Water extraction method, device, equipment, storage medium and program product

Abstract

The present invention relates to the field of image processing technology, and in particular to a water body extraction method, device, equipment, storage medium and program product. The method comprises: obtaining a first water body image of a target area, and obtaining a second water body image of the target area, wherein a first shadow area caused by a non-water body is distributed in at least one first water body unit area, and a second shadow area caused by a non-water body is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different; superimposing the first water body image and the second water body image to obtain the overlapping area of ​​each first water body unit area and the corresponding second water body unit area; and determining the final water body map of the target area according to at least one overlapping area. The present invention is used to solve the defect of low precision of water body extraction results caused by shadows in water body images in the prior art, and realize high-precision and accurate extraction of surface water bodies.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a water body extraction method, device, equipment, storage medium and program product. Background Art

[0002] Surface water refers to the general term for water bodies on the surface of the earth, such as rivers, lakes, ponds and glaciers. It plays an important role in agricultural irrigation, human survival and development, and maintaining the stability of the ecosystem. Therefore, it is particularly important to monitor the dynamics of surface water in a detailed and high-frequency manner. Timely and accurate acquisition of dynamic information on surface water is very important for government departments to formulate policies and adjust measures, and it also has important applications in disaster prevention. With the development of satellite remote sensing technology, the use of remote sensing images for surface water monitoring has become an important method, including optical remote sensing and radar remote sensing.

[0003] In the existing technology, optical remote sensing is affected by clouds, and most images cannot be used due to the large amount of clouds, so it is difficult to achieve continuous dynamic monitoring of water bodies. More importantly, when identifying water bodies, it will be affected by snow, buildings, soil and mountain shadows, making it difficult to accurately portray water bodies. Radar remote sensing has the ability to penetrate clouds and fog, is not affected by climatic conditions and light levels, and can obtain data all day and all weather. However, due to its complex image processing and the characteristics of its side-view imaging, it will cause geometric distortion due to the undulations of the terrain, such as overlapping, perspective contraction, shadows, etc. The backscattering coefficient of the shadow is similar to that of the water body, which also leads to confusion between the shadow and the water body when extracting the water body. When extracting water bodies, how to reduce the impact of shadows has become an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a water body extraction method, device, equipment, storage medium and program product, which are used to solve the defect of low accuracy of water body extraction results caused by shadows in water body images in the prior art, and realize high-precision and accurate extraction of surface water bodies.

[0005] The present invention provides a water body extraction method, comprising: obtaining a first water body image of a target area, and obtaining a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by a non-water body is distributed in the at least one first water body unit area, and a second shadow area caused by the non-water body is distributed in the at least one second water body unit area, and the distribution of the first shadow area and the second shadow area are different; superimposing the first water body image and the second water body image to obtain overlapping areas of each of the first water body unit areas and the corresponding second water body unit areas; and determining a final water body map of the target area according to at least one of the overlapping areas.

[0006] According to a water body extraction method provided by the present invention, the final water body map of the target area is determined based on at least one overlapping area, including: performing the following processing on each overlapping area: calculating a first ratio of the overlapping area to the corresponding first water body unit area, and calculating a second ratio of the overlapping area to the corresponding second water body unit area; when the first ratio is greater than a first threshold value, and the second ratio is greater than a second threshold value, extracting the first water body unit area corresponding to the overlapping area in the first water body image; and determining the final water body map of the target area based on the extracted at least one first water body unit area.

[0007] According to a water body extraction method provided by the present invention, after superimposing the first water body image and the second water body image to obtain the overlapping area between each of the first water body unit areas and the corresponding second water body unit areas, the method further includes: forming a first intermediate result based on each of the overlapping areas; superimposing the first intermediate result with preset river and lake vector data, and removing the overlapping areas belonging to river water bodies in the first intermediate result, wherein the river and lake vector data includes regional distribution data of the river water bodies; and performing the following processing on each of the overlapping areas: calculating a first ratio of the overlapping area to the corresponding first water body unit area, and calculating the first ratio of the overlapping area to the corresponding The method comprises the following steps: calculating a first proportion of the overlapping area to the corresponding first water body unit area, and calculating a second proportion of the overlapping area to the corresponding second water body unit area; when the first proportion is greater than a first threshold value, and the second proportion is greater than a second threshold value, extracting the first water body unit area corresponding to the overlapping area in the first water body image, including: performing the following processing on each overlapping area in the first intermediate result after removing the river water body: calculating a first proportion of the overlapping area to the corresponding first water body unit area, and calculating a second proportion of the overlapping area to the corresponding second water body unit area; when the first proportion is greater than a first threshold value, and the second proportion is greater than a second threshold value, extracting the first water body unit area corresponding to the overlapping area in the first water body image.

[0008] According to a water body extraction method provided by the present invention, the final water body map of the target area is determined based on at least one extracted first water body unit area, including: forming a second intermediate result based on at least one extracted first water body unit area; superimposing the second intermediate result with preset land cover data, and eliminating the first water body unit areas belonging to the preset land type in the second intermediate result to form a third intermediate result, wherein the land cover data includes regional distribution data of the preset land type; and determining the final water body map of the target area based on the third intermediate result.

[0009] According to a water body extraction method provided by the present invention, determining the final water body map of the target area based on the third intermediate result includes: superimposing the third intermediate result and the regional distribution data of the river water body, and using the superimposed result as the final water body map of the target area.

[0010] According to a water body extraction method provided by the present invention, the first water body image is a radar image, and the radar image is formed by collecting water bodies in the target area through radar equipment; the second water body image is an optical image, and the optical image is formed by collecting water bodies in the target area through optical equipment.

