Small beach shoreline accurate extraction method based on bare soil-water body double-index cooperation

Abstract

The invention relates to the technical field of remote sensing image processing, in particular to a small beach shoreline accurate extraction method and system based on bare soil-water body double-index collaboration, and the method comprises the following steps: obtaining a multi-temporal Landsat surface reflectance image set which covers a target area and is subjected to radiometric calibration and wave band unified naming; obtaining a clean image set; generating a binary beach mask based on the surface reflectance of each pixel in the clean image set in a blue band, a green band, a red band, a near-infrared band and a short-wave infrared band; performing morphological closing operation on the binary beach mask to fill the fine holes to obtain a continuous mask; vectorizing and geometrically simplifying the continuous mask to obtain an initial beach face vector; and performing post-processing operation on the initial beach face vector to obtain a final beach boundary vector. According to the technical scheme, the beach can be effectively distinguished from adjacent water bodies, vegetation, buildings and other ground objects, and false lifting and missing lifting are remarkably reduced.

Description

Technical Field

[0001] This disclosure relates to the field of remote sensing image processing technology, specifically to a method for accurate extraction of small-scale beach shorelines based on the synergistic interaction of bare soil and water bodies. Background Technology

[0002] As important geomorphic units and ecological transition zones in river systems, the accurate extraction of boundary information of islands and shoals (such as river islands and shoals) is of great significance for river morphology analysis, ecological environment assessment, land spatial planning, and flood control. Although traditional manual visual interpretation methods have high accuracy, they are time-consuming, labor-intensive, and highly subjective, making it difficult to meet the needs of large-scale, long-term dynamic monitoring.

[0003] In recent years, with the development of remote sensing technology, automated shoreline extraction methods based on satellite imagery have become a research hotspot. Among them, the single-exponential thresholding method and machine learning / deep learning methods are two mainstream technical routes. However, when faced with small, complex, and spectrally mixed sandbars, these methods still have obvious limitations: the single-exponential thresholding method is difficult to effectively distinguish sandbars from adjacent bare soil, buildings, and other features, which can easily lead to boundary misjudgment or the omission of small sandbars; while machine learning methods can integrate multiple features, they usually rely on a large number of labeled samples, the model's generalization ability is greatly affected by the training data, and the extraction results are often pixel-level classification maps with jagged boundaries, requiring complex post-processing to obtain smooth vector shorelines.

[0004] Furthermore, existing automated methods are mostly based on single-phase imagery analysis, which is susceptible to cloud cover, water level fluctuations, or transient flooding conditions within that phase, resulting in the extracted beach areas lacking typicality or stability. Regarding the vectorization and optimization of the extracted results, most studies stop at binary raster output or simple vector conversion, lacking a systematic morphological and topological post-processing workflow tailored to the polygonal characteristics of the beaches. This leads to the final vector output potentially containing numerous fragmented polygons, internal holes, and irregular boundaries, affecting subsequent spatial analysis and practical applications. Summary of the Invention

[0005] To address the problems in related technologies, this disclosure provides a method for accurate extraction of small-scale beach shorelines based on the synergistic effect of bare soil and water body dual indices.

[0006] In a first aspect, embodiments of this disclosure provide a method for accurately extracting the shoreline of small sandbars based on the synergistic effect of bare soil and water body dual indices, comprising the following steps:

[0007] Acquire a multi-temporal Landsat surface reflectance image set covering the target area, after radiometric calibration and unified band naming. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and shortwave infrared band, as well as the code of each pixel in the quality assessment band. Based on the code of the quality assessment band and the preset blue band reflectivity threshold, cloud and invalid pixel masks are applied to the image set to obtain a clean image set. Based on the surface reflectance of each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, the bare soil index (BSI) and normalized water index (NDWI) are calculated pixel by pixel, and a binary beach mask that satisfies the condition BSI>0 and NDWI≤0 is generated. Perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; The continuous mask is vectorized and geometrically simplified to obtain the initial beach surface vector; Post-processing operations are performed on the initial beach surface vector to obtain the final beach boundary vector; wherein, the post-processing operations include one or more of the following: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

[0008] According to embodiments of this disclosure, the radiometric calibration and unified band naming process includes: uniformly converting the reflectance values ​​of optical bands corresponding to different Landsat satellite sensors to the surface reflectance range of 0-1, and uniformly naming the surface reflectances of optical bands corresponding to different Landsat satellite sensors as Blue, Green, Red, NIR, and SWIR according to their physical meaning; wherein, Blue is the surface reflectance of the blue band, Green is the surface reflectance of the green band, Red is the surface reflectance of the red band, NIR is the surface reflectance of the near-infrared band, and SWIR is the surface reflectance of the shortwave infrared band.

[0009] According to embodiments of this disclosure, the cloud and invalid pixel masking of the image set based on the code of the quality assessment band and a preset blue band reflectance threshold includes: Mask the bits that identify clouds and cloud shadows in the decoding quality assessment band; Pixels with a surface reflectance of less than 0.1 in the blue band are identified as invalid pixels and masked.

[0010] According to an embodiment of this disclosure, the morphological closing operation uses a 3×3 pixel circular structure kernel to perform a first expansion followed by erosion operation to fill holes with an area of ​​less than 9 pixels.

[0011] According to embodiments of this disclosure, the vectorization is implemented using the reduceToVectors function on the Google Earth Engine platform, where the scale parameter is set to 30 meters and the maxPixels parameter is set to 1×10. 8 .

[0012] According to embodiments of this disclosure, the geometric simplification employs the Douglas-Peucker algorithm.

[0013] According to embodiments of this disclosure, the adjacent surface fusion operation in the post-processing operation is performed according to the image acquisition year and quarter field.

