A Dynamic Monitoring Method for Surface Water Bodies Based on Multi-Source Remote Sensing and Spatial Similarity Reconstruction

Abstract

This invention provides a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction, belonging to the field of remote sensing technology. The method acquires a sequence of water body distribution maps to be reconstructed from optical remote sensing sensors; it then searches a reference image set for the image with the smallest spatial structural difference between the reconstructed water body distribution map and the original image, using this image as the optimal reference water body map; finally, it replaces missing pixels in each reconstructed water body distribution map using the pixel categories in the optimal reference image, resulting in a reconstructed sequence of water body distribution maps. This method utilizes a pre-constructed set of reference images without missing pixels to perform pixel-level completion of the water body distribution maps acquired by the optical remote sensing sensors, ensuring the integrity of the temporal water body information images acquired by the optical remote sensing sensors over a period of time, thereby guaranteeing the temporal continuity of water body information and meeting the monitoring needs of large-scale, highly dynamic water bodies.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and in particular to a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction. Background Technology

[0002] Surface water bodies are a crucial component of global water resources, and their spatial and temporal distribution and dynamic changes directly impact critical decision-making processes such as watershed water resource allocation, ecosystem health, and flood disaster early warning. In recent years, influenced by the combined effects of global climate change and human activities, the dynamic changes in surface water bodies have become increasingly frequent, exhibiting significant seasonal fluctuations, sudden volatility, and long-term evolutionary characteristics. Therefore, high-frequency dynamic monitoring of surface water bodies has emerged. High-frequency dynamic monitoring refers to a monitoring method that utilizes high-temporal-resolution data acquisition techniques to continuously track target objects at high frequencies. Its core lies in "high frequency + dynamic updates" to promptly detect subtle changes and quickly respond to anomalies.

[0003] Currently, remote sensing satellite data used for surface water monitoring can be mainly divided into two categories: optical remote sensing data and synthetic aperture radar (SAR) data. Optical remote sensing data, such as Sentinel-2 and Landsat, has high spatial resolution and can depict water body boundaries and morphology in detail. However, due to limitations such as clouds, haze, and solar altitude angle, many temporal phases are unusable or even unavailable. SAR (synthetic aperture radar) data, such as Sentinel-1, can acquire surface information around the clock and in all weather conditions, demonstrating stable water body identification capabilities even in cloudy or rainy weather. However, SAR imagery has identification errors in areas with vegetation obscuring, blurred water-land boundaries, and undulating terrain. Therefore, multi-source data fusion strategies are currently widely used to improve the accuracy and coverage integrity of water body identification.

[0004] However, even with the fusion of multi-source remote sensing data, it is still difficult to completely eliminate the significant data loss caused by factors such as cloud cover, orbital misalignment, sensor malfunction, and image intervals during remote sensing monitoring. These missing pixels not only affect the results of single-period water body extraction but also cause discontinuities and distortions in time-series water body changes, limiting high-frequency dynamic monitoring capabilities. Currently, some studies have begun to attempt to introduce image inpainting and interpolation methods (such as temporal interpolation, neighborhood mean imputation, and deep learning reconstruction) to compensate for missing areas. However, existing methods lack stability when faced with large-scale, heavily obscured, or long-interval missing images, and the reconstruction results suffer from error propagation and boundary blurring, making it difficult to meet the practical needs of high-frequency, long-term sequence monitoring of surface water bodies. Summary of the Invention

[0005] This invention provides a dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction. This method addresses the problems of existing water body image reconstruction methods being unable to handle large-scale pixel loss in a single water body image, and being unable to handle long-term image loss in a sequence of water body images. It ensures the integrity and continuity of water body images and meets the needs of high-frequency, long-term sequence monitoring.

[0006] This invention provides a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction, comprising the following steps.

[0007] Obtain the sequence of water body distribution maps to be reconstructed, collected by optical remote sensing sensors;

[0008] For any water body distribution map in the sequence of water body distribution maps to be reconstructed, the reference image with the smallest spatial structure difference between it and the water body distribution map to be reconstructed is found in the reference image set and used as the optimal reference water body map for the water body distribution map to be reconstructed; wherein, all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area;

[0009] Using the pixel categories in the optimal reference water body map, the missing areas in the water body distribution map to be reconstructed corresponding to the optimal reference water body map are replaced with pixels to obtain the reconstructed water body distribution map; the reconstructed water body distribution map sequence corresponding to all the water body distribution map sequences to be reconstructed is used as the dynamic monitoring result of surface water bodies.

[0010] According to the present invention, a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction is provided, wherein the reference image set is constructed in the following manner:

[0011] Obtain a set of optical water body distribution maps and a set of radar water body distribution maps for the same area;

[0012] For each radar water distribution map in the aforementioned radar water distribution map set, areas prone to misjudgment are identified.

[0013] Based on the proportion of missing areas in each optical water distribution map in the optical water distribution map set, or the proportion of the missing areas in the easily misjudged areas of the corresponding radar water distribution map, high-quality optical water distribution maps are selected.

[0014] Based on the spatial structure difference between the radar water distribution map in the radar water distribution map set and the high-quality optical water distribution map, the radar water distribution map in the radar water distribution map set is used to fill the missing area in the high-quality optical water distribution map to obtain a high-quality, missing-free fused water distribution map.

[0015] The reference image set is constructed based on the high-quality, complete, and fused water body distribution map.

[0016] According to the present invention, a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction is provided, wherein the spatial structure difference includes symmetry difference; the step of finding the reference image with the smallest spatial structure difference between the reference image set and the water body distribution map to be reconstructed, as the optimal reference water body map for the water body distribution map to be reconstructed, includes:

[0017] Calculate the symmetry difference between the water body distribution map to be reconstructed and all reference water body distribution maps in the reference image set, and find the reference water body distribution map with the smallest symmetry difference as the optimal reference water body map for the water body distribution map to be reconstructed.