[0011] The present invention also provides a water body extraction device, comprising: an acquisition module, used to acquire a first water body image of a target area, and to acquire a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by non-water bodies is distributed in the at least one first water body unit area, and a second shadow area caused by non-water bodies is distributed in the at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different; a superposition module, used to superimpose the first water body image and the second water body image to obtain the overlapping area of ​​each of the first water body unit areas and the corresponding second water body unit areas; a determination module, used to determine the final water body map of the target area based on at least one of the overlapping areas.

[0012] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the water body extraction methods described above are implemented.

[0013] 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 water body extraction methods described above are implemented.

[0014] The water body extraction method, device, equipment, storage medium and program product provided by the present invention obtain a first water body image of a target area, and after obtaining a second water body image of the target area, the first water body image and the second water body image are superimposed to obtain the overlapping area between each first water body unit area and the corresponding second water body unit area, and then determine the final water body map of the target area based on at least one overlapping area.

[0015] In the above process, there is a first shadow area caused by a non-water body in the first water body image. There is a second shadow area caused by a non-water body in the second water body image. The first shadow area is distributed in at least one first water body unit area, and the second shadow area is distributed in at least one second water body unit area. Since the first shadow area and the second shadow area are distributed differently, when the first water body image and the second water body image of the same target area are superimposed, the actual distribution area of ​​the same water body will overlap, that is, there will be overlapping areas, but the shadow areas with different distributions in the two water body images will not overlap. Determining the final water body map based on the overlapping area reduces the impact of removing the shadows in the first water body map and / or the second water body map, and improves the accuracy of water body extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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.

[0017] Figure 1 It is a schematic flow chart of the water extraction method provided by the present invention;

[0018] Figure 2 is an example diagram of overlapping areas provided by the present invention;

[0019] Figure 3 It is a specific implementation example diagram of the water extraction method provided by the present invention;

[0020] Figure 4 This is an example diagram of preliminary extraction results of water bodies from radar images provided by the present invention;

[0021] Figure 5 This is an example diagram of preliminary extraction results of water bodies from optical images provided by the present invention;

[0022] Figure 6 This is an example diagram of the intersection of the preliminary water extraction results of Sentinel-1 and Sentinel-2 provided by the present invention;

[0023] Figure 7 This is an example diagram of the result of eliminating rivers from the intersection provided by the present invention;

[0024] Figure 8 This is an example diagram of the result of Sentinel-1 preliminary water body removal of rivers provided by the present invention;

[0025] Fig. 9 This is an example diagram of the result of Sentinel-2 preliminary water body removal of rivers provided by the present invention;

[0026] Fig.10 This is an example diagram of water body results when percent_s1>37% and percent_s2>40% provided by the present invention;

[0027] Fig.11 This is an example diagram of the result of superimposing the forest land provided by the present invention;

[0028] Fig.12 This is an example diagram of the results of superimposing the forest land and grassland provided by the present invention;

[0029] Fig.13 This is an example diagram of the results of superimposing the forest land, grassland and building provided by the present invention;

[0030] Fig.14 This is an example of a water body where Sentinel-1 provided by the present invention intersects with the result after removing small patches;

[0031] Fig.15 This is an example diagram of adding a river water body to obtain a final water body map provided by the present invention;

[0032] Fig.16 It is a structural schematic diagram of the water extraction device provided by the present invention;

[0033] Fig.17 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0034] 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.

[0035] With the development of satellite remote sensing technology, the use of remote sensing images for surface water monitoring has become an important way, that is, to extract water bodies by using the spectral differences between water bodies and other land objects in different bands. There are mainly threshold methods (single-band threshold and multi-band threshold) and classifier methods (object-oriented method, decision tree method and support vector machine method, etc.). The threshold method is simple to operate but has low accuracy, while the classifier method has high accuracy but is complex to operate. Optical remote sensing is affected by clouds, and most images cannot be used due to large cloud cover, so it is difficult to achieve continuous dynamic monitoring of water bodies; at the same time, water body identification is also affected by snow, buildings, soil and mountain shadows, making it difficult to finely depict water bodies. Radar remote sensing has the ability to penetrate clouds and fog, is not affected by climatic conditions and light levels, and can obtain data all day and all weather. However, due to its complex image processing and the characteristics of side-view imaging, it will cause geometric distortion due to the undulation of the terrain, such as overlapping, perspective contraction, shadows, etc. The backscattering coefficient of the shadow is similar to that of the water body, resulting in confusion between the shadow and the water body when extracting water bodies.

[0036] For example, the geographical location of region Z is southward, located between 20°54′-26°24′ north latitude and 104°26′-112°04′ east longitude, with a total area of ​​about 236,700 km2. It is located in the transition zone from the Yunnan-Guizhou Plateau to the hills of Guangdong and Guangxi. It is a typical karst landform. The overall terrain shows a trend of high northwest and low southeast. It is surrounded by mountains and plateaus, and is basin-shaped. There are many mountains and few plains, which has a unique karst landform. Region Z is dominated by mountains and hills. Mountains and hills are mainly distributed in the west, north and east of region Z. Mountains are mainly medium and low mountains. Mountainous terrain accounts for about 40% of the total area of ​​region Z. Due to the collision and compression of the Indian Ocean Plate and the Eurasian Plate, the mountains in region Z are mostly distributed in arc veins, and the strike direction is mainly northeast southwest and northwest southeast. The minimum elevation of region Z is -134m, the maximum elevation is 2111m, and the terrain is undulating. Optical images are greatly affected by mountain shadows when identifying water bodies, and simple threshold methods cannot separate water bodies and shadows. Synthetic Aperture Radar (SAR) uses side-view imaging. The back of high mountains cannot be illuminated by radar and has no echo signals. It appears as a very dark shadow in SAR images, namely radar shadow. The backscattering coefficient of the shadow is similar to that of the water body, which leads to confusion between the water body and the shadow when extracting the water body, increasing the difficulty of water body extraction. Optical images are easily affected by clouds. There are more cloudy and rainy weather in area Z, and most images are too cloudy to be used.