[0014] Secondly, embodiments of this disclosure provide a small-scale beach shoreline precision extraction system based on bare soil-water body dual-index synergy, comprising: The acquisition module is configured to acquire a multi-temporal Landsat surface reflectance image set covering the target area, which has been radiometrically calibrated and uniformly named by bands. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and shortwave infrared band, as well as the code of each pixel in the quality assessment band. The masking module is configured to perform cloud and invalid pixel masking on the image set based on the code of the quality assessment band and a preset blue band reflectivity threshold to obtain a clean image set; The binary beach mask generation module is configured to calculate the bare soil index (BSI) and normalized water index (NDWI) for each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, and generate a binary beach mask that satisfies the conditions of BSI>0 and NDWI≤0. A continuous mask generation module is configured to perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; The initial beach surface generation module is configured to vectorize and geometrically simplify the continuous mask to obtain the initial beach surface vector; The beach boundary generation module is configured to perform post-processing operations on the initial beach surface vector to obtain the final beach boundary vector; wherein the post-processing operations include one or more of the following operations: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

[0015] Thirdly, embodiments of this disclosure provide an electronic device including a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method as described in any of the first aspects.

[0016] Fourthly, this disclosure provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the method as described in any of the first aspects.

[0017] The technical effects provided by the embodiments of this disclosure may include the following beneficial effects: According to the technical solution provided in this disclosure, a method for accurately extracting the shoreline of small sandbars based on the dual-index synergy of bare soil and water includes the following steps: acquiring a multi-temporal Landsat surface reflectance image set covering the target area, after radiometric calibration and unified band naming, wherein the image set contains the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band, and shortwave infrared band, and the code of each pixel in the quality assessment band; performing cloud and invalid pixel masking on the image set based on the code of the quality assessment band and a preset blue band reflectance threshold to obtain a clean image set; and performing cloud and invalid pixel masking on each pixel in the clean image set. For surface reflectance in the blue, green, red, near-infrared, and shortwave infrared bands, calculate the bare soil index (BSI) and normalized water index (NDWI) pixel by pixel, and generate a binary beach mask that satisfies the condition BSI > 0 and NDWI ≤ 0. Perform morphological closing operations on the binary beach mask to fill small holes to obtain a continuous mask. Vectorize the continuous mask and perform geometric simplification to obtain an initial beach surface vector. Perform post-processing operations on the initial beach surface vector to obtain the final beach boundary vector. The post-processing operations include one or more of the following: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

[0018] The aforementioned technical solution, by synergistically utilizing the Bare Soil Index (BSI) and the Normalized Difference Water Index (NDWI) and setting a dual-condition judgment rule of "BSI>0 and NDWI≤0", can effectively distinguish between sandbars and adjacent water bodies, vegetation, buildings, and other land features, significantly reducing false positives and false negatives. It is particularly suitable for sandbar areas with complex shapes, small areas, and where traditional single-index methods are difficult to accurately extract. Morphological closing operations are used to process the binary mask, effectively filling in small holes and breaks within the sandbar caused by sensor noise or mixed pixels. Subsequent vector post-processing operations, such as geometric simplification and bidirectional buffer smoothing, can generate continuous, smooth shoreline vectors that conform to visual perception and actual geological features, overcoming the shortcomings of "jagged" extraction results based on pixel boundaries. Using multi-temporal Landsat surface reflectance imagery as input data, and combining quality assessment bands and blue band reflectance thresholds for cloud and invalid pixel masking, this method maximizes the acquisition of effective observation data and reduces interference from factors such as clouds, shadows, and sensor anomalies in a single temporal phase. This ensures the reliability and consistency of the beach extraction results under different seasons and hydrological conditions. Through a series of automated post-processing operations, including fragment removal, in-plane void elimination, and adjacent surface fusion, the method effectively eliminates small noisy polygons in the extraction results, integrates spatially adjacent independent beach surfaces, and generates clean, complete, and directly usable final beach boundary vector results for GIS analysis and mapping, significantly reducing the workload of manual editing. Based on publicly available Landsat imagery, this method has a clear workflow, and thresholds and parameters can be adjusted according to different geographical environments. It is easily adaptable and applicable to automated mapping and dynamic monitoring of beaches or similar wet-dry transition zones in other rivers, lakes, and coastal areas.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0020] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments, taken in conjunction with the accompanying drawings.

[0021] Figure 1 A flowchart illustrating a method for precise extraction of small sandbar shorelines based on bare soil-water body dual-index synergy according to an embodiment of the present disclosure is shown.

[0022] Figure 2 This diagram illustrates a comparison between extracting shoals based on a bare soil-water dual index and a normalized water index, according to embodiments of the present disclosure.

[0023] Figure 3 A schematic diagram showing the comparison effect before and after post-processing of the initial beach surface vector according to an embodiment of the present disclosure is provided.

[0024] Figure 4 A structural block diagram of a small-scale beach shoreline precision extraction system based on bare soil-water body dual-index synergy, according to an embodiment of the present disclosure, is shown.

[0025] Figure 5 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0026] Figure 6 A schematic diagram of the structure of a computer system suitable for implementing the method according to embodiments of the present disclosure is shown. Detailed Implementation

[0027] In the following, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings to enable those skilled in the art to readily implement them. Furthermore, for clarity, portions unrelated to the description of exemplary embodiments have been omitted from the drawings.

[0028] In this disclosure, it should be understood that terms such as “comprising” or “having” are intended to indicate the presence of features, figures, steps, behaviors, components, parts or combinations thereof disclosed in this specification, and are not intended to exclude the possibility of the presence or addition of one or more other features, figures, steps, behaviors, components, parts or combinations thereof.

[0029] It should also be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0030] For ease of understanding, the following explains some key terms in this embodiment: The Landsat Surface Reflectance Image Set refers to the collection of Earth's surface reflectance data acquired by Landsat series satellite sensors, after atmospheric correction and radiometric calibration. This image set typically contains data in multiple spectral bands, such as blue, green, red, near-infrared, and shortwave infrared bands, as well as quality assessment band codes for evaluating data quality.

[0031] The Bare Soil Index (BSI) is a remote sensing index used to identify and quantify bare soil cover on land surfaces. Calculated using a combination of specific spectral bands, this index effectively distinguishes exposed soil from other land cover types. Its values ​​typically vary within a certain range, with positive values ​​generally indicating bare soil areas.