[0018] According to the present invention, a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction, after constructing the reference image set based on the high-quality, complete, and unmissing fused water body distribution map, further includes:

[0019] After removing the high-quality optical water body distribution map from the optical water body distribution map set, the remaining optical water body distribution maps are used as the set of water body distribution maps to be reconstructed.

[0020] From each of the water body distribution maps to be reconstructed in the set of water body distribution maps to be reconstructed, areas with a slope less than or equal to a preset slope threshold and a water body inundation frequency within a preset frequency range are selected as key areas based on slope and water body inundation frequency.

[0021] The missing areas in the key region are filled in using reference images from the reference image set to obtain a reconstructed optical water distribution map.

[0022] According to the present invention, a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction is provided, wherein missing regions in the key area are filled in using reference images from a reference image set to obtain a reconstructed optical water body distribution map, including:

[0023] If water pixels and non-water pixels exist simultaneously in the critical region, the missing areas in the critical region are filled in using images from the reference image set to obtain a reconstructed optical water distribution map.

[0024] If the critical region does not contain both water body pixels and non-water body pixels simultaneously, then the water body distribution map corresponding to that critical region to be reconstructed is discarded.

[0025] According to the present invention, a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction is provided, wherein acquiring a set of optical water body distribution maps and a set of radar water body distribution maps for the same area includes:

[0026] Acquire the set of raw optical remote sensing images and the set of raw radar remote sensing images for the same area;

[0027] The polarization channel is extracted from each original radar remote sensing image in the original radar remote sensing image set to obtain the polarization index.

[0028] The Otsu algorithm is used to determine the water extraction threshold from the polarization index;

[0029] Based on the water body extraction threshold, radar water body distribution maps for each time phase are generated, resulting in the radar water body distribution map set;

[0030] Masking and water pixel extraction are performed on each original optical remote sensing image in the original optical remote sensing image set to obtain optical water distribution maps for each time phase.

[0031] Based on the optical water body distribution maps of each time phase, the set of optical water body distribution maps is obtained.

[0032] This invention also provides a surface water dynamic monitoring system based on multi-source remote sensing and spatial similarity reconstruction, the system comprising:

[0033] Optical remote sensing sensors are used to acquire sequences of optical water body distribution maps for target water areas.

[0034] The processor is used to execute the steps in any of the above-mentioned methods for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction as described above.

[0036] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction as described above.

[0037] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction as described above.

[0038] The present invention provides a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction. This method acquires a sequence of water body distribution maps to be reconstructed from optical remote sensing sensors. For any water body distribution map in the sequence, the method searches a reference image set for the reference image with the smallest spatial structure difference from the original water body distribution map, using this reference image as the optimal reference water body map. All reference images in the reference image set are images without missing pixels. Each reference image is obtained by image fusion based on images acquired by at least two remote sensing sensors targeting the same water area. Using the pixel categories in the optimal reference water body map, missing regions in the corresponding water body distribution map to be reconstructed are replaced with pixels to obtain the reconstructed water body distribution map. The reconstructed water body distribution map sequence corresponding to all the water body distribution map sequences to be reconstructed is used as the result of dynamic monitoring of surface water bodies. This method utilizes a pre-constructed, missing reference image library to perform pixel-level completion on the water body distribution map to be reconstructed acquired by the optical remote sensing sensor. This ensures the integrity of the temporal water body information images acquired by the optical remote sensing sensor over a period of time, thereby guaranteeing the high temporal continuity of water body information. It also ensures the image integrity and temporal continuity in high-frequency dynamic monitoring of surface water bodies, and can meet the monitoring needs of large-scale, highly dynamic water bodies. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction provided by the present invention.

[0041] Figure 2 This is a schematic diagram of the overall process of the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction provided by the present invention.

[0042] Figure 3 This is a schematic diagram illustrating the principle of water distribution map reconstruction using symmetry difference provided by the present invention.

[0043] Figure 4 This is one of the schematic diagrams illustrating the construction process of the reference image set provided by the present invention.

[0044] Figure 5 This is the second schematic diagram of the construction process of the reference image set provided by the present invention.

[0045] Figure 6 This is one of the schematic diagrams of the process for reconstructing the distribution map of a water body to be reconstructed, provided by the present invention.

[0046] Figure 7 This is the second schematic diagram of the process for reconstructing the distribution map of the water body to be reconstructed, provided by the present invention.

[0047] Figure 8 This is a schematic diagram of the preprocessing process for raw optical remote sensing image sets and raw radar remote sensing image sets provided by the present invention.

[0048] Figure 9 This is a schematic diagram of the surface water dynamic monitoring system based on multi-source remote sensing and spatial similarity reconstruction provided by the present invention.

[0049] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0051] The following is combined with Figures 1-8 This invention describes a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction.

[0052] Figure 1 This is a flowchart illustrating the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps:

[0053] Step 101: Obtain the sequence of water body distribution maps to be reconstructed collected by the optical remote sensing sensor.

[0054] Optical remote sensing sensors are devices that detect and identify the characteristics of ground features by receiving and recording the energy generated by the interaction between targets on the Earth's surface and electromagnetic waves. Their working principle involves the sensor collecting electromagnetic wave energy reflected or radiated from the target, converting it into electrical signals, and then processing it to form images or datasets. Simply put, they are cameras installed on platforms such as satellites, aircraft, and drones. The sequence of water body distribution maps to be reconstructed refers to a sequence of water body images captured by the aforementioned optical remote sensing sensors over a period of time, after preprocessing to remove non-water body images and low-quality water body images that have no reconstruction value. This includes water body images from various time phases within that period.

