation threshold, the image is divided into a foreground image and a background image. The inter-class variance of the foreground image and the background image obtained by the Otsu method is the largest.

[0092] The expression for the ratio of the number of pixels of the foreground image to the entire image is as follows:

[0093] ω 0 =N 0 / (M×N)

[0094] In the formula, ω 0 Indicates the ratio of the number of pixels of the foreground image to the entire image, N 0 It represents the number of pixels in the image whose grayscale value is less than the threshold T, M represents the number of pixels in the horizontal direction of the image, and N represents the number of pixels in the vertical direction of the image.

[0095] The expression for the ratio of the number of pixels of the background image to the whole image is as follows:

[0096] ω 1 =N 1 / (M×N)

[0097] In the formula, ω 1 Indicates the ratio of the number of pixels of the background image to the entire image, N 1 It represents the number of pixels in the image whose grayscale is greater than the threshold T, M represents the number of pixels in the horizontal direction of the image, and N represents the number of pixels in the vertical direction of the image.

[0098] The relationship between the number of pixels whose grayscale values ​​in the image are less than the threshold, the number of pixels whose grayscale values ​​in the image are less than the threshold, and the size of the image is as follows:

[0099] N 0 +N 1 =M×N

[0100] Where N 0 Indicates the number of pixels in the image whose grayscale value is less than the threshold T, N 1 It represents the number of pixels in the image whose grayscale is greater than the threshold T, M represents the number of pixels in the horizontal direction of the image, and N represents the number of pixels in the vertical direction of the image.

[0101] The relationship between the ratio of the number of pixels in the foreground image to the ratio of the number of pixels in the background image to the ratio of the number of pixels in the background image is as follows:

[0102] ω 0 +ω 1 =1

[0103] In the formula, ω 0 Indicates the ratio of the number of pixels of the foreground image to the entire image, ω 1 Indicates the ratio of the number of pixels of the background image to the entire image.

[0104] The expression of the total average grayscale of the image is as follows:

[0105] μ=ω 0 *μ 0 +ω 1 *μ 1

[0106] In the formula, μ represents the total average grayscale of the image, ω 0 Indicates the ratio of the number of pixels in the foreground to the entire image, μ 0 represents the average grayscale of the foreground, ω 1 Indicates the ratio of background pixels to the entire image, μ 1 Represents the average grayscale of the background.

[0107] The expression for between-class variance is as follows:

[0108] g=ω 0 *(μ 0 -μ) 2 +ω 1 *(μ 1 -μ) 2

[0109] In the formula, g represents the between-class variance, ω 0 Indicates the ratio of the number of pixels in the foreground to the entire image, μ 0 represents the average grayscale of the foreground, μ represents the total average grayscale of the image, ω 1 Indicates the ratio of background pixels to the entire image, μ 1 Represents the average grayscale of the background.

[0110] The expression for a pixel in an image that is water is as follows:

[0111] Water={P|P>T}={P|P>max(g)}

[0112] Where Water indicates that the pixel in the image is water, P indicates the pixel whose water index is greater than the threshold T, g indicates the inter-class variance, and T indicates the segmentation threshold of the foreground image and the background image.

[0113] Mountain shadows or building shadows are the biggest interference to water surface extraction. According to the relatively horizontal characteristics of the surface water slope, the water surface usually has a low slope. The slope and aspect information were calculated based on the SRTM DEM data, and 10° was used as the threshold to mask out areas with a slope greater than 10°.

[0114] Remote sensing water body images taken by satellites may be blocked by clouds or incompletely photographed. Incomplete water surface images affected by clouds and detection range are called water surface missing images. Therefore, it is necessary to reconstruct water surface missing images.

[0115] Optionally, reconstructing the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image, including:

[0116] Masking a water body occurrence rate image using the water surface missing image to obtain a water body occurrence rate histogram of the water surface missing image;

[0117] Determine a water body occurrence rate threshold value according to a water body occurrence rate histogram of the water surface missing image;

[0118] Water body reconstruction processing is performed on pixels in the water surface missing image that are greater than the water body occurrence rate threshold to obtain a reconstructed time-series binary water body image.

[0119] Specifically, the water body occurrence rate refers to the probability that the actual situation corresponding to the pixel in the image is a water body.

[0120] Water surface restoration based on water body occurrence has the following assumption: if all non-missing pixels with the same water body occurrence rate α are classified as water, then all missing pixels with a water body occurrence rate greater than or equal to α should also be covered by water. Missing pixels refer to pixels that are affected by clouds or detection range and appear as data gaps in the image.

[0121] Optionally, the water body incidence image is obtained by the following steps:

[0122] Water body extraction is performed on multi-source remote sensing water body images over many years to obtain time series binary water body images;

[0123] The time series binary water body images are superimposed to obtain the water body occurrence rate image.

[0124] Specifically, the water surface data are screened, and the screening threshold is determined according to the land area included in the study area. Then, the multi-source remote sensing water body images that are less affected by the loss of water surface are screened out according to the screening threshold.