[0032] The Normalized Difference Water Index (NDWI) is a remote sensing index used to identify and quantify surface water cover. Calculated using reflectance in the near-infrared and green bands, it effectively distinguishes between water and non-water bodies. Its value typically ranges from -1 to 1, with negative values ​​often indicating non-water areas.

[0033] A binary beach mask is an image in which identified beach areas in remote sensing are labeled with a specific value (e.g., 1), while African beach areas are labeled with another value (e.g., 0). This mask simplifies complex beach information into a binary form, facilitating subsequent morphological processing and vectorization.

[0034] Morphological closing is an image processing operation typically used to fill in small holes and broken areas in an image. This operation involves two basic steps: dilation followed by erosion, which can smooth image boundaries and eliminate noise.

[0035] Vectorization refers to the process of converting raster image data into vector graphic data. In beach extraction, this process transforms the beach region boundary in a binary mask into vector geometric objects composed of points, lines, and surfaces, facilitating geometric analysis and management.

[0036] Geometric simplification refers to thinning vector graphics to reduce the number of vertices, thereby reducing data complexity while preserving key morphological features. This operation helps eliminate jagged edges in the extracted results, making shorelines smoother.

[0037] Post-processing refers to a series of operations performed on the initial extracted results to further optimize the morphology and topological relationships of the shoal boundaries. These operations may include removing small-area debris, eliminating in-plane voids, smoothing boundaries, and merging adjacent regions to improve the accuracy and continuity of the final results.

[0038] Figure 1 A flowchart illustrating a method for precise extraction of small sandbar shorelines based on bare soil-water body dual-index synergy according to an embodiment of the present disclosure is shown.

[0039] like Figure 1 As shown, the method for accurate extraction of small sandbar shorelines based on the synergistic effect of bare soil and water body dual indices includes the following steps: Step S110: Obtain a multi-temporal Landsat surface reflectance image set covering the target area, after radiometric calibration and unified band naming. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and short-wave infrared band, as well as the code of each pixel in the quality assessment band. Step S120: Based on the code of the quality assessment band and the preset blue band reflectivity threshold, perform cloud and invalid pixel masking on the image set to obtain a clean image set; Step S130: Based on the surface reflectance of each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band respectively, calculate the bare soil index BSI and normalized water index NDWI for each pixel, and generate a binary beach mask that satisfies the conditions of BSI>0 and NDWI≤0. Step S140: Perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; Step S150: Vectorize the continuous mask and perform geometric simplification to obtain the initial beach surface vector; Step S160: Perform post-processing operations on the initial beach surface vector to obtain the final beach boundary vector; wherein, the post-processing operations include one or more of the following operations: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

[0040] Existing technologies for extracting beach boundaries using remote sensing imagery primarily rely on water body indices such as the Normalized Difference Water Index (NDWI) and the Modified Normalized Difference Water Index (MNDWI) for threshold segmentation, distinguishing between water bodies and non-water bodies to retrieve the shoreline. However, this method suffers from the following core technical problems: Inaccurate identification of sandbar boundaries: Traditional water body indices struggle to effectively distinguish between exposed surfaces such as sandy beaches and mudflats. This leads to small-scale sandbar boundaries (less than 30 square kilometers) being easily misidentified as water bodies or inland areas, resulting in the "missing" or incorrect identification of sandbars. This is because water body indices primarily focus on water characteristics and lack the ability to differentiate between exposed surfaces, failing to fundamentally reconcile the index model with the characteristics of bare land features.

[0041] Unsmooth shoreline: Due to the spatial resolution of 30 meters in Landsat imagery, the extracted shoreline appears jagged and cannot accurately and smoothly reflect the natural tortuous shape of the sandbar, making it difficult to meet the requirements of high-precision applications.

[0042] Data voids and discontinuities: Sensor failures (such as SLC-off in Landsat 7) can cause stripe voids and missing data in images. Traditional interpolation or stitching methods have limited repair effects and often leave pseudo voids or discontinuous areas in the beach surface data, affecting the integrity and usability of the data.

[0043] These issues collectively mean that existing methods cannot meet the requirements of high-precision applications in terms of accuracy and continuity in the extraction of small islands.

[0044] This disclosure aims to solve problems such as inaccurate extraction of small sandbars, jagged shorelines, and data gaps. By combining automated preprocessing with Google Earth Engine (GEE) platform and refined post-processing with ArcGIS software, it achieves accurate and smooth extraction of sandbar shorelines.

[0045] GEE platform for automated generation of beach masks: First, a multi-temporal Landsat surface reflectance image set covering the target area is acquired. This image set needs to undergo radiometric calibration and standardized band naming to ensure comparability of data from different sensors and time periods. On the GEE platform, multi-source remote sensing image data from Landsat 5 / 7 / 8 and other sources are integrated, filtered according to user-defined spatiotemporal ranges, and standardized band naming and radiometric calibration are performed to ensure data consistency. For example, appropriate calibration parameters can be manually selected to calibrate the original images, and the band names from different sensors can be manually standardized to standard names, thereby obtaining image data containing surface reflectance in blue, green, red, near-infrared, and shortwave infrared bands, as well as quality assessment band codes.

[0046] Secondly, cloud and invalid pixel masking is performed on the acquired image set to obtain a clean image set. This masking process can be implemented in various ways, such as visually identifying clouds and cloud shadow areas in the image and manually drawing the mask, or setting a fixed blue band reflectance threshold, marking all pixels below the threshold as invalid and masking them.

[0047] Next, based on surface reflectance data from the clean image set, the Bare Soil Index (BSI) and Normalized Difference Water Index (NDWI) are calculated pixel-by-pixel. BSI can be calculated using a combination of red, shortwave infrared, green, and blue bands, while NDWI can be calculated using a combination of green and near-infrared bands. Subsequently, by setting the condition that BSI > 0 and NDWI ≤ 0, pixels meeting the condition are identified as beach areas, and a binary beach mask is generated.