[0037] Therefore, in the same type of cloudy and rainy mountainous areas, the extraction of surface water bodies is greatly affected by the mountainous areas. How to reduce the impact of shadows has become an urgent problem to be solved. Figure 1-Figure 15 The water extraction method of the present invention is described.

[0038] In one embodiment, if Figure 1 As shown, the implementation process of the water extraction method is as follows:

[0039] Step 101, obtaining a first water body image of a target area, and obtaining a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by a non-water body is distributed in at least one first water body unit area, a second shadow area caused by a non-water body is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different.

[0040] In this embodiment, the target area refers to the area where water body information needs to be extracted. For example, the target area can refer to the area Z mentioned above. When extracting water body information of the target area, it is necessary to implement it through the water body image of the target area. However, due to the influence of non-water bodies, a shadow area will appear in the water body image. The shadow area is confused with the actual area of ​​the water body, which will cause the various water body unit areas presented in the water body image to be inaccurate. For example, a water body unit area is actually the shadow of a mountain in the actual terrain, not an actual water body. For another example, a part of a water body unit area is an actual water body in the actual terrain, while the rest is actually the shadow of grass.

[0041] A target area can obtain different water body images through various means, such as radar acquisition, optical equipment acquisition, infrared equipment acquisition, and any other means capable of acquiring water body distribution information, and the first water body image and the second water body image can be images acquired by any two of the above means. Different acquisition means are affected differently by external factors, resulting in different distributions of shadow areas in the acquired water body images.

[0042] In this embodiment, by using the first water body image and the second water body image with different shadow area distributions, the method provided by the present invention is adopted to better avoid the influence of shadows and achieve fine extraction of water bodies in the target area.

[0043] In one embodiment, the first water body image is a radar image, which is formed by collecting water in a target area through radar equipment; the second water body image is an optical image, which is formed by collecting water in a target area through optical equipment.

[0044] More specifically, the first water body image is a radar image obtained by observing the Sentinel-1 satellite. Sentinel-1 is an active microwave remote sensing satellite equipped with a C-band synthetic aperture radar sensor, which can observe the ground and perform radar imaging around the clock and in all weather conditions. The Sentinel-1 satellite is an earth observation satellite of the Copernicus program of the European Space Agency (ESA). It consists of two satellites, A and B. The revisit period of a single satellite is 12 days, and the two satellites complement each other with a revisit period of 6 days. Sentinel-1 has a total of four imaging modes: Strip Map Mode (SM), Interferometric Wide Swath (IW), Extra Wide Swath (EW), and Wave Mode (WV).

[0045] In this embodiment, the Ground Range Detected (GRD) product using Sentinel-1A IW imaging mode includes two polarization modes, VV and VH, where VV refers to a co-polarization and VH refers to a cross-polarization. Sentinel-1 covers area Z and requires a total of 12 images, which can achieve 2 to 3 repeated observations per month.

[0046] The second water body image is an optical image obtained by observation by the Sentinel-2 satellite. Sentinel-2 is a high-resolution multispectral imaging satellite, mainly composed of two satellites, 2A and 2B. The revisit period of one satellite is 10 days, and the revisit period of the dual satellite is 5 days. Sentinel-2 carries a multi-spectrometer with 13 spectral bands spanning visible light, near-infrared and short-wave infrared, with a resolution of 10m for blue, green, red and near-infrared bands. Sentinel-2 has high temporal resolution and spectral resolution, and has the advantage of concentrated and continuous monitoring of the earth's surface.

[0047] In this embodiment, the optical image is affected by clouds. There is more cloudy and rainy weather in area Z. Most of the images taken by Sentinel-2 during transit are too cloudy to be used. Moreover, 44 images are required to fully cover area Z with Sentinel-2 images. Therefore, one clear sky data every three months is selected for water extraction.

[0048] In this embodiment, Sentinel-2 is easily affected by clouds when collecting images, especially in cloudy and rainy weather similar to area Z. It takes three months or even six months to determine a usable image to achieve water extraction. Since the time consumed for the acquisition of available images is too long, it is difficult to achieve dynamic monitoring of surface water bodies using Sentinel-2 alone. However, radar data has all-weather and all-day monitoring capabilities in cloudy and rainy mountainous areas. It can achieve the ability of multiple satellite radar data to achieve high-frequency observation under spatial resolution conditions, and realize the dynamic monitoring capability of water bodies in mountainous areas to meet the needs of end users. For example, Sentinel 1A can achieve revisit capabilities within 12 days.

[0049] The radar image collected by Sentinel 1 and the optical image collected by Sentinel 2 are combined for processing. For example, Sentinel 1 obtains a radar image every 12 days, which is processed together with the optical image collected by Sentinel 2 during the same period as the first water body image and the second water body image, respectively. This ensures high-precision extraction of surface water bodies while also enabling dynamic monitoring.

[0050] In this embodiment, after Sentinel-1 acquires a first water body image each time within a period of time, it can be used as a basis for extracting water bodies using the method provided by the present invention based on the second water body image acquired by Sentinel-2 in the same period of time, thereby realizing dynamic observation of water bodies in the target area.

[0051] In one embodiment, before water body extraction, the original radar data collected by Sentinel-1 and the original optical data collected by Sentinel-2 may be preprocessed to filter out the influence of factors such as image noise and improve the accuracy of water body extraction results.

[0052] In one embodiment, the pre-processing of the raw radar data includes but is not limited to orbit correction, radiation calibration, radiation terrain correction, filtering, geometric terrain correction and / or decibelization.