[0125] Water bodies are extracted from multiple images of multi-source remote sensing water body images by calculating water body index, edge detection, morphological dilation operation, threshold segmentation and slope correction to obtain time series binary water body images.

[0126] Multiple time-series binary water body images are superimposed, for example, 100 images, and the number of times each pixel appears as a water body in the multiple time-series binary water body images and the total number of times the pixel appears are recorded.

[0127] Calculate the ratio of the number of times a pixel appears as water to the total number of times the pixel appears, and then normalize the ratio to obtain the water body occurrence rate of the pixel.

[0128] The ratio is normalized to make the water body occurrence rate distributed between 0 and 1, thereby improving the sensitivity of the water body occurrence rate.

[0129] Furthermore, the water body occurrence rate of each pixel in the image is obtained, thereby obtaining a water body occurrence rate image.

[0130] The calculation method of water body occurrence rate was clarified, laying the foundation for the subsequent reconstruction of missing water surface images based on water body occurrence rate.

[0131] After obtaining the water body incidence image, the water body incidence image is masked using the water surface missing image, the water body incidence of the pixels containing water in the water surface missing image is retained, and the water body incidence of the masked pixels is counted to obtain a water body incidence histogram of the water surface missing image.

[0132] Figure 2 is a schematic diagram of determining the water body occurrence rate threshold provided by an embodiment of the present invention, such as Figure 2 As shown, the horizontal axis is the water body occurrence value, and the vertical axis is the frequency of the water body occurrence value.

[0133] First, calculate the ratio of the total number of pixels in the water surface missing image to the number of histogram columns in the water body incidence histogram of the water surface missing image, and then multiply the ratio by the empirical value of 0.17 as the counting threshold.

[0134] The water body occurrence rate values ​​in the water body occurrence rate histogram are traversed, and the minimum water body occurrence rate value with an occurrence frequency greater than the counting threshold is selected as the water body occurrence rate threshold.

[0135] Under the assumption of water surface restoration based on water body occurrence rate, pixels whose values ​​are greater than the water body occurrence rate threshold in the image to be restored and reconstructed are selected for water body reconstruction processing, and pixels whose values ​​are greater than the water body occurrence rate threshold are confirmed as water bodies, thereby obtaining a reconstructed time series binary water body image.

[0136] By determining the water body occurrence rate threshold and reconstructing water bodies for pixels whose water body occurrence rate is greater than the water body occurrence rate threshold, it is possible to reconstruct the water surface in the image where the water surface is missing when the remote sensing image is affected by clouds and detection range.

[0137] Optionally, reconstructing the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image, including:

[0138] Determine the ratio of a first water area in the water surface missing image to a second water area in the water surface missing image; the first water area is the water area between the water body boundary and the lower depth contour of the adjacent depth contour in the non-missing area of ​​the water surface missing image; the second water area is the water area between the minimum neighboring depth contour areas in the non-missing area of ​​the water surface missing image;

[0139] Determine the position of boundary points between adjacent depth contours according to the ratio;

[0140] The boundary point positions are connected to obtain the reconstructed time-series binary water body image.

[0141] Specifically, using the water body incidence reconstruction method for multi-source remote sensing water body images with too much water surface missing or uneven spatial distribution will seriously interfere with the determination of the water body incidence threshold. Therefore, such multi-source remote sensing water body images are not suitable for reconstruction using water body incidence.

[0142] However, the images in this type of multi-source remote sensing water body images may still contain information about the land-water boundary, suggesting the elevation of the land-water boundary.

[0143] The water body boundary is equivalent to the isobath. If the water-land boundary of the missing part of the water surface is between the lower isobath and the higher isobath, the complete water area can be estimated as the water area between the minimum neighboring isobath areas.

[0144] A straight line is drawn from the center point of the water surface along different angles to the adjacent isobaths of the non-water surface missing area in the water surface missing image, and intersects with the higher isobath among the adjacent isobaths at point A, intersects with the lower isobath among the adjacent isobaths at point B, and intersects with the water body boundary at point C.

[0145] In the water surface missing image, the first water area can be estimated as the water area between the water body boundary in the non-missing area and the lower isobath in the adjacent isobaths, and the second water area can be estimated as the water area between the minimum neighboring isobath areas in the non-missing area. The first water area and the second water area in the water surface missing image are in proportion k, and the proportion k is equal to the ratio of the length of line segment BC to the length of line segment AB.

[0146] The expression for the ratio k is as follows:

[0147]

[0148] Where k represents the ratio of the first water area in the water surface missing image to the second water area in the water surface missing image, l BC represents the length of line segment BC, l AB Represents the length of line segment AB.

[0149] The ratio k can be used to determine the position of the land-water boundary between points AB. The coordinates of the water surface reconstruction point C′ can be calculated as follows:

[0150] x C′ =x B +kl AB cosθ

[0151] y C′ =y B +kl AB sinθ

[0152] In the formula, x C′ and C′ Respectively represent the horizontal and vertical coordinates of the water surface reconstruction point C′; B and B represents the horizontal and vertical coordinates of point B respectively; k represents the ratio of the first water area in the water surface missing image to the second water area in the water surface missing image; l AB represents the length of line segment AB; θ represents the angle between the straight line drawn along the center point of the water surface and the horizontal axis.