[0048] Then, a morphological closing operation is performed on the generated binary beach mask. This operation aims to fill the small holes in the mask and connect broken areas, thereby obtaining a more continuous mask. For example, a simple square structuring element can be used to perform an expansion operation on the binary mask, followed by an erosion operation, to fill the holes and improve the integrity of the beach area.

[0049] Subsequently, the continuous mask is vectorized and geometrically simplified to obtain the initial beach surface vector. Vectorization converts the raster-style mask into polygonal vector data. For example, the automatic tracking function in image processing software can convert the boundaries of a binary image into vector lines or surfaces. Geometric simplification removes redundant vertices on the vector boundaries, making the boundary lines smoother. For example, a simplification algorithm based on a distance threshold can be used to merge or delete vertices that are too close together.

[0050] ArcGIS software fine-grained post-processing: Finally, post-processing operations are performed on the initial beach surface vector to obtain the final beach boundary vector. This post-processing aims to further optimize the accuracy and topological relationships of the beach boundaries. For example, small surface fragments can be manually checked and deleted, or voids within the surface can be eliminated through manual editing. Additionally, boundaries can be smoothed through simple buffer operations, or adjacent beach surfaces can be manually merged.

[0051] By employing the dual-index synergy of the Bare Soil Index (BSI) and the Normalized Difference Water Index (NDWI), the boundaries between bare soil and water can be effectively distinguished, avoiding misjudgment of small-scale sandbar boundaries. Combining morphological closing operations, vectorization and geometric simplification, along with various post-processing operations, data gaps are effectively filled, jagged edges in the extracted results are eliminated, and the topological relationships of the sandbar surface are optimized, thereby significantly improving the accuracy, continuity, and natural smoothness of small-scale sandbar boundary extraction.

[0052] The following example will provide a more detailed explanation of the above technical solution: A research team needed to automate and accurately extract the shorelines of small sandbars with an area of ​​less than 30 km² in a large river system to monitor their dynamic changes. The sandbar topography in this area is complex, including exposed sandy beaches and mudflats, and is often affected by cloud cover and image quality issues.

[0053] First, on the GEE platform, the study area was set as a specific sandbar in the middle and lower reaches of the ×× River, with the time range set from 1990 to 2024. The system automatically called and integrated multi-source remote sensing image datasets such as Landsat 5, Landsat 7, and Landsat 8 within this spatiotemporal range.

[0054] Next, these images undergo automated preprocessing: 1. Unified band naming and radiometric calibration: Unify the band naming of all Landsat images and apply radiometric calibration functions (such as applyScaleFactors) to convert the raw digital quantization values ​​into surface reflectance to ensure data consistency and comparability.

[0055] 2. Cloud Masking and Invalid Value Filtering: Based on the image quality assessment band code and blue band reflectance threshold, the cloudRemoval function automatically identifies and removes clouds, cloud shadows, and other abnormal pixels in the image to obtain high-quality usable data.

[0056] 3. Construction of the BSI and NDWI dual-exponential collaborative model: The Bare Soil Index (BSI) and Normalized Difference Water Index (NDWI) for each image are calculated using the following formulas.

[0057] =(( + ) ( + )) / (( + )+( + )); =( ) / ( + ); Wherein, Blue represents the surface reflectance in the blue band, Green represents the surface reflectance in the green band, Red represents the surface reflectance in the red band, NIR represents the surface reflectance in the near-infrared band, and SWIR represents the surface reflectance in the shortwave infrared band.

[0058] Based on the set beach identification conditions (BSI>0 and NDWI≤0), a preliminary binary beach mask is generated.

[0059] For example, in a certain quarter, after this step, the initial number of beach surface components generated was approximately 1,153.

[0060] 4. Morphological Closure Operation for Hole Filling: Perform a 3×3 pixel window morphological closure operation on the generated binary beach mask to fill the small holes in the mask, making the beach area more complete.

[0061] 5. Vectorization and Geometric Simplification: The processed raster beach mask is converted into a vector surface using the reduceToVectors function, and then polygonal geometric simplification with a 30-meter tolerance is performed to reduce redundant vertices, generating preliminary beach surface vector data, i.e., the initial beach surface vector.

[0062] The initial beach surface vector generated by the GEE platform is imported into ArcGIS software for further refinement and post-processing. 1. Small-area surface fragment removal: Using ArcGIS's select-by-attribute tool, filter and delete small surface fragments with an area less than 0.1 square kilometers. For example, after this step, the number of beach surface components was reduced from 1153 to approximately 208, effectively eliminating noise.

[0063] 2. Eliminate voids within the surface: Using ArcGIS's Eliminate tool, with a threshold area ratio set to less than 80%, small voids within the sandbar surface are automatically merged into adjacent main surfaces, resolving the issue of discontinuity within the surface.

[0064] 3. Multi-ring buffer expansion and fusion: A positive 75-meter buffer is created to fill the gaps between surfaces caused by data gaps or resolution limitations, and to connect broken areas. Subsequently, a fusion tool is executed based on the year-quarter field (e.g., label attribute) to merge adjacent surface components from the same period, generating continuous beach surfaces.

[0065] 4. Secondary Fragment Removal and Reverse Buffer Contraction: The small-area surface fragment removal step is performed again to further eliminate any small-area anomalies that may remain after fusion. Finally, a negative 75-meter buffer contraction is applied to the fused beach surface to precisely offset the boundary extension caused by the previous expansion, while further smoothing the shoreline and restoring its original, natural shape.

[0066] Through the above steps, accurate, smooth, and continuous shoreline data of the sandbars in the middle and lower reaches of the ×× River are finally obtained. For example, the final number of sandbar surface components is 119, and the total area error is less than 3%, significantly improving the accuracy and quality of sandbar extraction.