[0053] Orbit correction. During the operation of the satellite, due to external interference or its own reasons, the flight attitude will be slightly shaken and offset, which will introduce certain systematic errors to data acquisition. The orbit state parameters provided in the SAR metadata are usually inaccurate. Therefore, in order to better remove the systematic errors caused by the orbit, it is necessary to use precise orbit information data to perform precise orbit correction to obtain accurate satellite geometric position and velocity information.

[0054] Radiometric calibration. Radiometric calibration is the process of converting the original image pixel brightness value (Digital Number, DN) into the radiometric brightness value. The radiometric calibration formula of Sentinel-1 is as follows:

[0055]

[0056] In formula (1), DN is the brightness value of the pixel (also called pixel), i represents the pixel position, and A is a calibration parameter, which is one of Sigma-naught (standard backscatter coefficient), Gamma-naught (normalized backscatter coefficient) or Beta-naught (radar reflectivity coefficient). In this embodiment, the image is calibrated to a Beta-naught product.

[0057] Radiometric terrain correction. Radiometric terrain correction is to eliminate the influence of terrain on remote sensing reflectivity. In complex mountainous terrain, different terrain slopes cause image radiation distortion, which increases the difficulty of extracting spectral information of ground objects. Radiometric terrain correction can eliminate radiation changes related to terrain while retaining radiation changes related to land cover, but the influence of incident angle still exists. This step eliminates the effects of perspective contraction, overlap and shadow to a certain extent.

[0058] Filtering. Since the SAR system itself is a coherent system, the superposition of the strengthening and weakening of coherent waves will cause periodic speckle noise on the radar image. Moreover, among the various noise factors that affect the quality of SAR images, the influence of coherent noise is much greater than other types of noise. Therefore, in order to maintain as much image detail information as possible and suppress the influence of noise on image interpretation, filtering processing is required. This method uses the Refined Lee filter (an improved image denoising filter) to process speckle noise.

[0059] Geometric terrain correction. Geometric terrain correction is the process of eliminating and correcting the geometric errors of remote sensing images. Due to the terrain changes of the scene and the tilt of the satellite sensor, the distance in the SAR image may be distorted, which will cause some distortion of the SAR image, such as the reversal of the north-south direction. This method uses the range Doppler algorithm to perform terrain correction to reduce the error impact of terrain distortion in remote sensing images.

[0060] Linear conversion to decibels

[0061] In order to avoid the problem of too small backscatter coefficient values, the preprocessing results are converted to logarithmic values ​​to make visualization and analysis easier. The conversion formula is as follows:

[0062]

[0063] In formula (2), γ 0 Indicates the decibel processing for the normalized backscatter coefficient.

[0064] At this point, the preprocessing of Sentinel-1 raw radar data is completed.

[0065] In one embodiment, after the preprocessing of the Sentinel-1 raw radar data is completed, preliminary water body extraction is performed on the preprocessed Sentinel-1 radar image.

[0066] Specifically, the dual polarization water index (SDWI) is used to perform preliminary water extraction on the preprocessed Sentinel-1 radar image to obtain the water body map of Sentinel-1 every 12 days. The basic principle is to perform band operations based on the VV and VH backscatter coefficients to expand the difference between water bodies and other landforms. The dividing point for dividing water bodies is determined according to the actual situation. For example, in one case, 0 is used as the dividing point, and the part with SDWI greater than 0 is divided into water bodies, and the part with SDWI less than 0 is divided into non-water bodies. For another example, based on experimental data, -0.2 is selected as the threshold segmentation point (i.e., the dividing point), the SDWI image is binarized, and the water body is extracted. The SDWI expression is as follows:

[0067] SDWI=ln(10×VV×VH)-8 (3);

[0068] In formula (3), SDWI is the water extraction index, VV and VH represent the VV polarization mode and VH polarization mode, respectively.

[0069] At this point, the result of preliminary water body extraction from the Sentinel-1 radar image is used as the first water body image.

[0070] In one embodiment, preprocessing of the raw optical data includes but is not limited to radiometric calibration, atmospheric correction, resampling and / or band synthesis.

[0071] Radiometric calibration and atmospheric correction. Radiometric calibration is the process of converting DN values ​​into radiometric brightness values ​​by eliminating the errors caused by the sensor itself. Atmospheric correction is the process of converting the apparent reflectance of the upper atmosphere into the surface reflectance to eliminate the errors caused by atmospheric scattering, atmospheric absorption, and atmospheric reflection.

[0072] Resampling. The atmospherically corrected Sentinel-2 data has three resolutions (10m, 20m, and 60m). The bilinear interpolation method is used to uniformly resample the resolution of all bands to 10m.

[0073] Band synthesis: The resampling process stores the Sentinel-2 data as one band corresponding to one image, and the band synthesis tool is used to synthesize them into a multi-band image for subsequent analysis.

[0074] At this point, the preprocessing of Sentinel-2 raw optical data is completed.

[0075] In one embodiment, after the preprocessing of the Sentinel-2 raw optical data is completed, preliminary water body extraction is performed on the preprocessed Sentinel-2 optical image.

[0076] Specifically, in this embodiment, the single-band threshold method is used to perform preliminary water extraction on the preprocessed Sentinel-2 optical image. The single-band threshold method uses the fact that the spectral characteristics of water bodies in the near-infrared band of remote sensing images are significantly different from those of other objects. In this band, water bodies show high absorption rates while other objects show high reflectivity. By comparing the differences in the spectral curves of different objects in this band, the trial-and-error method is used to find the optimal threshold that can separate water bodies from background objects. The expression is generally:

[0077] ρ NIR <T (4);

[0078] In formula (4), ρ NIR represents the reflectance of water in the near-infrared band; T represents the threshold selected in water extraction. According to experimental data, the threshold is selected as 800.