[0055] It is worth mentioning that this application focuses on the reconstruction of optical remote sensing images of target water bodies. This is mainly because while optical remote sensing data, such as Sentinel-2 and Landsat, possess high spatial resolution and can meticulously depict water body boundaries and morphology, they are limited by conditions such as clouds, haze, and solar altitude angle. A large number of temporal optical remote sensing images suffer from pixel loss and distortion, making them difficult to use or even unusable. In contrast, SAR (Synthetic Aperture Radar) data, such as Sentinel-1, while lacking the high resolution of optical remote sensing, can acquire surface information around the clock and in all weather conditions, demonstrating stable water body identification capabilities even in cloudy or rainy weather. Therefore, radar remote sensing data can be used to fill in the distorted, erroneous, or missing parts of optical remote sensing data, forming a complete, clear, and continuously distributed dynamic image of the water body over time.

[0056] Specifically, in combination Figure 2 The overall flowchart of the surface water dynamic monitoring method based on multi-source remote sensing and spatial similarity reconstruction is described below. Step 101 includes: acquiring the original water distribution map sequence collected by optical remote sensing sensors for the target water area, for example, using the Sentinel-2 satellite to collect the original water distribution map sequence of water area A over a period of time, and preprocessing these original water distribution map sequences (such as...). Figure 2 The process involves using indicators such as MNDWI / NDVI / EVI to filter water bodies of interest, removing non-water body images, and determining whether they meet the selection requirements of the reference image set (based on the proportion of missing areas in each optical water body distribution map in that optical water body distribution map, or the proportion of the missing areas in the easily misjudged areas in the corresponding radar water body distribution map, to distinguish high-quality optical water body distribution maps from water body distribution maps to be reconstructed, as described later). If they do not meet the requirements, they are used as water body distribution maps to be reconstructed. The preprocessing steps include removing invalid images with high cloud cover or lacking clear land-water boundaries, retaining water body images, and using them as water body distribution maps to be reconstructed.

[0057] Sentinel-2 is a high-resolution multispectral imaging satellite (i.e., an optical imaging satellite) specifically designed for land monitoring. Its main mission is to provide high-resolution imagery of land and coastal areas worldwide for use in various fields such as environmental monitoring, land use planning, agricultural management, and disaster monitoring.

[0058] Step 102: For any water body distribution map to be reconstructed in the sequence of water body distribution maps to be reconstructed, find the reference image with the smallest spatial structure difference between it and the water body distribution map to be reconstructed in the reference image set, and use it as the optimal reference water body map for the water body distribution map to be reconstructed; wherein, all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area.

[0059] The reference image set is a collection of images without missing pixels obtained by fusing images of the same body of water (the target body of water mentioned above in this application) acquired by at least two remote sensing sensors (especially radar and optical sensors). Spatial structure differences refer to the differences between the two images in terms of the geometric layout, shape, size, orientation, relative positional relationship, and topological structure of objects in the scene.

[0060] Specifically, this application quantifies the spatial structure differences of different water body distribution maps within the effective area (referring to the non-missing area, i.e., the area without missing pixels), and selects the most similar reference image to fill in the missing or distorted pixels in the aforementioned water body distribution map to be reconstructed, thereby restoring the water body information in the missing or distorted areas. The index for measuring spatial structure differences can be pixel-based and statistical, such as mean squared error (MSE), which calculates the average of the squared differences between corresponding pixel values ​​in two images; the smaller the value, the more similar the images. It can also be the structural similarity index (SSIM), which compares images from three dimensions: brightness, contrast, and structure. Other examples include edge structure similarity (such as ESIM) and gradient magnitude similarity deviation (GMSD). This application selects symmetric difference (DP) as the index to describe spatial structure differences. Symmetry difference (DP) is defined as the number of pixels in two water body distribution maps that differ in classification within the valid area (i.e., the non-missing area, where no pixels are missing). For example, if the same location is classified as "water body" in image A but "non-water body" in image B, then the pixels at that location can be considered to have different classifications. The smaller the DP value, the more similar the spatial structure of the two images.

[0061] For any water body distribution map to be reconstructed in step 101 above, calculate the DP (symmetric difference) value between it and all water body distribution maps in the reference image set, and find the reference image with the smallest DP value as the optimal reconstruction basis for the map (i.e. the optimal reference water body map).

[0062] Step 103: Using the pixel categories in the optimal reference water body map, replace the missing areas in the water body distribution map to be reconstructed corresponding to the optimal reference water body map with pixels to obtain the reconstructed water body distribution map; use the reconstructed water body distribution map sequence corresponding to all the water body distribution map sequences to be reconstructed as the dynamic monitoring results of surface water bodies.

[0063] Specifically, by utilizing the pixel categories in the optimal reference water body map (categories can be water body, non-water body, gaps, etc.), the missing areas in each water body distribution map to be reconstructed are replaced with pixels to restore the water body information in the missing areas. Finally, the reconstructed water body distribution map sequences corresponding to all the water body distribution map sequences to be reconstructed are used as the dynamic monitoring results of surface water bodies.

[0064] The above embodiment involves acquiring a sequence of water body distribution maps to be reconstructed from optical remote sensing sensors; for any water body distribution map in the sequence, finding the reference image with the smallest spatial structure difference between it and the water body distribution map to be reconstructed in a reference image set, which is then used as the optimal reference water body map for that water body distribution map; wherein all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images acquired by at least two remote sensing sensors for the same water area; using the pixel categories in the optimal reference water body map, pixel replacement is performed on the missing areas in the water body distribution map to be reconstructed corresponding to the optimal reference water body map to obtain the reconstructed water body distribution map; the reconstructed water body distribution map sequence corresponding to all water body distribution map sequences to be reconstructed is used as the dynamic monitoring result of surface water bodies. This method utilizes a pre-constructed, missing reference image library to perform pixel-level completion on the water body distribution map to be reconstructed acquired by the optical remote sensing sensor. This ensures the integrity of the temporal water body information images acquired by the optical remote sensing sensor over a period of time, thereby guaranteeing the high temporal continuity of water body information. It also ensures the image integrity and temporal continuity in high-frequency dynamic monitoring of surface water bodies, and can meet the monitoring needs of large-scale, highly dynamic water bodies.