[0067] Figure 2 This diagram illustrates a comparison between extracting sandbars based on a bare soil-water dual index and a normalized water index, according to embodiments of this disclosure. Figure 2 (a) in the image is a true-color composite of the original remote sensing image. Figure 2 (b) shows the results of beach extraction based on the bare soil-water dual index. In the figure, red represents land and blue represents water. This result can accurately identify small exposed beaches (such as mudflats and sandbars), avoiding misclassification of shallow water or shaded areas as land, and reducing missed extractions. The figure shows more complete boundaries, better continuity, and... Figure 2 The actual location of the shoal is highly consistent with that in (a) of the diagram. Figure 2 (c) shows the results of beach extraction based on the Normalized Difference Water Index (NDWI). In the figure, red represents land and blue represents water. Since NDWI mainly distinguishes between water and non-water, it cannot effectively identify the difference between bare soil and non-water bodies such as vegetation / buildings. In small beach areas, it is easy to misclassify shallow water or low reflectivity areas as land, or to misclassify dry mudflats with spectra close to water bodies as water, resulting in missed extractions. The extraction results have problems such as inaccurate boundaries, fragmentation, and morphological distortion.

[0068] Figure 3 A schematic diagram showing the comparison effect before and after post-processing of the initial beach surface vector according to an embodiment of the present disclosure is provided. Figure 3 (a) in the figure is a schematic diagram of the effect before post-processing. The beach extraction results show obvious strip-shaped missing data (blue circles in the figure), and the shoreline (blue crescent-shaped line in the figure) is obviously jagged and stepped. Figure 3 (b) in the figure is a schematic diagram of the post-processing effect. The strip voids are effectively filled, and the beach surface is continuous and complete. After multi-level buffer and geometric simplification operations, the shoreline (red crescent-shaped line in the figure) becomes smooth, continuous and closer to the real landform, while retaining the main geometric features.

[0069] In some embodiments of this disclosure, it is necessary to acquire a multi-temporal Landsat surface reflectance image set covering the target area, which has been radiometrically calibrated and uniformly named by bands. However, due to the differences between different Landsat satellite sensors, if uniform radiometric calibration and band naming are not performed, inconsistencies may occur between image data acquired at different times or by different sensors, thereby affecting the accuracy and reliability of subsequent beach extraction.

[0070] In response, this application further proposes a specific implementation method for the radiometric calibration and unified band naming process. The radiometric calibration and unified band naming process includes: uniformly converting the reflectance values ​​of optical bands corresponding to different Landsat satellite sensors to surface reflectance values ​​within the range of 0-1, and uniformly naming the surface reflectance values ​​of the optical bands corresponding to different Landsat satellite sensors as Blue, Green, Red, NIR, and SWIR according to their physical meaning; where Blue represents the surface reflectance of the blue band, Green represents the surface reflectance of the green band, Red represents the surface reflectance of the red band, NIR represents the surface reflectance of the near-infrared band, and SWIR represents the surface reflectance of the shortwave infrared band.

[0071] Specifically, the radiometric calibration aims to eliminate physical differences in raw imagery data acquired by different Landsat satellite sensors (e.g., Landsat 5, Landsat 7, Landsat 8, Landsat 9) and convert them into standardized, comparable surface reflectance. This conversion typically involves atmospheric correction and sensor calibration to ensure data consistency in physical dimensions and numerical ranges. For example, for Landsat data, this can be achieved using its provided calibration coefficients and atmospheric correction models, or by employing other radiometric correction methods known in the art.

[0072] Meanwhile, the standardized band naming process ensures that subsequent algorithms can reference specific band data using consistent band names when processing multi-source data, thereby improving the algorithm's versatility and portability. For example, programming scripts map the original band names (such as Landsat 8's B2, B3, B4, B5, and B6) to the standardized names Blue, Green, Red, NIR, and SWIR. This explicit naming convention provides a clear definitional basis for subsequent index calculations based on these bands (such as the Bare Soil Index (BSI) and the Normalized Difference Water Index (NDWI).

[0073] Furthermore, the specific implementation of the radiometric calibration involves multiplying the SR band reflectance of Landsat 5 / 7 / 8 / 9 by 0.0000275 and subtracting 0.2. This linear transformation formula is determined according to the recommendations of Landsat data providers (such as USGS) or industry standards, and aims to eliminate systematic biases between different sensors, ensuring data consistency in physical dimensions and numerical ranges. Applying this linear transformation formula to the SR band reflectance value of each pixel during the data preprocessing stage achieves consistency across multi-sensor data.

[0074] By employing the aforementioned technical solutions, unified radiometric calibration and band naming of multi-temporal Landsat surface reflectance image sets can effectively eliminate data differences between different Landsat satellite sensors, ensuring the physical comparability and consistency of image data from different time phases. This lays a solid foundation for the subsequent accurate calculation of the Bare Soil Index (BSI) and the Normalized Difference Water Index (NDWI), thereby significantly improving the accuracy and reliability of automated and precise extraction of smooth shorelines on small sandbars and avoiding extraction errors caused by data inconsistencies.

[0075] In some embodiments of this disclosure, masking the image set with clouds and invalid pixels is a key step in obtaining a high-quality, clean image set. However, failure to accurately identify and effectively remove clouds, cloud shadows, and other low-quality data from the images may lead to deviations in the subsequent calculations of the bare soil index and normalized water index, thereby affecting the accurate extraction of the sandbar area.

[0076] In response, this application further proposes the method of applying cloud and invalid pixel masks to the image set based on the code of the quality assessment band and a preset blue band reflectance threshold, including: Mask the bits that identify clouds and cloud shadows in the decoding quality assessment band; Pixels with a surface reflectance of less than 0.1 in the blue band are identified as invalid pixels and masked.

[0077] Specifically, Quality Assessment (QA) bands are auxiliary bands in remote sensing imagery used to indicate the data quality of each pixel. They typically contain multiple bits, each bit or combination of bits representing specific quality information, such as the presence of clouds, cloud shadows, snow, or water bodies. By decoding these bits, areas in the image affected by clouds and cloud shadows can be accurately identified and located. For example, for Landsat satellite imagery, its QA bands (such as the QA_PIXEL band in Landsat 8 / 9) use bit encoding, and specific combinations of bits may represent "high-confidence clouds" or "high-confidence cloud shadows." By reading and parsing these bits, a binary mask can be generated, in which pixels identified as clouds or cloud shadows are marked as invalid and thus excluded in subsequent processing.