[0079] At this point, the result of preliminary water body extraction of the Sentinel-2 optical image is used as the second water body image.

[0080] Step 102: superimpose the first water body image and the second water body image to obtain overlapping areas between each first water body unit area and the corresponding second water body unit area.

[0081] In this embodiment, when the first water body image is a radar image and the second water body image is an optical image, due to the different angles of sunlight and the different optical and radar imaging methods, in the remote sensing image, the optical shadow (corresponding to the second shadow area) and the radar shadow (corresponding to the first shadow area) are not in the same orientation, and the two overlap less or even do not overlap, while the actual position of the water body is roughly the same. The first water body image and the second water body image are superimposed to obtain the overlapping area of ​​each first water body unit area and the corresponding second water body unit area, so as to achieve the purpose of removing the shadow area.

[0082] In this embodiment, a single water body in the target area, such as a river, a lake, etc., is distributed in a dispersed manner. Therefore, the first water body image includes at least one first water body unit area, and each first water body unit area is distributed in a dispersed manner. Similarly, the second water body image includes at least one second water body unit area, and each second water body unit area is distributed in a dispersed manner.

[0083] For the same target area, theoretically, the image distribution of the same single water body in the first water body image and the second water body image should be consistent. However, due to the existence of shadows, the distribution of the same single water body in the first water body image and the second water body image is different.

[0084] When the first water body image and the second water body image are superimposed, the first water body unit area and the second water body unit area corresponding to the same single water body will overlap. Figure 2 As shown, for a single water body, the area A within the solid circle represents the first water body unit area of ​​the single water body in the first water body image, and the area B within the dotted circle represents the second water body unit area of ​​the single water body in the second water body image. After the first water body image and the second water body image are superimposed, the dotted line and the area C obtained by the intersection are an overlapping area.

[0085] After the first water body image and the second water body image are superimposed, at least one overlapping area can be obtained.

[0086] Step 103: Determine a final water body map of the target area according to at least one overlapping area.

[0087] In this embodiment, fine extraction of the water body in the target area is achieved based on at least one overlapping area, and a final water body map of the target area is obtained.

[0088] In one embodiment, a final water body map of a target area is determined based on at least one overlapping area. The specific implementation process is as follows: the following processing is performed for each overlapping area: a first ratio of the overlapping area to the corresponding first water body unit area is calculated, and a second ratio of the overlapping area to the corresponding second water body unit area is calculated; when the first ratio is greater than a first threshold value and the second ratio is greater than a second threshold value, the first water body unit area corresponding to the overlapping area in the first water body image is extracted; based on the extracted at least one first water body unit area, a final water body map of the target area is determined.

[0089] In this embodiment, based on Figure 2 Taking an overlapping area as an example, the first ratio of area C to area A is calculated, and the second ratio of area C to area B is calculated. When the first ratio is greater than the first threshold, for example, the first threshold value is 37%, and the second ratio is greater than the second threshold value, for example, the second threshold value is 40%, area A is considered to be a water body and area A is extracted. If the first ratio is less than or equal to the first threshold, and / or the second ratio is less than or equal to the second threshold, area A is considered to be a shadow and area A is removed.

[0090] Based on each extracted first water body unit area, a final water body map of the target area is determined.

[0091] In this embodiment, the ratio of the first threshold value to the second threshold value can be set according to experimental data and / or actual conditions. For example, based on the sample radar image and the sample optical image, the first threshold value and the second threshold value are respectively taken at intervals of 2% and are taken between 0 and 100%, until the predicted final sample water body map reaches the preset error range, and the final values ​​of the first threshold value and the second threshold value are determined. Of course, other methods can also be used to determine the values ​​of the first threshold value and the second threshold value, and the protection scope of the present invention is not limited to the specific values ​​and value-taking methods of the first threshold value and the second threshold value.

[0092] In one embodiment, among the water bodies on the ground, river water bodies are flowing water bodies relative to lakes, and there may be seasonal dry-ups, changes in river channels, and the like. In the process of dynamic observation of the water bodies in the target area, in order to avoid the influence of the river water bodies on the observation, river and lake vector data are used to process the river water bodies. Specifically, the first water body image and the second water body image are superimposed to obtain the overlapping areas of each first water body unit area and the corresponding second water body unit area, and then a first intermediate result is formed based on each overlapping area; the first intermediate result is superimposed with the preset river and lake vector data, and the overlapping areas belonging to the river water bodies in the first intermediate result are removed, wherein the river and lake vector data include regional distribution data of the river water bodies. Each overlapping area is processed separately, and specifically, each overlapping area in the first intermediate result after the river water bodies are removed is processed separately.

[0093] In this embodiment, the river and lake vector data uses a global multi-party river and lake dataset as source data, combined with domestic high-resolution image data, to generate a set of global river and lake datasets, which is an editable and high-precision lake and river vector dataset.

[0094] In this embodiment, the regional distribution data of the river water body in the river and lake vector data is extracted in advance, that is, the preset river and lake vector data is the regional distribution data of the river water body, and the first intermediate result formed by the overlapping area and the preset river and lake vector data are superimposed, and the overlapping area belonging to the river water body is eliminated, thereby avoiding the influence of the river.

[0095] In one embodiment, after at least one first water body unit area is extracted based on the first ratio and the second ratio, most of the shadows can be removed. However, some small patch shadows with an area less than a certain value cannot be removed. In order to avoid removing small patch shadows, the final water body map of the target area is determined based on the extracted at least one first water body unit area. The specific implementation process is as follows: based on the extracted at least one first water body unit area, a second intermediate result is formed; the second intermediate result is superimposed with the preset land cover data, and the first water body unit area of ​​the second intermediate result belonging to the preset land type is removed to form a third intermediate result, wherein the land cover data includes the regional distribution data of the preset land type; based on the third intermediate result, the final water body map of the target area is determined.