[0065] In one embodiment, the aforementioned spatial structure difference includes the symmetry difference (DP), and step 102 includes: calculating the symmetry difference between the water body distribution map to be reconstructed and all reference water body distribution maps in the reference image set, and finding the reference water body distribution map with the smallest symmetry difference as the optimal reference water body map for the water body distribution map to be reconstructed.

[0066] Specifically, such as Figure 3 As shown, Figure 3 This diagram illustrates the principle of reconstructing water distribution maps using symmetry difference. Figure 3 In the middle, the water body distribution map to be reconstructed Let represent the i-th water body distribution map to be reconstructed, where blue squares represent pixels identified as "water," yellow squares represent pixels identified as "not water," and gray squares represent pixels identified as "gap." This is to reconstruct the water body distribution map. To reconstruct and fill in the missing pixels, it is necessary to find the distribution map of the water body to be reconstructed in the reference image set. The reference image with the smallest symmetry difference (DP) can be selected. For example, each reference image in the above reference image set can be compared with the water distribution map to be reconstructed. Symmetric difference (DP) between them, such as Figure 3 As shown, the reference image is calculated respectively. Map of water bodies to be reconstructed Symmetry difference (DP) between reference images Map of water bodies to be reconstructed The symmetric difference (DP) between them..., the symmetric differences obtained are respectively , ...from these, the reference image with the smallest symmetry difference is selected as the optimal reference water body image. Figure 3 In, it has the smallest symmetric difference (i.e. Figure 3 As shown The reference image is Then As the optimal reference water body map.

[0067] Furthermore, such as Figure 3 As shown, using the optimal reference water body map To reconstruct the water body distribution map Pixel-level filling is performed on the missing pixels to obtain the reconstructed water distribution map.

[0068] The above embodiments, by constructing a symmetry difference (DP) index, can rigorously quantify the spatial difference between the missing map and the reference map within the effective area, and accurately select the optimal reference image for pixel-level filling. Existing technologies generally rely on time-series interpolation, water body probability models, or topographic assumptions, making it difficult to accurately recover the spatial distribution of water bodies when there are strong dynamic changes or observational breaks. Compared to existing methods, this application places greater emphasis on maintaining the true consistency of water body boundaries and morphology (primarily through pre-constructed high-quality reference images for pixel-level filling, without altering pixels in non-missing areas, and using high-quality reference images to fill pixels in missing areas, ensuring image integrity). It is particularly suitable for areas with complex water body morphology and significant seasonal changes, such as lakes and river networks, resulting in a more reasonable spatial morphology and more accurate trend prediction in the reconstruction.

[0069] In one embodiment, such as Figure 4 As shown, Figure 4 One of the schematic diagrams illustrating the construction process of the reference image set is shown. The reference image set is constructed in the following manner:

[0070] Step 401: Obtain the set of optical water distribution maps and the set of radar water distribution maps for the same area.

[0071] The optical and radar water distribution maps are both pre-processed images, and the specific pre-processing steps are described in detail later. Both sets of images are time-series images collected within the same time frame for the same water body region. This water body region is also the area requiring high-frequency dynamic monitoring, which is the target water body region monitored in step 101 above.

[0072] Specifically, such as Figure 5 As shown, Figure 5 The second illustration shows the construction process of the reference image set. For the same area, the original optical remote sensing image sequence is acquired using an optical remote sensing sensor (e.g., Sentinel-2), and these sequence images are preprocessed to obtain the optical water body distribution map set. Similarly, the original radar remote sensing image sequence is acquired using a radar remote sensing sensor (e.g., Sentinel-1), and these sequence images are preprocessed to obtain the radar water body distribution map set.

[0073] Step 402: For each radar water distribution map in the radar water distribution map set, identify areas prone to misjudgment.

[0074] It should be noted that in radar imagery, steep terrain (large slopes) or high-altitude areas far from waterways (large hands) may be mistakenly identified as "body of water" due to the special reflection or shadowing effects of radar signals. For example, radar may generate low-echo signals resembling water bodies on steep slopes or at high altitudes, causing automatic algorithms to misidentify them as water surfaces. Areas prone to misidentification are those easily mistaken for bodies of water.

[0075] Specifically, this embodiment introduces auxiliary terrain information to identify areas prone to misjudgment in the radar water distribution map. More specifically, it calculates the slope using NASADEM data and combines it with HAND data to calculate the relative drainage elevation. When the slope exceeds a preset slope threshold (e.g., 15°), or the HAND value exceeds a preset HAND threshold (e.g., this threshold is set to 30m), the area is considered to have a high risk of misjudgment in the radar water distribution map and is marked as a misjudgment-prone area.

[0076] NASA ADEM (NASA Digital Elevation Model) is the name of a global digital elevation model (DEM). As the name suggests, it is a set of high-precision, near-global coverage Earth's land elevation data developed and released by NASA, providing elevation (DEM) data, slope, aspect data, and more.

[0077] HAND (Height Above Nearest Drainage) refers to the vertical difference in elevation between any point on the Earth's surface and the elevation of the bed of its nearest drainage channel (such as a river or stream). HAND data is typically generated automatically using a digital elevation model (DEM).

[0078] Step 403: Based on the proportion of missing regions in each optical water distribution map in the optical water distribution map set, or the proportion of the missing regions in the easily misjudged regions of the corresponding radar water distribution map, high-quality optical water distribution maps are selected.

[0079] Specifically, the optical water distribution maps for each time phase in the optical water distribution map set are screened: if the proportion of invalid pixels (i.e., missing pixels) in the entire optical water distribution map is less than a first proportion threshold (e.g., 5%), or if the proportion of invalid pixels (i.e., missing pixels) located in the easily misjudged area of ​​the corresponding radar water distribution map is less than a second proportion threshold (e.g., 5%), then the optical water distribution map is considered a high-quality optical water map (which can be used as the basis for constructing the reference image set). This type of high-quality optical water map can be used to fill in missing regions using the spatial similarity reconstruction algorithm described above, generating a high-quality, missing-free fused water map, based on the corresponding radar water distribution map. The remaining low-quality optical water distribution maps are then used as water distribution maps to be reconstructed for the next step of reconstruction.