[0153] Traverse all angles contained in the missing part of the water surface to obtain the positions of all water-land boundary points, and connect all boundary points to reconstruct the water boundary of the missing area of ​​the water surface.

[0154] The water-land boundary point is determined by the ratio of the first water area to the second water area between adjacent isobaths, which enables the reconstruction of the water surface in the image with missing water surface when the remote sensing image is affected by clouds and detection range.

[0155] Optionally, reconstructing the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image, including:

[0156] Reconstruct the missing water surface image based on the water body occurrence rate to obtain a reconstructed first time-series binary water body image;

[0157] The reconstructed first time-series binary water body image is corrected based on the water surface isobath to obtain a reconstructed second time-series binary water body image.

[0158] Specifically, when fewer available pixels appear in the middle of the water body and do not contain water-land boundary information, it is necessary to comprehensively consider the reconstruction method of water body incidence and the reconstruction method of water surface isobaths. First, use the reconstruction method based on water body incidence to restore the approximate boundary of the water body, and then use the reconstruction method based on water surface isobaths to correct the water body boundary using adjacent isobaths.

[0159] By combining the reconstruction method of water body incidence with the reconstruction method of water surface isobaths, it is possible to reconstruct the missing water surface image when the remote sensing image is affected by clouds and detection range.

[0160] Step 102: Acquire a time-series river width dataset based on the reconstructed time-series binary water body image.

[0161] Specifically, the reconstructed time-series binary water body image is processed to obtain the river centerline image, and then the river centerline image is processed to obtain the normal direction of the river centerline. Finally, the intersection of the normal direction of the river centerline and the river boundary is used to obtain the time-series river width dataset.

[0162] Optionally, based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained, including:

[0163] Performing distance transformation, image gradient and morphological skeletonization processing on the reconstructed time-series binary water body image to obtain a river channel centerline image;

[0164] The time series river width dataset is obtained based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary.

[0165] Specifically, the distance transformation is to transform the pixel value of each river pixel in the reconstructed time-series binary water body image into the distance between the pixel and the non-river pixel closest to the pixel, thereby obtaining a distance transformation image.

[0166] The pixel value at the center of the river has a local maximum, while the gradient has a minimum. The image gradient is calculated by convolving the distance transformed image with a pair of 3×3 pixel kernels. The pair of 3×3 pixel kernels is shown below:

[0167]

[0168] The pixels with gradient values ​​less than the threshold are taken as the river centerline. A lower threshold will increase the algorithm speed, while a higher threshold will enhance the connectivity of the centerline. Therefore, in order to balance the algorithm speed and the connectivity of the centerline, the threshold is set to 0.9 based on experience. Connect the pixels with gradient values ​​less than 0.9 to obtain the initial river centerline.

[0169] In the direction perpendicular to the initial river centerline, there is a situation where the gradient value of more than one pixel is less than 0.9. In order to ensure that the river centerline is only one pixel wide, the initial river centerline is skeletonized or refined three times to obtain the river centerline image.

[0170] Optionally, based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary, acquiring a time series river width dataset includes:

[0171] Convolving the river centerline image with a convolution kernel to obtain a convolution image;

[0172] Determining the normal direction of the centerline of the river channel based on the number of intersections between the river centerline image and the edge of the convolution kernel and the convolution value in the convolution image;

[0173] The time series river width dataset is obtained based on the intersection of the normal line of the river centerline along the normal line direction and the river boundary.

[0174] Specifically, the river centerline image is convolved with a 9×9 convolution kernel to obtain a convolution image. The 9×9 convolution kernel is as follows:

[0175]

[0176] The convolution kernel starts from the rightmost center position moving counterclockwise, and the outer ring value of the convolution kernel is the angle between the horizontal line and the line connecting this value and the center point of the kernel. Except for the outer ring value of the convolution kernel, the other values ​​of the convolution kernel are zero.

[0177] When the pixel of the center line of the river channel has two intersection points with the edge of the convolution kernel, the normal direction of the local river channel center point is the direction of the convolution value divided by 2.

[0178] When the pixel of the center line of the river has an intersection with the edge of the convolution kernel, the intersection is the pixel near the end of the center line of the river, and the normal direction of the local center point of the river is the direction of the convolution value plus 90°.

[0179] When there are more than two intersection points between the pixel of the center line of the river and the edge of the convolution kernel, which may occur near the confluence of the river, these intersection points are discarded and do not participate in the calculation of the normal direction.

[0180] For example, when the center line of the river channel intersects the edge of the convolution kernel at 90° and 270°, the normal direction of the local river channel center point is (270°+90°) / 2, that is, 180°.

[0181] For example, when the center line of the river channel only intersects the edge of the convolution kernel at 45°, the normal direction of the local river channel center point is 45°+90°, that is, 135°.