[0096] In this embodiment, the land cover data may be ESA land cover data, which is a global 10-meter land cover product for 2020 produced by ESA in collaboration with a number of scientific research institutions around the world based on Sentinel-1 and Sentinel-2 data. The land feature categories are divided into 11 categories: woodland, shrubs, grassland, cultivated land, buildings, deserts, ice and snow, water bodies, wetlands, mangroves, mosses and lichens. The overall accuracy of the product is 74%.

[0097] In this embodiment, the small patch shadows are mainly distributed in the area of ​​the preset land type. The second intermediate result formed by the first water body unit area is superimposed on the preset land cover data, and the first water body unit area belonging to the preset land type is eliminated, thereby removing the small patch shadows.

[0098] In this embodiment, the preset land type is determined according to experimental data and / or actual conditions. For example, by superimposing the sample results of the preliminary extraction of water bodies from the Sentinel-1 sample radar image and the ESA land cover data, it is found that the shadows mistakenly extracted in the Sentinel-1 sample results are mostly distributed in the woodland, grassland, building and desert types of the ESA land cover data. Therefore, the woodland, grassland and building types are extracted as masks according to the ESA land cover data, and are superimposed and analyzed with the water body results of Sentinel-1 in turn to remove the mistaken parts of the water body. The desert of the ESA land cover data is defined as the exposed soil, sand and rock with no more than 10% of the vegetation coverage area at any time of the year, and its water body is defined as a geographical area covered by water for more than 9 months a year, including lakes, reservoirs and rivers. Some small rivers or ponds may only have water during the flood season. In order to preserve the integrity of the water body, the desert type is not used as a mask to remove the shadow. For another example, only one preset land type, such as woodland, is determined to remove small patch shadows. The protection scope of the present application is not limited to the specific content and determination method of the preset land type. In this embodiment, the influence of small patch shadows is removed through land cover data, further improving the accuracy of water body extraction.

[0099] In one embodiment, when the river water body is eliminated by the river and lake vector data in the above embodiment, when the final water body map is generated, the river water body is added to the final result. Specifically, based on the third intermediate result, the final water body map of the target area is determined, and the implementation process is as follows: superimpose the third intermediate result and the regional distribution data of the river water body, and use the superimposed result as the final water body map of the target area.

[0100] In a specific embodiment, Figure 3 As shown, the specific implementation of water extraction is as follows:

[0101] The ground range detected (GRD) data (using VV and VH polarization modes) of Sentinel-1A IW imaging mode is used for preliminary water extraction. Specifically, it includes orbit correction, radiation calibration, radiation terrain correction, filtering, geometric terrain correction and decibel processing, as well as threshold segmentation using SDWI to complete preliminary water extraction and obtain the first water image. The first water image is a water map obtained by Sentinel-1 every 12 days.

[0102] The impact data obtained by Sentinel-2 is used to perform preliminary water extraction. Specifically, it includes pre-processing radiation calibration, atmospheric correction, resampling and band synthesis, and the use of threshold method to complete the preliminary water extraction and obtain the second water image. The second water image is a quarterly (three-month) water map obtained by Sentinel-2.

[0103] The first water body image and the second water body image are superimposed to obtain the intersecting water body of Sentinel-1 and Sentinel-2 (i.e., the first intermediate result in the above embodiment). The river water body in the first intermediate result is eliminated by the distribution area data of the river water body extracted in advance from the river and lake vector data. Then, the overlap of each static single water body is calculated, such as lakes, reservoirs, and ponds. The overlap is represented by a first ratio and a second ratio. When an overlapping area is determined to be a shadow by the first ratio and the second ratio, the first water body unit area corresponding to the overlapping area is eliminated; when an overlapping area is determined to be a water body, the first water body unit area corresponding to the overlapping area is retained.

[0104] Then, using the ESA land cover data, the parts of each extracted first water unit area that overlap with woodland, grassland and buildings are removed to finally obtain the required final water map. If necessary, the accuracy of the final water map can be verified using a preset method.

[0105] In a specific embodiment, Figure 4 As shown, Figure 4 To represent the preliminary extraction results of water bodies (Sentinel-1_water) in radar images (Sentinel-1); Figure 5 As shown, Figure 5 This is the initial extraction result of water body (Sentinel-2_water) in optical image (Sentinel-2). Figure 4 and Figure 5 , and get Figure 6 The results shown, Figure 6 Represents the intersection of the preliminary water extraction results of Sentinel-1 and Sentinel-2 (water_Intersect).

[0106] Next, the river water bodies are removed, and this process can be achieved using river and lake vector data. Figure 7 It represents the result of intersection excluding rivers (water_lake). Figure 8 It represents the result of Sentinel-1 preliminary water body removal of rivers (S1_water_lake). Fig. 9 Represents the result of Sentinel-2 preliminary water body removal of rivers (S2_water_lake).

[0107] like Fig.10 As shown in the figure, after removing the river water body, the first and second proportions of the individual water bodies in Sentinel-1 and Sentinel-2 are calculated. percent_s1 represents the first proportion, percent_s2 represents the second proportion, and percent_s1>37% and percent_s2>40% are used as the split points for extraction. Fig.10 It indicates the water extraction results (result) of percent_s1>37% and percent_s2>40%.