[0080] Step 404: Based on the spatial structure difference between the radar water distribution map in the radar water distribution map set and the high-quality optical water distribution map, the radar water distribution map in the radar water distribution map set is used to fill the missing areas in the high-quality optical water distribution map to obtain a high-quality, missing-free fused water distribution map.

[0081] Specifically, for any high-quality optical water body distribution map, the DP (symmetric difference) value between it and the radar water body distribution maps in the radar water body distribution map set is calculated. The radar water body distribution map with the minimum DP value is found as the optimal reconstruction basis for the high-quality optical water body distribution map. The pixels in the radar water body distribution map with the minimum DP value that differ from those in the high-quality optical water body distribution map are used to fill the missing areas in the high-quality optical water body distribution map, thus obtaining a high-quality, missing-free fused water body distribution map.

[0082] Step 405: Based on the high-quality, complete, and fused water body distribution map, construct the reference image set.

[0083] Specifically, the set of high-quality optical water distribution maps (i.e., high-quality, missing-free fused water distribution maps) constructed above is used as a reference image set.

[0084] The above embodiments utilize radar imagery to assist in constructing an optical water body distribution map. By setting appropriate screening ratio thresholds to eliminate high-false-positive and high-missing data, a high-quality, missing-free fused water body distribution map with continuous time series and no missing spatial coverage is formed. A reference image set is constructed to improve the accuracy and reliability of image reconstruction during subsequent dynamic monitoring.

[0085] In one embodiment, such as Figure 6 As shown, Figure 6One of the flowcharts illustrates the reconstruction process for an optical water body distribution map. It should be noted that, to ensure the temporal continuity of the optical water body distribution map sequence, the low-quality optical water body distribution maps selected during the construction of the reference image set also need to be reconstructed. This ensures the temporal continuity of the water body distribution maps, thus providing a complete data foundation for subsequent ecological monitoring. Therefore, after step 405, the following steps are also included:

[0086] Step 601: Remove the high-quality optical water distribution map from the optical water distribution map set and use the remaining optical water distribution map as the set of water distribution maps to be reconstructed.

[0087] Specifically, in combination Figure 7 The second flowchart illustrating the process of reconstructing the water body distribution map is as follows: In step 601 above, optical water body distribution maps with smaller missing areas or smaller proportions of missing areas in easily misjudged areas are selected from the set of optical water body distribution maps as high-quality optical water body distribution maps. The remaining low-quality optical water body distribution maps (i.e., optical water body distribution maps with larger missing areas, especially those with a large proportion of missing areas in easily misjudged areas) are used as the set of water body distribution maps to be reconstructed.

[0088] Step 602: From each water body distribution map to be reconstructed in the set of water body distribution maps to be reconstructed, select areas whose slope is less than or equal to a preset slope threshold and whose water body inundation frequency is within a preset frequency range as key areas based on slope and water body inundation frequency.

[0089] Specifically, considering that the spatial distribution of seasonal water bodies in flat areas best represents the overall water pattern of the study area, masking can be performed based on slope and water inundation frequency. Specifically: areas with a slope greater than a preset slope threshold (e.g., a preset slope threshold of 15°) and a water inundation frequency less than 0.02 or greater than 0.98 are considered non-critical areas, and the water changes in these areas are not representative; the remaining areas are considered critical areas, that is, areas with a slope less than or equal to a preset slope threshold (e.g., a preset slope threshold of 15°) and a water inundation frequency within the closed interval [0.02, 0.98] are considered critical areas.

[0090] Mask filtering refers to the method of extracting the part of interest from the original data using a preset template (i.e., a mask).

[0091] Optionally, mask screening can also be performed based on three factors: slope, water inundation frequency, and spatial variation rate of water inundation frequency. Specifically, areas with a slope greater than a preset slope threshold (e.g., the preset slope threshold is set to 15°), a water inundation frequency less than 0.02 or greater than 0.98, and a spatial variation rate of water inundation frequency greater than 0.1 are considered non-critical areas, while the remaining areas are considered critical areas.

[0092] Step 603: Fill in the missing areas in the key region using reference images from the reference image set to obtain the reconstructed optical water distribution map.

[0093] Specifically, based on the spatial similarity reconstruction algorithm described above, the high-quality, missing reference image set obtained in steps 401-405 can be used to select the optimal reference image to fill in the missing areas of the target water body image suitable for reconstruction at the pixel level, thereby restoring its complete water body distribution information.

[0094] The above embodiments obtain complete water distribution information by performing missing reconstruction on the low-quality optical water distribution map, i.e., the water distribution map to be reconstructed, so as to fuse it with the high-quality, missing-free water distribution in the aforementioned reference image set. Figure 1 This forms a high-frequency dynamic water distribution map of the water area, ensuring the spatial integrity and temporal continuity of water body observations in the area.

[0095] In one embodiment, step 603 includes:

[0096] If water pixels and non-water pixels exist simultaneously in the critical region, the missing areas in the critical region are filled in using images from the reference image set to obtain a reconstructed optical water distribution map.

[0097] If the critical region does not contain both water body pixels and non-water body pixels simultaneously, then the water body distribution map corresponding to that critical region to be reconstructed is discarded.

[0098] In this context, a pixel refers to the smallest unit of a digital image, also known as a water body pixel. In this application, a water body pixel refers to a pixel that appears as a water body, and a non-water body pixel refers to any other pixel besides the aforementioned water body pixel.

[0099] In this embodiment, during the process of filling in the missing areas in the key areas of the above-mentioned non-high-quality optical water distribution map (i.e., the water distribution map to be reconstructed), it is necessary to determine whether these missing areas are suitable for reconstruction. In this embodiment, the determination of whether they are suitable for reconstruction is based on the distribution of water pixels and non-water pixels.