[0182] For example, when the center line of the river intersects the edge of the convolution kernel at 45°, 225°, and 270°, the intersection points are discarded and do not participate in the calculation.

[0183] Figure 3 is a schematic diagram of determining the river width provided by an embodiment of the present invention, such as Figure 3As shown in the figure, after determining the normal direction, a straight line is drawn along the normal direction of the local river centerline (perpendicular to the direction of the local river centerline) to intersect the river boundary at one point. The line segment between the intersection point and the river center point is Wm. The length of the line segment Wm is equal to 3 times the distance value Wb of the distance transformation of the river center point. In GEE, a mean reducer operation is performed on Wm to calculate the ratio of the intersection distance between the foreground of the binary water body image and Wm to the length of Wm, and then multiplied by the length of Wm to obtain the distance between Wm and the foreground of the binary water body image, thereby obtaining the river width.

[0184] Furthermore, the river width calculation method mentioned above is used to obtain the river width corresponding to each binary water surface image, and a time series river width dataset is constructed.

[0185] By first obtaining the centerline of the river channel, then using the centerline of the river channel to determine the normal direction of the centerline of the river channel, and finally using the normal direction of the centerline of the river channel to determine the river width, the time-series river width dataset can be automatically constructed without manual labeling, which further improves the automation of the river flow monitoring program and is also conducive to the subsequent construction of the relationship curve between river width and river flow.

[0186] After obtaining the river centerline image, the DEM data is sampled to obtain the elevation value of each center point pixel corresponding to the position in the river centerline image.

[0187] Figure 4 Schematic diagram of determining hydraulic gradient provided by an embodiment of the present invention, such as Figure 4 As shown in the figure, the horizontal axis is the river distance and the vertical axis is the elevation. Hydraulic gradient refers to the ratio of the water level difference between any two points to the distance between the two points.

[0188] A starting point of the center line of the river channel is preset, and the center points are sorted along the river direction from the starting point. The next center point is found and recorded according to the closest spatial distance. After traversing all the center points, the distances of all the center points from the starting point and the elevations of the center points are visualized and fitted into a first-order polynomial. Figure 4 The straight line in the figure is a fitted first-order polynomial, and the slope of the straight line is the hydraulic gradient of the river section.

[0189] Step 103, simulating the underwater terrain based on the elevation loop iteration to obtain a relationship curve between the river width and the river flow.

[0190] Specifically, before simulating the underwater terrain, the main direction of the river channel needs to be determined.

[0191] Optionally, determine the main direction of the river channel, including:

[0192] transforming the river channel edge image from the spatial domain to the frequency domain;

[0193] In the frequency domain, the origin of the river channel edge image is used as the starting point to divide the sectors and determine the energy of each sector;

[0194] The direction corresponding to the sector with the largest energy is determined as the main direction of the river channel.

[0195] Specifically, the morphological structure of the river channel is complex and changeable. In order to fully automate the preprocessing of river section terrain data, the first issue is to determine the direction of the cross-section and longitudinal section of the river channel, that is, to determine the main direction of the river channel.

[0196] Binarization, low-pass filtering, edge detection, and fast Fourier transform are performed on multi-source remote sensing images to obtain edge images, which are then transformed from the spatial domain to the frequency domain.

[0197] The geometric contour of the river channel edge is very characteristic, roughly composed of two straight lines intersecting at similar angles. It can be seen that such a contour has a clear gradient direction of pixel values.

[0198] In the frequency domain, sector division is performed with the origin of the edge image as the starting point. For example, the division unit is 1°, and the total energy of the area contained in each small sector is calculated, that is, the energy contained at this angle, and the direction (angular direction) corresponding to the sector with the largest energy is determined as the main direction of the river.

[0199] By first transforming the edge image from the spatial domain to the frequency domain, and then selecting the direction corresponding to the sector with the largest energy in the frequency domain as the main direction of the river channel, the main direction of the river channel can be determined fully automatically without human intervention, further improving the automation of the river flow monitoring solution and facilitating the subsequent simulation of underwater terrain.

[0200] Optionally, before determining the deepest point of the river, include:

[0201] Determine the main direction of the river;

[0202] Rotating the river channel edge image according to the main direction of the river channel until the cross section of the river channel edge image is horizontal;

[0203] Averaging the elevation values ​​in the river channel edge image along a direction perpendicular to the cross section line to obtain a river channel cross section;

[0204] The water surface of the river cross section is masked to obtain the underwater terrain area to be simulated.

[0205] Specifically, Figure 5 Schematic diagram of river section topographic data preprocessing provided by an embodiment of the present invention. Figure 5As shown, there are river banks on both sides and water in the middle. The river cross section is rotated and fitted so that the river cross section coincides with the x-axis, and then the elevation values ​​perpendicular to the river cross section are averaged.

[0206] After the main direction of the river channel is determined, the edge image is rotated according to the main direction of the river channel until the cross section of the edge image is approximately parallel to the x-axis.