[0078] Meanwhile, blue band surface reflectance is typically low in water bodies, shadows, or very dark areas. Pixels with blue band surface reflectance less than 0.1 are classified as invalid pixels, effectively removing low-quality or anomalous data caused by atmospheric scattering, water bodies, deep shadows, or sensor noise. In practice, an empirical threshold, such as 0.1, is set. For each pixel in the image, its blue band surface reflectance value is checked. If this value is less than 0.1, the pixel's data quality is considered unreliable or does not represent the true surface information, and it is marked as an invalid pixel. This threshold can be adjusted based on the actual conditions of the target area and experience, but 0.1 is a commonly used and relatively robust empirical value that can effectively exclude most low-quality pixels, such as those in deep water, permanent shadows, or sensor-saturated areas.

[0079] Through the above technical solutions, this application can more precisely and accurately identify and remove interference information in the original image. Specifically, the decoding quality assessment band ensures that clouds and cloud shadow areas are effectively excluded, avoiding the negative impact of these interference factors on the calculation of the Bare Soil Index (BSI) and the Normalized Difference Water Index (NDWI). Simultaneously, further screening using the blue band reflectance threshold can remove low-quality pixels caused by water bodies, deep shadows, or sensor anomalies, thereby obtaining a cleaner image set. This dual and complementary masking strategy significantly improves the accuracy and reliability of subsequent beach identification, ensuring the accuracy of beach boundary extraction and avoiding misjudgments or omissions due to data quality issues.

[0080] In some embodiments of this disclosure, a binary beach mask can be initially generated by calculating the Bare Soil Index (BSI) and Normalized Difference Water Index (NDWI) pixel by pixel and combining them with specific conditions. However, due to factors such as noise in remote sensing images, spectral obfuscation of ground features, or classification errors, the generated binary beach mask may contain some small, discontinuous holes or gaps. If these small holes are not effectively processed, they will affect the accuracy and continuity of subsequent beach surface vectorization, resulting in insufficiently smooth beach boundaries or the presence of unnecessary internal voids, thereby reducing the accuracy of beach extraction.

[0081] To address this, this application further proposes performing a morphological closing operation on the binary beach mask to fill small holes, thereby obtaining a continuous mask. Specifically, the morphological closing operation uses a 3×3 pixel circular core and performs a first expansion followed by erosion operation to fill holes with an area of ​​less than 9 pixels.

[0082] Morphological closing is an image processing technique whose main function is to fill small holes and connect broken regions in a binary image. This operation is achieved by first dilating the image and then eroding the resulting image. Dilation expands the foreground region, filling small holes and connecting nearby breaks; erosion shrinks the foreground region, restoring the boundary expansion caused by dilation, while removing small protrusions to maintain the overall structure of the image.

[0083] A structural kernel is a template used in morphological operations to define the operational neighborhood. Using a 3×3 pixel circular structural kernel means that during dilation and erosion operations, pixels within a circular neighborhood with a radius of approximately one pixel centered on the current pixel are considered. The advantage of a circular structural kernel lies in its isotropy, ensuring similar processing effects in all directions and avoiding the introduction of directional biases, which is crucial for maintaining the smoothness of the natural morphology of sandbars. The relatively small 3×3 size is suitable for processing fine holes and gaps in images, effectively filling them while avoiding over-smoothing or altering the shape of larger features.

[0084] Perform a single expansion-then-erosion operation. The expansion operation enlarges the foreground (shoal) region in the binary beach mask, filling small voids and connecting nearby fractures. Subsequently, the erosion operation shrinks the expanded foreground region, removing minor protrusions created by the expansion and restoring the original shape of the beach boundary while preserving the filled voids. This combination of single operations efficiently handles minor defects and avoids overprocessing that could distort the beach boundary.

[0085] In binary beach masking, holes smaller than 9 pixels are typically considered minor defects caused by noise, inaccurate classification, or data acquisition errors. By employing a 3×3 pixel circular kernel for closing operations, these small holes can be accurately identified and filled, ensuring the internal continuity of the beach area while avoiding misfilling of larger, realistic African beach areas (such as water bodies), thus improving the accuracy of the beach masking.

[0086] By employing the aforementioned technical solution and performing a morphological closing operation involving expansion followed by erosion using a 3×3 pixel circular kernel, this application can accurately identify and fill small holes with an area of ​​less than 9 pixels in a binary beach mask. This parameter-specific closing operation avoids misfilling of larger, more realistic African beach areas, while effectively eliminating internal discontinuities caused by noise or classification errors. The resulting continuous mask exhibits higher internal integrity and boundary smoothness, providing more reliable foundational data for subsequent vectorization and geometric simplification steps, thereby significantly improving the automation and accuracy of smooth shoreline extraction for small beaches.

[0087] According to embodiments of this disclosure, the vectorization is implemented using the reduceToVectors function on the Google Earth Engine platform, where the scale parameter is set to 30 meters and the maxPixels parameter is set to 1×10. 8 .

[0088] Specifically, vectorization is the process of converting raster data into vector data, which is crucial for subsequent geospatial analysis. Google Earth Engine (GEE), as a cloud-based geospatial analysis platform, provides powerful capabilities for processing large-scale remote sensing imagery data. Vectorization within this platform fully leverages its distributed computing resources, effectively addressing the performance bottlenecks faced by traditional local computing when processing massive amounts of data. The `reduceToVectors` function is a core tool in the Google Earth Engine platform used to convert raster imagery into a set of vector features. This function identifies connected regions with the same pixel values ​​in the imagery and converts them into polygon vectors, thus achieving the conversion from pixel-level representation to geometric object representation. By calling this function, the conversion from continuous masks to initial landsat surface vectors can be automated, simplifying the operation process. The `scale` parameter defines the minimum resolution or sampling interval of the output vector geometry during vectorization. Setting it to 30 meters means that the vectorization operation will be sampled and processed at a spatial resolution of 30 meters. This helps ensure that the generated landsat surface vectors accurately reflect the detailed information of the original Landsat imagery, while avoiding data expansion and computational burden caused by excessively high resolution, thus achieving a balance between accuracy and efficiency. The `maxPixels` parameter limits the maximum number of pixels the `reduceToVectors` function processes in a single operation. This is a critical resource control parameter, especially when processing large-scale or high-resolution images. Setting `maxPixels` to 1×10... 8 This allows the system to process up to 100 million pixels in a single vectorization task, which helps prevent memory overflow or computation timeouts caused by processing too many pixels, thus ensuring the stability and reliability of the vectorization process.