[0100] Specifically, if both water body pixels and non-water body pixels exist in the aforementioned key areas, they are considered suitable for reconstruction. Therefore, the spatial similarity reconstruction algorithm described above can be used to reconstruct the missing areas. That is, the high-quality, missing-free fused water body distribution map from the aforementioned reference image set is used to fill in the missing areas suitable for reconstruction at the pixel level, restoring their complete water body distribution information. The specific process will not be elaborated here. Otherwise, they are considered not worth reconstructing, and this will not be elaborated here either.

[0101] The above embodiment further assesses missing areas by determining their suitability for reconstruction based on the presence of both water and non-water pixels. If suitable, pixel-level filling is performed to obtain a complete image; otherwise, the image is discarded to avoid wasting resources. This embodiment helps ensure the integrity and temporal continuity of water distribution maps for the same area.

[0102] In one embodiment, such as Figure 8 As shown, Figure 8 This diagram illustrates the preprocessing workflow for the raw optical remote sensing image set and the raw radar remote sensing image set. Step 401 includes:

[0103] Acquire a set of original optical remote sensing images and a set of original radar remote sensing images for the same area; extract the polarization channel from each original radar remote sensing image in the original radar remote sensing image set to obtain the polarization index; use the Otsu algorithm to determine the water body extraction threshold from the polarization index; generate radar water body distribution maps for each time phase based on the water body extraction threshold to obtain the set of radar water body distribution maps;

[0104] Each original optical remote sensing image in the original optical remote sensing image set is masked and water body pixels are extracted to obtain optical water body distribution maps for each time phase; based on the optical water body distribution maps for each time phase, the optical water body distribution map set is obtained.

[0105] In this context, polarization channels refer to the different combinations of polarization modes employed by a radar system when transmitting and receiving electromagnetic waves. Each combination constitutes an independent observation channel, providing unique information. There are four main basic polarization channel combinations:

[0106] 1. HH Channel: Transmits horizontal (H) waves and receives horizontal (H) waves.

[0107] 2. HV channel: Transmits horizontal (H) waves and receives vertical (V) waves.

[0108] 3. VH channel: Transmits vertical (V) waves and receives horizontal (H) waves.

[0109] 4. VV Channel: Transmits vertical (V) waves and receives vertical (V) waves.

[0110] By analyzing and combining different polarization channels, it is easy to distinguish between water bodies and non-water bodies.

[0111] Specifically, radiometric correction, geometric correction, and range cropping are performed on Sentinel-1 (radar) images (i.e., the original set of radar remote sensing images), and VV and VH polarization channels are extracted to construct the VV×VH index. The Otsu algorithm is used to determine the water body extraction threshold of the index, and the initial water body distribution map corresponding to each temporal phase of Sentinel-1 (radar) images is generated based on the water body extraction threshold.

[0112] The Otsu algorithm, also known as the maximum inter-class variance method, is a statistical method that automatically determines the optimal threshold for image binarization by maximizing the inter-class variance between the foreground and background. Its main steps include: 1. Calculating the grayscale histogram; 2. Iterating through all possible thresholds; 3. Calculating the inter-class variance at the current threshold; 4. Finding the optimal threshold, i.e., finding the threshold that maximizes the inter-class variance. This threshold is the optimal threshold calculated by the Otsu algorithm, which is the water extraction threshold proposed in this embodiment.

[0113] Furthermore, for Sentinel-2 (optical) imagery (i.e., the original set of optical remote sensing images), pre-defined cloud and shadow masks are applied to remove invalid images with high cloud cover or lacking clear land-water boundaries, retaining only high-quality images for water body extraction. Water body distribution is extracted based on the Modified Normalized Difference Water Index (MNDWI), and further removed by combining the Enhanced Vegetation Index (EVI) and the Normalized Difference Vegetation Index (NDVI), ultimately generating optical water body distribution maps corresponding to Sentinel-2 (optical) imagery at each temporal phase.

[0114] The Normalized Difference Water Index (MNDWI) is a method that uses multispectral data from remote sensing satellites to enhance water body information while suppressing non-water body information (such as soil, buildings, and vegetation) through a simple mathematical calculation. Its core purpose is to more accurately and quickly identify and extract water body ranges, such as lakes, rivers, and reservoirs, from satellite imagery.

[0115] The Enhanced Vegetation Index (EVI) is an indicator that quantifies the greenness and density of surface vegetation using multispectral data from remote sensing satellites. Its core objective is to minimize the interference of atmospheric conditions (such as aerosols) and soil background on vegetation index results while retaining the advantage of NDVI's sensitivity to vegetation, thereby providing more accurate vegetation information, especially in areas with high biomass (dense vegetation).

[0116] The Normalized Difference Vegetation Index (NDVI) is an indicator that uses multispectral data from remote sensing satellites to quantify land cover through a simple mathematical calculation. Its core purpose is to rapidly and over a large area assess the quantity, health status, and biomass of vegetation.

[0117] In the above embodiments, by performing preprocessing operations on the original optical water distribution map set and the original radar water distribution map set respectively, the final optical water distribution map set and radar water distribution map set are obtained. This allows for further water extraction or missing reconstruction of these preprocessed optical water distribution map sets and radar water distribution map sets, thereby constructing a complete high-quality image.

[0118] To verify the effectiveness and applicability of the surface water dynamic monitoring method based on multi-source remote sensing and spatial similarity reconstruction proposed in this application, the following experiments were conducted.

[0119] (1) Large-scale application experiments.

[0120] The proposed method was applied in the D Lake area, where water dynamics changed frequently and hydrological conditions were complex, verifying the method's broad applicability and stability. Results show that the proposed method can effectively recover water information in cloud-obscured areas, ensuring the integrity and continuity of monitoring data.

[0121] (2) Reconstruction accuracy verification.

[0122] To further verify the accuracy and stability of this method in water loss restoration, the following two aspects were analyzed:

[0123] a. Analysis of the relationship between water surface area and water level.