[0207] Since the cross-section of the river is not a regular rectangle, there will still be a small angle between the rotated cross-section and the x-axis. In order to facilitate the subsequent mean processing of the river longitudinal section, the rotated cross-section needs to be precisely rotated.

[0208] The precise rotation takes the minimum circumscribed rectangle of the cross section as the operation object, calculates the angle between the vectors of each corner point of the circumscribed rectangle and the x-axis, and rotates the minimum circumscribed rectangle to make the cross section completely parallel to the x-axis.

[0209] The minimum circumscribed rectangle of the cross section first realizes the simple circumscribed rectangle of the polygon. The simple circumscribed rectangle refers to the circumscribed rectangle whose sides are parallel to the x-axis or y-axis.

[0210] When a polygon's circumscribed rectangle has a side that is collinear with the side of the original polygon, first detect the direction with the same angle as the polygon's side length, find the simple circumscribed rectangle of the polygon after rotating it by different degrees, record the area and vertex coordinates of the simple circumscribed rectangle corresponding to each rotation angle, then select the simple circumscribed rectangle with the smallest area as the target simple circumscribed rectangle, and obtain the vertex coordinates of the target simple circumscribed rectangle. Finally, calculate the length of each side of the target simple circumscribed rectangle and the angle with the x-axis based on the vertex coordinates of the circumscribed rectangle.

[0211] According to the general shape of the river section, the minimum angle corresponding to the maximum side length is selected as the angle of precise rotation, and the cross-section of the river section is rotated to almost coincide with the x-axis.

[0212] After rotating the cross section of the river channel to nearly coincide with the x-axis, the elevation values ​​of the edge image along the direction perpendicular to the cross section (i.e., the y-axis direction) are counted, and the elevation values ​​along the y-axis direction are averaged to obtain the cross section of the river channel;

[0213] The water surface of the river cross section is masked to obtain the area of ​​the river cross section that needs underwater terrain simulation.

[0214] The cross-section of the river channel edge image is first rotated to horizontal, and then the elevation values ​​in the river channel edge image along the direction perpendicular to the cross-section are averaged to obtain the river channel cross-section. Finally, the water surface of the river channel cross-section is masked to obtain the underwater terrain area to be simulated. This realizes the automatic determination of the underwater terrain area to be simulated without human intervention, further improving the automation of the river flow monitoring program and facilitating the subsequent simulation of the underwater terrain.

[0215] Optionally, before determining the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels, the following steps are included:

[0216] Determine the deepest point of the river;

[0217] Taking the deepest point of the river channel as the boundary, the underwater terrain on the corresponding side is simulated according to the elevation and slope information of the above-water pixels.

[0218] Specifically, after determining the underwater terrain area to be simulated, it is necessary to determine the deepest point of the river.

[0219] Assuming that the deepest point of the river is the result of the combined effect of the river banks on both sides, find the local extreme values ​​on both sides of the river banks, consider the river bank area between the two extreme points closest to the water surface as the two river banks that affect the deepest point, and fit the two river banks.

[0220] Figure 6 is a schematic diagram for determining the deepest point of a river provided by an embodiment of the present invention, such as Figure 6 As shown, the horizontal axis is the width of the river channel, and the vertical and horizontal axes are the elevation values. Figure 6 The points marked by the inverted triangle are the extreme points of the river banks on both sides. The river banks on both sides are fitted according to the extreme points of the river banks on both sides. Figure 6 The line segments in the boxes on the left and right sides are the areas on both sides of the river that need to be fitted.

[0221] The ratio of the distances from the deepest point of the river channel to both sides of the river bank is equal to the ratio of the inverse of the slopes of the fitting line segments on both sides of the river bank. Figure 6 The circle in the figure is the deepest point of the river.

[0222] After determining the deepest point of the river channel, the underwater terrain on the left side of the deepest point of the river channel is simulated based on the elevation and slope information of the above-water pixels on the left side of the river channel, and the underwater terrain on the right side of the deepest point of the river channel is simulated based on the elevation and slope information of the above-water pixels on the right side of the river channel.

[0223] By first determining the deepest point of the river channel and then simulating the underwater terrain on the corresponding side using the deepest point of the river channel as the boundary, the underwater terrain can be zoned for simulation, which is beneficial to improving the accuracy of underwater terrain simulation.

[0224] Optionally, the underwater terrain is simulated based on the elevation cycle iteration to obtain a relationship curve between the river width and the river flow, including:

[0225] Determine the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels; the adjacent pixels are pixels adjacent to the pixel to be simulated;

[0226] Based on the elevation of the pixel to be simulated, the underwater terrain is obtained;

[0227] Based on the underwater topography, river width and hydraulic gradient, the relationship curve between river width and river flow is obtained.

[0228] Specifically, first determine the pixel boundary adjacent to the boundary to be simulated, and find a plurality of pixels with higher pixel values ​​than the pixel to be simulated at the pixel boundary, for example, the number of the plurality of pixels is ten.

[0229] According to the elevation values ​​of multiple pixels adjacent to the pixel to be simulated and the horizontal distance of the ten pixels, multiple pixels are simulated to obtain the slopes of the simulated line segments of the multiple pixels.