[0089] By utilizing the powerful cloud computing capabilities of the GEE platform and employing a dedicated `reduceToVectors` function for vectorization, continuous masks can be efficiently converted into initial beach surface vectors. Specifically, setting the `scale` parameter to 30 meters ensures that the vectorization result matches the original resolution of the Landsat imagery, thus maintaining the detail accuracy of the beach boundaries while avoiding unnecessary data redundancy. Furthermore, by setting the `maxPixels` parameter to 1×10... 8This method effectively controls the computational resource consumption of a single vectorization operation, avoiding potential memory overflow or computation timeout issues when processing large-scale image data, and significantly improving the stability and efficiency of the vectorization process. This vectorization method, combining cloud computing platforms and parameter optimization, overcomes the limitations of traditional local vectorization processing of large-scale remote sensing images, ensuring the automation and accuracy of beach boundary extraction, and providing high-quality input data for subsequent geometric simplification and post-processing operations.

[0090] According to embodiments of this disclosure, the geometric simplification employs the Douglas-Peucker algorithm.

[0091] Specifically, the Douglas-Peucker algorithm is a widely used algorithm for simplifying line features. Its core idea is to iteratively remove vertices that contribute little to the overall shape, thereby significantly reducing the number of vertices while preserving the basic geometric features. The algorithm first establishes a virtual straight line between the start and end points of a line segment, then finds the point on the line segment farthest from this virtual line. If the distance from this farthest point to the virtual line is less than a preset tolerance, all intermediate points on the line segment are removed; conversely, if the distance is greater than the tolerance, the farthest point is retained, and the original line segment is divided into two parts using this point as the boundary. The same simplification process is recursively performed on these two parts until all points meet the tolerance condition. This method effectively removes fine jagged structures, making the line segment smoother. In this algorithm, the tolerance is a key parameter that defines the allowed degree of simplification. A larger tolerance value results in a higher degree of simplification, fewer retained vertices, and a smoother line segment; a smaller tolerance value results in a lower degree of simplification and more retained details. Setting the tolerance to 30 meters is based on the 30-meter spatial resolution of the original Landsat imagery and the 30-meter scale parameter set during vectorization. This setting ensures that the simplification process removes geometric details that are considered small or unnecessary at 30-meter resolution, while preserving the main morphological features of the beach shoreline at the 30-meter scale.

[0092] According to embodiments of this disclosure, the adjacent surface fusion operation in the post-processing operation is performed according to the image acquisition year and quarter field.

[0093] Specifically, adjacent surface fusion is a commonly used spatial data processing function in Geographic Information Systems (GIS). Its main function is to merge multiple spatially adjacent or overlapping areal features into a single isal feature. In the scenario of beach boundary extraction, this operation helps to eliminate small gaps caused by image noise, classification errors, or incomplete morphological processing, connecting parts that originally belonged to the same beach but were separated, thereby forming a more complete and continuous beach surface. Based on this, this application further limits the execution conditions of adjacent surface fusion, namely, it is performed according to the image acquisition year and quarter fields. This means that when performing the fusion operation, the system will not only consider the spatial proximity of the areal features, but also check the metadata information of the image data associated with these areal features, especially the image acquisition year and quarter. Fusion is only allowed when adjacent beach surface vectors all come from image data of the same year and the same quarter. For example, if one beach surface vector originates from an image from the first quarter of 2020, while its adjacent beach surface vector originates from an image from the second quarter of 2020 or the first quarter of 2021, they will not be merged even if they are spatially adjacent. This time-field-based constraint ensures the accuracy and temporal consistency of the fusion operation.

[0094] By combining the adjacent surface fusion operation with the year and quarter fields of image acquisition, the above technical solution effectively solves the erroneous fusion problem caused by time differences in multi-temporal beach data processing. This restrictive fusion strategy ensures that the extracted beach boundaries remain consistent across time, avoiding the erroneous merging of beach features from different years or seasons, thereby improving the accuracy and reliability of the final beach boundaries. This is of great significance for subsequent applications such as beach dynamic change analysis, area statistics, and ecological environment assessment, enabling a more realistic reflection of the beach evolution process.

[0095] In some embodiments of this disclosure, an initial beach surface vector can be obtained by masking the image set with clouds and invalid pixels, calculating the Bare Soil Index (BSI) and Normalized Difference Water Index (NDWI) to generate a binary beach mask, performing morphological closing operations, and vectorizing and geometrically simplifying the vector. However, in practical applications, this initial beach surface vector may still contain some small, irregular fragments, or there may be voids inside the surface that have not been completely eliminated. Even between adjacent beach surfaces, there may be tiny gaps or breaks, resulting in insufficient smoothness and continuity of the extracted beach boundaries, thus affecting the accuracy of the final beach morphology.

[0096] To address this, this application further proposes performing post-processing operations on the initial beach surface vector to obtain the final beach boundary vector. The post-processing operation includes a series of collaborative steps designed to refine and optimize the beach surface vector.

[0097] First, small-area surface fragment removal is performed. This operation aims to remove isolated surface features with excessively small areas from the initial beach surface vector. These surface features are often image noise or artifacts generated during processing. Specifically, by setting an area threshold, such as 0.1 square kilometers, all surface features are filtered, retaining only those with an area greater than or equal to the threshold, while deleting those with an area smaller than the threshold. This effectively cleanses the data and improves the accuracy of beach boundaries.