[0124] Using measured water level data from Lake D as a reference, the changes in water surface area extracted before and after water body image reconstruction were statistically analyzed, and their correlation with water level changes was examined. By comparing the correlation between water surface area and measured water level, as well as the time series change trend, it was shown that the reconstructed water body map has a higher consistency with water level changes and can accurately reflect the spatial changes in water body caused by water level changes, thus verifying the rationality and effectiveness of the proposed method.

[0125] b. Assessment of reconstruction accuracy under different cloud cover scenarios.

[0126] Historical cloudless Sentinel-2 optical images were selected, covering different seasons and water areas. Different cloud cover masks were artificially added to construct various cloud pollution scenarios before water body reconstruction experiments were conducted. The images were then compared with the original images, and quantitative evaluation was performed using three indicators: overall accuracy (OA), recall, and precision.

[0127] Table 1 shows the reconstruction results under different cloud cover conditions. The experimental results show that the method in this application maintains high reconstruction accuracy under low, medium, and high cloud cover scenarios, demonstrating good robustness and universality.

[0128] Table 1. Accuracy of the method in this application under different cloud cover conditions.

[0129]

[0130] like Figure 9 As shown, Figure 9 A schematic diagram of a surface water dynamic monitoring system based on multi-source remote sensing and spatial similarity reconstruction is shown. The system includes the following modules:

[0131] Optical remote sensing sensor 901 is used to collect optical water body distribution map sequences for target water areas;

[0132] Processor 902 is used to perform the steps in any of the above embodiments of the dynamic monitoring method for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction.

[0133] In one embodiment, the spatial structure difference includes a symmetry difference; the processor 902 is further configured to:

[0134] Calculate the symmetry difference between the water body distribution map to be reconstructed and all reference water body distribution maps in the reference image set, and find the reference water body distribution map with the smallest symmetry difference as the optimal reference water body map for the water body distribution map to be reconstructed.

[0135] In one embodiment, the system further includes a radar remote sensing sensor for collecting a set of radar water distribution maps for the same area; and the optical remote sensing sensor 901 for collecting a set of optical water distribution maps for the same area.

[0136] The aforementioned processor 902 is further configured to: acquire the optical water distribution map set and the radar water distribution map set;

[0137] For each radar water distribution map in the aforementioned radar water distribution map set, areas prone to misjudgment are identified.

[0138] Based on the proportion of missing areas in each optical water distribution map in the optical water distribution map set, or the proportion of the missing areas in the easily misjudged areas of the corresponding radar water distribution map, high-quality optical water distribution maps are selected.

[0139] Based on the spatial structure difference between the radar water distribution map in the radar water distribution map set and the high-quality optical water distribution map, the radar water distribution map in the radar water distribution map set is used to fill the missing area in the high-quality optical water distribution map to obtain a high-quality, missing-free fused water distribution map.

[0140] The reference image set is constructed based on the high-quality, complete, and fused water body distribution map.

[0141] In one embodiment, the processor 902 is further configured to:

[0142] After removing the high-quality optical water body distribution map from the optical water body distribution map set, the remaining optical water body distribution maps are used as the set of water body distribution maps to be reconstructed.

[0143] From each of the water body distribution maps to be reconstructed in the set of water body distribution maps to be reconstructed, areas with a slope less than or equal to a preset slope threshold and a water body inundation frequency within a preset frequency range are selected as key areas based on slope and water body inundation frequency.

[0144] The missing areas in the key region are filled in using images from the reference image set to obtain a reconstructed optical water distribution map.

[0145] In one embodiment, the processor 902 is further configured to:

[0146] If water pixels and non-water pixels exist simultaneously in the critical region, the missing areas in the critical region are filled in using images from the reference image set to obtain a reconstructed optical water distribution map.

[0147] If the critical region does not contain both water body pixels and non-water body pixels simultaneously, then the water body distribution map corresponding to that critical region to be reconstructed is discarded.

[0148] In one embodiment, the processor 902 is further configured to:

[0149] Acquire the set of raw optical remote sensing images and the set of raw radar remote sensing images for the same area;

[0150] The polarization channel is extracted from each original radar remote sensing image in the original radar remote sensing image set to obtain the polarization index.

[0151] The Otsu algorithm is used to determine the water extraction threshold from the polarization index;

[0152] Based on the water body extraction threshold, radar water body distribution maps for each time phase are generated, resulting in the radar water body distribution map set;

[0153] Masking and water pixel extraction are performed on each original optical remote sensing image in the original optical remote sensing image set to obtain optical water distribution maps for each time phase.

[0154] Based on the optical water body distribution maps of each time phase, the set of optical water body distribution maps is obtained.

[0155] Figure 10 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 10 As shown, the electronic device may include: a processor 1010, a communications interface 1020, a memory 1030, and a communications bus 1040, wherein the processor 1010, the communications interface 1020, and the memory 1030 communicate with each other through the communications bus 1040. The processor 1010 can call logical instructions in the memory 1030 to execute a method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction. This method includes: acquiring a sequence of water body distribution maps to be reconstructed collected by an optical remote sensing sensor; for any water body distribution map in the sequence, finding the reference image with the smallest spatial structure difference between it and the water body distribution map in a reference image set, and using this reference image as the optimal reference water body map; wherein all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area; using the pixel categories in the optimal reference water body map, replacing missing pixels in the water body distribution map to be reconstructed corresponding to the optimal reference water body map to obtain a reconstructed water body distribution map; and using the reconstructed water body distribution map sequence corresponding to all water body distribution map sequences as the result of dynamic monitoring of surface water bodies.