[0230] The slope of the adjacent pixel is determined based on the slope of the simulated line segment of multiple pixels and the simulated function. The expression of the slope of the adjacent pixel is as follows:

[0231] k(x)=k0×f(x)

[0232] Where k(x) is the slope of the adjacent pixel, x is the ratio of the horizontal distance from the pixel to be simulated to the deepest point to the horizontal distance from the land-water boundary to the deepest point, k0 is the slope of the simulated line segment of multiple pixels at the pixel boundary, and f(x) is a simulation function, such as a trigonometric function.

[0233] After determining the slope of the adjacent pixel, the elevation of the pixel to be simulated is determined based on the elevation, slope and pixel size of the adjacent pixel. The expression of the elevation of the pixel to be simulated is as follows:

[0234] Z(i)=h1 i -k1 i ×cellsize

[0235] In the formula, Z(i) represents the elevation value of pixel i to be simulated, h1 i Indicates the elevation value of the first pixel adjacent to the pixel to be simulated, k1 i It represents the slope of the first pixel adjacent to the pixel to be simulated, and cellsize represents the pixel size of the adjacent pixels adjacent to the pixel to be simulated.

[0236] Furthermore, the elevation values ​​of all pixels to be simulated are calculated, and the elevation decrement step parameter of each iteration is set to one pixel, so that the iterative calculation continues until the number of pixels to be simulated is zero.

[0237] If the underwater terrain simulation based on the two riverbanks does not overlap at the deepest point, the slope value is optimized, such as changing the simulation function, and using the equal elevation of the deepest point as a constraint condition for the next iteration.

[0238] After completing the underwater terrain simulation, the intersection of the current water level or the elevation value corresponding to the water surface and the above-water and underwater terrain data is taken as the origin, the horizontal direction is set as the x-axis, and the vertical direction is set as the y-axis to calculate the river cross-sectional area and wetted perimeter.

[0239] There are two methods to calculate the cross-sectional area and wetted perimeter. Method 1: Integrate the function expression: Calculate the high-order polynomial of the underwater terrain, so that the function shape is as close to the simulation shape as possible. The cross-sectional area can be obtained by taking the definite integral of the high-order polynomial, and the wetted perimeter can be expressed as the definite integral form of the derivative expression of the high-order polynomial.

[0240]

[0241] Where L represents the wetted perimeter, a and b represent the range of the definite integral, x represents the distance on the x-axis, and f′(x) represents the derivative expression of the high-order polynomial of the underwater topography.

[0242] Method 2: Integrate discrete values: Directly use the elevation value of underwater simulation. The cross-sectional area can be calculated using the trapezoidal rule to obtain the area enclosed by the discrete points and the x-axis. The wetted perimeter is calculated by accumulating the distance between the coordinates of two adjacent points.

[0243] After obtaining the cross-sectional area and wetted perimeter, the roughness coefficient needs to be obtained. The roughness coefficient depends on factors such as vegetation type and structure, cross-sectional area, meander degree, and river obstacles.

[0244] The roughness coefficient n is intended to be calculated with reference to the commonly used river roughness calculation formula through the river morphological parameters and landform types within the section.

[0245] The expression of roughness coefficient n is as follows:

[0246] n=(n 0 +n 1 +n 2 +n 3 +n 4 )m 5

[0247] Where n is the roughness coefficient, n 0 is the basic roughness coefficient of a naturally straight, smooth and uniform channel; n 1 is the coefficient to consider the influence of water surface irregularity; n 2 is the influence coefficient of the cross-sectional shape and size change of the river channel; n 3 is the influence coefficient of water barrier; n 4 is the influence coefficient of the plant; m 5 is the influence coefficient of river channel meanders.

[0248] Table 1 is a table of natural river flood roughness coefficients. As shown in Table 1, the values ​​of various parameters can refer to the values ​​in the table of natural river flood roughness coefficients inverted and summarized based on a large amount of measured data of various rivers.

[0249] Table 1 is the roughness coefficient table of natural river flood

[0250]

[0251]

[0252] After obtaining the hydraulic gradient, river cross-sectional area, wetted perimeter and roughness coefficient, the average flow velocity of the river can be calculated according to the Manning formula. The expression of the average flow velocity is as follows:

[0253]

[0254] In the formula, represents the average flow velocity of the river, k st is the roughness coefficient, R is the hydraulic radius, which can be obtained by dividing the river cross-sectional area by the wetted perimeter, and I is the hydraulic gradient.

[0255] The river flow can be determined based on the average flow velocity and cross-sectional area of ​​the river. The expression of river flow is as follows:

[0256]

[0257] Where Q represents the river flow, represents the average flow velocity of the river, and A represents the cross-sectional area of ​​the river.

[0258] In order to minimize the consumption of computing resources, we choose to construct a curve of the relationship between river width and river flow. According to the intersection of different water level planes and the river cross-section terrain, the distance between the two intersections is the river width, and the cross-sectional area and wetted perimeter under the water level are calculated, and the relationship curve between river width and river flow is obtained by substituting them into the above river flow estimation formula.