[0108] like Figures 11 to 13 As shown, the next step is to remove small patch shadows. Specifically, the forest, grassland, and building types of ESA land cover data are used for overlay analysis to remove small patch shadows. Based on the results of the previous step, Fig.11 Indicates the result of overlaying with the forest land (result_tree), Fig.12 Indicates the result of overlaying with woodland and grassland (result_tree_grass), Fig.13 Represents the result of superimposing with woodland, grassland and building (result_tree_grass_built).

[0109] like Fig.14 As shown in the figure, the results after removing small patches are intersected with the results of Sentinel-1 water bodies, and the water bodies that have intersections between the results of Sentinel-1 water bodies and the results after removing small patches are extracted. Fig.14 Indicates the water body (s1_lake_result) where the results after extracting Sentinel-1 and removing small patches intersect.

[0110] like Fig.15 As shown, the obtained Sentinel-1 lake water body is merged with the Sentinel-1 river water body to obtain the final result. Fig.15 Indicates adding river water bodies to obtain the final water body map (last_result).

[0111] The water body extraction method provided by the present invention obtains a first water body image of a target area, and after obtaining a second water body image of the target area, the first water body image and the second water body image are superimposed to obtain the overlapping area of ​​each first water body unit area and the corresponding second water body unit area, and then the final water body map of the target area is determined according to at least one overlapping area. In the above process, there is a first shadow area caused by a non-water body in the first water body image. There is a second shadow area caused by a non-water body in the second water body image. The first shadow area is distributed in at least one first water body unit area, and the second shadow area is distributed in at least one second water body unit area. However, due to the different distribution of the first shadow area and the second shadow area, when the first water body image and the second water body image of the same target area are superimposed, the actual distribution area of ​​the same water body will overlap, that is, there will be an overlapping area, but the shadow areas with different distributions in the two water body images will not overlap. Determining the final water body map based on the overlapping area reduces the influence of removing the shadows in the first water body map and / or the second water body map, and improves the accuracy of water body extraction.

[0112] Especially for mountainous areas dominated by mountainous and hilly terrain, for example, the area Z mentioned in the above embodiment, the mountains will cause a large number of shadows in the radar image and the optical image, and the method provided by the present invention can effectively extract the water area and avoid the influence of shadows, thereby greatly reducing the influence of mountainous terrain on water extraction, and realizing high-precision and accurate extraction of surface water in mountainous areas.

[0113] Of course, when water body extraction is difficult in a region due to other factors, such as clouds, woodlands, grasslands, etc., if a first water body image and a second water body image with different distribution of shadow areas can be obtained, the method provided by the present invention can also be used to achieve fine extraction of surface water bodies.

[0114] The water extraction device provided by the present invention is described below. The water extraction device described below and the water extraction method described above can be referred to in correspondence with each other. Fig.16 As shown, the water extraction device comprises:

[0115] An acquisition module 1601 is used to acquire a first water body image of a target area, and acquire a second water body image of a target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by a non-water body is distributed in at least one first water body unit area, a second shadow area caused by a non-water body is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different;

[0116] A superposition module 1602 is used to superimpose the first water body image and the second water body image to obtain an overlapping area between each first water body unit area and the corresponding second water body unit area;

[0117] The determination module 1603 is used to determine a final water body map of the target area according to at least one overlapping area.

[0118] In one embodiment, the determination module 1603 is specifically used to perform the following processing on each overlapping area: calculate a first ratio of the overlapping area to the corresponding first water body unit area, and calculate a second ratio of the overlapping area to the corresponding second water body unit area; when the first ratio is greater than a first threshold and the second ratio is greater than a second threshold, extract the first water body unit area corresponding to the overlapping area in the first water body image; based on at least one extracted first water body unit area, determine the final water body map of the target area.

[0119] In one embodiment, the water body extraction device further includes a fine processing module 1604. The fine processing module 1604 is specifically used to superimpose the first water body image and the second water body image, and after obtaining the overlapping area between each first water body unit area and the corresponding second water body unit area, it also includes: forming a first intermediate result based on each overlapping area; superimposing the first intermediate result with preset river and lake vector data, and removing the overlapping area belonging to the river water body in the first intermediate result, wherein the river and lake vector data includes the regional distribution data of the river water body; performing the following processing on each overlapping area, including: performing the following processing on each overlapping area in the first intermediate result after removing the river water body.

[0120] In one embodiment, the determination module 1603 is specifically used to form a second intermediate result based on at least one extracted first water body unit area; superimpose the second intermediate result with preset land cover data, and eliminate the first water body unit area belonging to the preset land type in the second intermediate result to form a third intermediate result, wherein the land cover data includes regional distribution data of the preset land type; based on the third intermediate result, determine the final water body map of the target area.

[0121] In one embodiment, the determination module 1603 is specifically used to superimpose the third intermediate result and the regional distribution data of the river water body, and use the superimposed result as the final water body map of the target area.

[0122] In one embodiment, the first water body image is a radar image, which is formed by collecting water in a target area through radar equipment; the second water body image is an optical image, which is formed by collecting water in a target area through optical equipment.

[0123] Fig.17 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.17 As shown, the electronic device may include: a processor 1701, a communication interface 1702, a memory 1703 and a communication bus 1704, wherein the processor 1701, the communication interface 1702 and the memory 1703 communicate with each other through the communication bus 1704. The processor 1701 may call the logic instructions in the memory 1703 to execute the water body extraction method, which includes: obtaining a first water body image of the target area, and obtaining a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, the first shadow area caused by the non-water body is distributed in at least one first water body unit area, the second shadow area caused by the non-water body is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different; superimposing the first water body image and the second water body image to obtain the overlapping area of ​​each first water body unit area and the corresponding second water body unit area; and determining the final water body map of the target area according to at least one overlapping area.

[0124] In addition, the logic instructions in the above-mentioned memory 1703 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when 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 several 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.