[0156] Furthermore, the logical instructions in the aforementioned memory 1030 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0157] On the other hand, the present invention also provides a computer program product, which includes a computer program that 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 surface water dynamic monitoring method based on multi-source remote sensing and spatial similarity reconstruction provided by the above methods. The method includes: acquiring a sequence of water body distribution maps to be reconstructed collected by an optical remote sensing sensor; for any water body distribution map to be reconstructed in the sequence, finding the reference image with the smallest spatial structure difference between it and the water body distribution map to be reconstructed in a reference image set, and using it as the optimal reference water body map for the water body distribution map to be reconstructed; wherein all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area; using the pixel categories in the optimal reference water body map, replacing the missing regions in the water body distribution map to be reconstructed corresponding to the optimal reference water body map with pixels to obtain the reconstructed water body distribution map; and using the reconstructed water body distribution map sequence corresponding to all water body distribution map sequences to be reconstructed as the surface water dynamic monitoring result.

[0158] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the surface water dynamic monitoring method based on multi-source remote sensing and spatial similarity reconstruction provided by the above methods. The method includes: acquiring a sequence of water body distribution maps to be reconstructed collected by an optical remote sensing sensor; for any water body distribution map to be reconstructed in the sequence, searching in a reference image set for the reference image with the smallest spatial structure difference between it and the water body distribution map to be reconstructed, and using it as the optimal reference water body map for the water body distribution map to be reconstructed; wherein all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area; using the pixel categories in the optimal reference water body map, replacing the missing regions in the water body distribution map to be reconstructed corresponding to the optimal reference water body map with pixels to obtain the reconstructed water body distribution map; and using the reconstructed water body distribution map sequence corresponding to all water body distribution map sequences to be reconstructed as the surface water dynamic monitoring result.

[0159] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0160] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction, characterized in that, The method comprises the following steps: acquiring a sequence of water body distribution maps to be reconstructed collected by an optical remote sensing sensor; for any water body distribution map to be reconstructed in the sequence of water body distribution maps to be reconstructed, searching for a reference image in a reference image set that has the smallest spatial structure difference with the water body distribution map to be reconstructed, as an optimal reference water body map of the water body distribution map to be reconstructed; all reference images in the reference image set are images without missing pixels; each reference image is obtained by image fusion based on images collected by at least two remote sensing sensors for the same water area; using the pixel categories in the optimal reference water body map to replace missing areas in the water body distribution map to be reconstructed corresponding to the optimal reference water body map, to obtain a reconstructed water body distribution map; and using the sequence of reconstructed water body distribution maps corresponding to all water body distribution maps to be reconstructed as a ground water body dynamic monitoring result; the reference image set is constructed according to the following method: acquiring a set of optical water body distribution maps and a set of radar water body distribution maps for the same area; identifying an easy misjudgment area for each radar water body distribution map in the set of radar water body distribution maps; based on the proportion of the missing area in each optical water body distribution map in the set of optical water body distribution maps, or the proportion of the missing area in the corresponding radar water body distribution map in the easy misjudgment area, high-quality optical water body distribution maps are screened out; based on the spatial structure difference between the radar water body distribution maps in the set of radar water body distribution maps and the high-quality optical water body distribution maps, the radar water body distribution maps in the set of radar water body distribution maps are used to fill in the missing areas in the high-quality optical water body distribution maps, to obtain high-quality no-missing fusion water body distribution maps; and based on the high-quality no-missing fusion water body distribution maps, the reference image set is constructed.

2. The method according to claim 1, characterized in that, The spatial structure difference includes a symmetric difference; and searching for a reference image in the reference image set that has the smallest spatial structure difference with the water body distribution map to be reconstructed, as an optimal reference water body map of the water body distribution map to be reconstructed, comprises: calculating the symmetric difference between the water body distribution map to be reconstructed and all reference water body distribution maps in the reference image set, and searching for a reference water body distribution map with the smallest symmetric difference as the optimal reference water body map of the water body distribution map to be reconstructed.

3. The method according to claim 1, wherein, After the reference image set is constructed based on the high-quality no-missing fusion water body distribution maps, the following steps are further included: using the optical water body distribution maps in the set of optical water body distribution maps except the high-quality optical water body distribution maps as a set of water body distribution maps to be reconstructed; from each water body distribution map to be reconstructed in the set of water body distribution maps to be reconstructed, regions with a slope less than or equal to a preset slope threshold and a water body submergence frequency within a preset frequency range are selected as key regions according to the slope and the water body submergence frequency; missing areas in the key regions are filled in using reference images in the reference image set, to obtain reconstructed optical water body distribution maps.

4. The method according to claim 3, wherein, The filling of the missing area in the key area by using the reference image in the reference image set to obtain the reconstructed optical water body distribution map comprises: If the water body pixels and the non-water body pixels exist in the key area at the same time, the missing area in the key area is filled by using the image in the reference image set to obtain the reconstructed optical water body distribution map. If the water body pixels and the non-water body pixels do not exist in the key area at the same time, the water body distribution map corresponding to the key area is discarded.

5. The method according to claim 1, wherein, The acquisition of the optical water body distribution map set and the radar water body distribution map set for the same area comprises: An original optical remote sensing image set and an original radar remote sensing image set for the same area are acquired. A polarization channel is extracted from each original radar remote sensing image in the original radar remote sensing image set to obtain a polarization index. A water body extraction threshold is determined from the polarization index by using the Otsu algorithm. The radar water body distribution map of each phase is generated based on the water body extraction threshold to obtain the radar water body distribution map set. Each original optical remote sensing image in the original optical remote sensing image set is subjected to mask removal and water body pixel extraction to obtain the optical water body distribution map of each phase. The optical water body distribution map set is obtained based on the optical water body distribution map of each phase.

6. A dynamic monitoring system for surface water bodies based on multi-source remote sensing and spatial similarity reconstruction, characterized in that, The system comprises: An optical remote sensing sensor is configured to collect a sequence of optical water body distribution maps for a target water area. A processor is configured to execute the method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction according to any one of claims 1 to 5.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The processor executes the computer program to implement the method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction according to any one of claims 1 to 5.

9. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction according to any one of claims 1 to 5. The computer program is executed by the processor to implement the method for dynamic monitoring of surface water bodies based on multi-source remote sensing and spatial similarity reconstruction according to any one of claims 1 to 5.

Images