[0259] By simulating the elevation of the simulated pixel and then further simulating the underwater terrain, the river cross-sectional area and wetted perimeter are obtained based on the underwater terrain, and finally the relationship curve between river width and river flow is obtained based on the obtained river channel parameter information. The relationship curve between river width and river flow is automatically obtained, which is conducive to the subsequent construction of a time-series river flow data set and further improves the automation of river flow monitoring solutions.

[0260] Step 104: acquiring a time-series river flow data set based on the time-series river width data set and the relationship curve between the river width and the river flow.

[0261] Specifically, based on the relationship curve between river width and river flow, the river width of a certain time phase in the time series river width dataset is matched, and the river flow corresponding to the river width closest to the curve is selected as the river flow value at that time. All time series river width data are traversed to obtain a long time series river flow dataset.

[0262] The time-series river flow data set is constructed through the time-series river width data set and the relationship curve between river width and river flow, which further improves the automation of the river flow monitoring program.

[0263] The river flow remote sensing monitoring method provided by the present invention reconstructs the water surface to obtain river surface distribution data without manual interpretation, simulates the underwater terrain to obtain river cross-section terrain data without actual measurement of the terrain, and constructs an automated river flow monitoring solution.

[0264] Figure 7 FIG. 1 is a schematic diagram of the structure of a river flow remote sensing monitoring device provided by an embodiment of the present invention. Figure 7 As shown, the present invention also provides a river flow remote sensing monitoring device, including: a first acquisition module 701, a second acquisition module 702, a third acquisition module 703 and a fourth acquisition module 704, wherein:

[0265] The first acquisition module 701 is used to reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath, and obtain the reconstructed time-series binary water body image;

[0266] The second acquisition module 702 is used to acquire a time-series river width dataset based on the reconstructed time-series binary water body image;

[0267] The third acquisition module 703 is used to simulate the underwater terrain based on the elevation cycle iteration to obtain the relationship curve between the river width and the river flow;

[0268] The fourth acquisition module 704 is used to acquire a time-series river flow data set based on the time-series river width data set and the relationship curve between the river width and the river flow.

[0269] Specifically, the river flow monitoring device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0270] Figure 8 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 8As shown, the electronic device may include: a processor 810, a communication interface 920, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute a river flow remote sensing monitoring method, the method comprising: reconstructing a missing water surface image based on a water body occurrence rate and / or a water surface isobath to obtain a reconstructed time-series binary water body image; obtaining a time-series river width data set based on the reconstructed time-series binary water body image; simulating underwater terrain based on elevation loop iteration to obtain a relationship curve between river width and river flow; obtaining a time-series river flow data set based on the time-series river width data set and the relationship curve between river width and river flow.

[0271] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0272] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the river flow remote sensing monitoring method provided by the above-mentioned methods, and the method includes: reconstructing the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image; obtaining a time-series river width data set based on the reconstructed time-series binary water body image; simulating the underwater terrain based on the elevation cycle iteration to obtain a relationship curve between the river width and the river flow; obtaining a time-series river flow data set based on the time-series river width data set and the relationship curve between the river width and the river flow.

[0273] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the above-mentioned river flow remote sensing monitoring methods, the methods comprising: reconstructing missing water surface images based on water body occurrence rate and / or water surface depth contours to obtain reconstructed time-series binary water body images; obtaining a time-series river width data set based on the reconstructed time-series binary water body images; simulating underwater terrain based on elevation cyclic iteration to obtain a relationship curve between river width and river flow; obtaining a time-series river flow data set based on the time-series river width data set and the relationship curve between river width and river flow.

[0274] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0275] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0276] In the embodiments of the present application, the terms "first", "second", etc. are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0277] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A river flow remote sensing monitoring method, characterized in that: include: Reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time-series binary water body image; Based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained; The underwater terrain is simulated based on elevation cycle iteration to obtain the relationship curve between river width and river flow; Based on the time-series river width dataset and the relationship curve between the river width and the river flow, obtaining a time-series river flow dataset; The underwater terrain is simulated based on the elevation cycle iteration to obtain the relationship curve between river width and river flow, including: Determine the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels; the adjacent pixels are pixels adjacent to the pixel to be simulated; Based on the elevation of the pixel to be simulated, the underwater terrain is obtained; Based on underwater topography, river width and hydraulic gradient, obtain the relationship curve between river width and river flow; Based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained, including: Performing distance transformation, image gradient and morphological skeletonization processing on the reconstructed time-series binary water body image to obtain a river channel centerline image; Acquire the time series river width dataset based on the intersection of the normal line of the river channel centerline and the river channel boundary in the river channel centerline image; Based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary, a time series river width dataset is obtained, including: Convolving the river centerline image with a convolution kernel to obtain a convolution image; Determining the normal direction of the centerline of the river channel based on the number of intersections between the centerline image of the river channel and the edge of the convolution kernel and the convolution value in the convolution image; The time series river width dataset is obtained based on the intersection of the normal line of the river centerline along the normal line direction and the river boundary.