[0125] On the other hand, the present invention also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the water body extraction method provided by the above-mentioned methods, and the method includes: obtaining a first water body image of the target area, and obtaining a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by non-water bodies is distributed in at least one first water body unit area, and a second shadow area caused by non-water bodies is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different; superimposing the first water body image and the second water body image to obtain the overlapping area between each first water body unit area and the corresponding second water body unit area; and determining the final water body map of the target area based on at least one overlapping area.

[0126] 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 water body extraction method provided by the above-mentioned methods, the method comprising: obtaining a first water body image of a target area, and obtaining a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by non-water bodies is distributed in at least one first water body unit area, a second shadow area caused by non-water bodies is distributed in at least one second water body unit area, and the distribution of the first shadow area and the second shadow area is different; superimposing the first water body image and the second water body image to obtain the overlapping area between each first water body unit area and the corresponding second water body unit area; and determining the final water body map of the target area based on at least one overlapping area.

[0127] 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.

[0128] 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.

[0129] 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 water extraction method, characterized in that: include: Acquire a first water body image of a target area, and acquire a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by a non-water body is distributed in the at least one first water body unit area, and a second shadow area caused by a non-water body is distributed in the at least one second water body unit area, and the distribution of the first shadow area and the second shadow area are different; Superimposing the first water body image and the second water body image to obtain overlapping areas of each first water body unit area and the corresponding second water body unit area; Determining a final water map of the target area based on at least one of the overlapping areas; Determining a final water map of the target area according to at least one overlapping area comprises: The following processing is performed for each overlapping area: a first ratio of the overlapping area to the corresponding first water body unit area is calculated, and a second ratio of the overlapping area to the corresponding second water body unit area is calculated; when the first ratio is greater than a first threshold value, and the second ratio is greater than a second threshold value, the first water body unit area corresponding to the overlapping area in the first water body image is extracted; Determining a final water body map of the target area based on the extracted at least one first water body unit area; After superimposing the first water body image and the second water body image to obtain the overlapping area between each of the first water body unit areas and the corresponding second water body unit areas, the method further includes: Based on each of the overlapping regions, forming a first intermediate result; Superimposing the first intermediate result with preset river and lake vector data, and removing the overlapping area belonging to the river water body in the first intermediate result, wherein the river and lake vector data includes regional distribution data of the river water body; The following processing is performed on each overlapping area, including: For each of the overlapping areas in the first intermediate result after removing the river water body, the following processing is performed respectively; The determining of the final water body map of the target area based on the at least one first water body unit area extracted comprises: forming a second intermediate result based on the at least one first water body unit area extracted; superimposing the second intermediate result with preset land cover data, removing the first water body unit area belonging to the preset land type in the second intermediate result, and forming a third intermediate result, wherein the land cover data includes regional distribution data of the preset land type; and determining the final water body map of the target area based on the third intermediate result; Determining the final water body map of the target area based on the third intermediate result includes: superimposing the third intermediate result and the regional distribution data of the river water body, and using the superimposed result as the final water body map of the target area.

2. The water extraction method according to claim 1, characterized in that: The first water body image is a radar image, and the radar image is formed by collecting water bodies in the target area through radar equipment; The second water body image is an optical image, and the optical image is formed by collecting water in the target area through optical equipment.

3. A water extraction device, characterized in that: include: an acquisition module, configured to acquire a first water body image of a target area, and acquire a second water body image of the target area, wherein the first water body image includes at least one first water body unit area, the second water body image includes at least one second water body unit area, a first shadow area caused by a non-water body is distributed in the at least one first water body unit area, and a second shadow area caused by a non-water body is distributed in the at least one second water body unit area, and the first shadow area and the second shadow area are distributed differently; A superposition module, used for superimposing the first water body image and the second water body image to obtain an overlapping area between each of the first water body unit areas and the corresponding second water body unit areas; A determination module, configured to determine a final water body map of the target area based on at least one of the overlapping areas; The determination module is specifically used for: The following processing is performed for each overlapping area: a first ratio of the overlapping area to the corresponding first water body unit area is calculated, and a second ratio of the overlapping area to the corresponding second water body unit area is calculated; when the first ratio is greater than a first threshold value, and the second ratio is greater than a second threshold value, the first water body unit area corresponding to the overlapping area in the first water body image is extracted; Determining a final water body map of the target area based on the extracted at least one first water body unit area; After superimposing the first water body image and the second water body image to obtain the overlapping area between each of the first water body unit areas and the corresponding second water body unit areas, the method further includes: Based on each of the overlapping regions, forming a first intermediate result; Superimposing the first intermediate result with preset river and lake vector data, and removing the overlapping area belonging to the river water body in the first intermediate result, wherein the river and lake vector data includes regional distribution data of the river water body; The following processing is performed on each overlapping area, including: For each of the overlapping areas in the first intermediate result after removing the river water body, the following processing is performed respectively; The determining of the final water body map of the target area based on the at least one first water body unit area extracted comprises: forming a second intermediate result based on the at least one first water body unit area extracted; superimposing the second intermediate result with preset land cover data, removing the first water body unit area belonging to the preset land type in the second intermediate result, and forming a third intermediate result, wherein the land cover data includes regional distribution data of the preset land type; and determining the final water body map of the target area based on the third intermediate result; Determining the final water body map of the target area based on the third intermediate result includes: superimposing the third intermediate result and the regional distribution data of the river water body, and using the superimposed result as the final water body map of the target area.

4. 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 program, the steps of the water body extraction method according to any one of claims 1 to 2 are implemented.

5. 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 water body extraction method according to any one of claims 1 to 2 are implemented.

6. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the water body extraction method according to any one of claims 1 to 2 are implemented.

Images