2. The river flow remote sensing monitoring method according to claim 1, characterized in that: Reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time series binary water body image, including: Masking a water body occurrence rate image using the water surface missing image to obtain a water body occurrence rate histogram of the water surface missing image; Determine a water body occurrence rate threshold value according to a water body occurrence rate histogram of the water surface missing image; Water body reconstruction processing is performed on pixels in the water surface missing image that are greater than the water body occurrence rate threshold to obtain a reconstructed time-series binary water body image.

3. The river flow remote sensing monitoring method according to claim 1, characterized in that: Reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time series binary water body image, including: Determine the ratio of a first water area in the water surface missing image to a second water area in the water surface missing image; the first water area is the water area between the water body boundary and the lower depth contour of the adjacent depth contour in the non-missing area of ​​the water surface missing image; the second water area is the water area between the minimum neighboring depth contour areas in the non-missing area of ​​the water surface missing image; Determine the position of boundary points between adjacent depth contours according to the ratio; The boundary point positions are connected to obtain the reconstructed time-series binary water body image.

4. The river flow remote sensing monitoring method according to claim 1, characterized in that: Reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath to obtain a reconstructed time series binary water body image, including: Reconstruct the missing water surface image based on the water body occurrence rate to obtain a reconstructed first time-series binary water body image; The reconstructed first time-series binary water body image is corrected based on the water surface isobath to obtain a reconstructed second time-series binary water body image.

5. The river flow remote sensing monitoring method according to claim 1, characterized in that: Based on the elevation, slope and pixel size of the adjacent pixels, the elevation of the pixel to be simulated is determined, including: Determine the deepest point of the river; Taking the deepest point of the river channel as the boundary, the underwater terrain on the corresponding side is simulated according to the elevation and slope information of the above-water pixels.

6. The river flow remote sensing monitoring method according to claim 5, characterized in that: Before determining the deepest point of a river, include: Determine the main direction of the river; Rotating the river channel edge image according to the main direction of the river channel until the cross section of the river channel edge image is horizontal; Averaging the elevation values ​​in the river channel edge image along a direction perpendicular to the cross section line to obtain a river channel cross section; The water surface of the river cross section is masked to obtain the underwater terrain area to be simulated.

7. The river flow remote sensing monitoring method according to claim 6, characterized in that: Determine the main direction of the river, including: transforming the river channel edge image from the spatial domain to the frequency domain; In the frequency domain, the origin of the river channel edge image is used as the starting point to divide the sectors and determine the energy of each sector; The direction corresponding to the sector with the largest energy is determined as the main direction of the river channel.

8. The river flow remote sensing monitoring method according to claim 2 is characterized in that: The water body incidence image is obtained by the following steps: Water body extraction is performed on multi-source remote sensing water body images over many years to obtain time series binary water body images; The time series binary water body images are superimposed to obtain the water body occurrence rate image.

9. A river flow remote sensing monitoring device, characterized in that: include: A first acquisition module is used to reconstruct the missing water surface image based on the water body occurrence rate and / or the water surface isobath, and obtain a reconstructed time-series binary water body image; A second acquisition module is used to acquire a time-series river width dataset based on the reconstructed time-series binary water body image; The third acquisition module is used to simulate the underwater terrain based on the elevation cycle iteration to obtain the relationship curve between the river width and the river flow; A fourth acquisition module, configured to acquire a time-series river flow data set based on the time-series river width data set and a relationship curve between the river width and the river flow; The underwater terrain is simulated based on the elevation cycle iteration to obtain the relationship curve between river width and river flow, including: Determine the elevation of the pixel to be simulated based on the elevation, slope and pixel size of the adjacent pixels; the adjacent pixels are pixels adjacent to the pixel to be simulated; Based on the elevation of the pixel to be simulated, the underwater terrain is obtained; Based on underwater topography, river width and hydraulic gradient, obtain the relationship curve between river width and river flow; Based on the reconstructed time-series binary water body image, a time-series river width dataset is obtained, including: Performing distance transformation, image gradient and morphological skeletonization processing on the reconstructed time-series binary water body image to obtain a river channel centerline image; Acquire the time series river width dataset based on the intersection of the normal line of the river channel centerline and the river channel boundary in the river channel centerline image; Based on the intersection of the normal line of the river channel centerline in the river channel centerline image and the river channel boundary, a time series river width dataset is obtained, including: Convolving the river centerline image with a convolution kernel to obtain a convolution image; Determining the normal direction of the centerline of the river channel based on the number of intersections between the centerline image of the river channel and the edge of the convolution kernel and the convolution value in the convolution image; The time series river width dataset is obtained based on the intersection of the normal line of the river centerline along the normal line direction and the river boundary.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the river flow remote sensing monitoring method according to any one of claims 1 to 8 are implemented.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the river flow remote sensing monitoring method according to any one of claims 1 to 8 are implemented.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the river flow remote sensing monitoring method according to any one of claims 1 to 8 are implemented.

Images