[0098] Secondly, an operation to eliminate in-surface voids is performed. This operation fills voids within the beach surface, ensuring the continuity and integrity of the beach surface representation. In practice, elimination tools in a Geographic Information System (GIS) can be used. By setting a threshold ratio of void area to the total area of ​​the surface component, such as less than 80%, relatively small internal voids can be identified and filled. This helps to treat the beach as a whole, avoiding misidentification of internal water bodies or small patches of African beach as voids.

[0099] Next, a multi-ring buffer expansion is performed. This operation involves extending the boundaries of beach surface components outward to connect adjacent, closely spaced beach surfaces and fill the small gaps between them. Specifically, a positive buffer zone, such as a 75-meter buffer zone, can be created, such that the boundary of each beach surface component expands outward by 75 meters. This expansion helps to bridge beach surface breaks caused by data resolution or initial processing errors and provides initial smoothing of the shoreline.

[0100] Building upon this, a field-based fusion operation is performed on surface components. This operation aims to merge adjacent or overlapping beach surface components belonging to the same period (e.g., the same year and quarter) into a larger, continuous beach surface. In practice, the year and quarter fields acquired from the imagery can be used as the fusion criteria, employing fusion tools to merge surface components with the same year-quarter attributes that are spatially adjacent or overlapping. This helps to integrate fragmented beach information, forming a more representative overall beach structure.

[0101] Subsequently, a second round of elimination of minor surface components is performed. After buffer zone expansion and surface component fusion, some new, smaller anomalous surface components or small fragments that were not completely removed before may reappear. This operation further cleans the fused result, screening and deleting surface components whose area does not meet the preset threshold (e.g., still less than 0.1 square kilometers) to ensure the purity of the final beach boundary.

[0102] Finally, reverse multi-ring buffer contraction is performed. This operation is the inverse of multi-ring buffer expansion, aiming to shrink the expanded and merged beach surface boundary inward to restore its original, more accurate shoreline morphology. Specifically, a negative buffer, such as a -75-meter buffer, can be created, causing the boundary of the beach surface component to shrink inward by 75 meters. This step maintains the connectivity and integrity of the beach surface while eliminating the over-smoothing effect of buffer expansion, making the extracted shoreline more realistic.

[0103] Through the aforementioned post-processing operations, this application effectively addresses the issues of fragmentation, voids, and breaks in the initial beach surface vector, significantly improving the accuracy and smoothness of beach boundary extraction. The small-area surface fragment removal and secondary surface component elimination steps effectively remove noise and artifacts, ensuring the purity of the beach data. The void elimination step guarantees the continuity of the beach surface, preventing misidentification of internal water bodies as beaches. The combined application of multi-ring buffer expansion and field-based surface component fusion connects and merges spatially adjacent but logically belonging-to-the-street broken surface components, forming a more complete beach entity, while simultaneously providing initial shoreline smoothing. Finally, the reverse multi-ring buffer contraction step accurately restores the original shoreline morphology of the beach while maintaining its overall connectivity, eliminating the geometric deviation introduced by buffer expansion. Overall, these synergistic steps ensure that the final beach boundary vector is not only continuous and complete but also smooth and highly accurate, providing a reliable data foundation for subsequent beach dynamic monitoring and management.

[0104] This application further specifies that the target area is the middle and lower reaches of the ×× River with an area of ​​less than 30 km². 2 Small sandbars.

[0105] The middle and lower reaches of the ×× River are renowned for their unique alluvial landforms, abundant sediment deposition, and significant seasonal hydrological variations. These characteristics have a crucial impact on the formation, evolution, and morphology of sandbars and shoals. Limiting the target area to the middle and lower reaches of the ×× River allows this method to fully utilize the common characteristics of sandbars and shoals in this region, thereby optimizing the algorithm's applicability. (Area less than 30 km²) 2 This further limits the scale of the extracted sandbars, emphasizing the characteristic of "small size". Setting this area threshold helps to focus attention on those sandbars that are small in scale, have varied shapes, and are sensitive to hydrological conditions, avoiding misclassification of large landmasses or vast bodies of water as sandbars. It also facilitates more refined identification and processing of small, scattered sandbars.

[0106] Figure 4A structural block diagram of a small-scale beach shoreline precision extraction system based on a bare soil-water body dual-index synergy, according to an embodiment of this disclosure, is shown. This system can be implemented as part or all of an electronic device through software, hardware, or a combination of both.

[0107] like Figure 4 As shown, the small-scale beach shoreline precision extraction system 400 based on bare soil-water body dual-index synergy includes: The acquisition module 410 is configured to acquire a multi-temporal Landsat surface reflectance image set covering the target area, which has been radiometrically calibrated and uniformly named by bands. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and short-wave infrared band, as well as the code of each pixel in the quality assessment band. The masking module 420 is configured to perform cloud and invalid pixel masking on the image set based on the code of the quality assessment band and a preset blue band reflectivity threshold to obtain a clean image set; The binary beach mask generation module 430 is configured to calculate the bare soil index (BSI) and normalized water index (NDWI) pixel by pixel based on the surface reflectance of each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, and generate a binary beach mask that satisfies the conditions of BSI>0 and NDWI≤0. A continuous mask generation module 440 is configured to perform a morphological closing operation on the binary beach mask to fill small holes and obtain a continuous mask. The initial beach surface generation module 450 is configured to vectorize and geometrically simplify the continuous mask to obtain the initial beach surface vector. The beach boundary generation module 460 is configured to perform post-processing operations on the initial beach surface vector to obtain the final beach boundary vector; wherein the post-processing operations include one or more of the following operations: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface fusion.

[0108] The specific technical details of the embodiments disclosed herein can be found in the above method embodiments, and will not be repeated here.

[0109] This disclosure also discloses an electronic device. Figure 5 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0110] like Figure 5 As shown, the electronic device includes a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method according to embodiments of the present disclosure.

[0111] The method for accurately extracting the shoreline of small sandbars based on the synergistic effect of bare soil and water body dual indices includes the following steps: Acquire a multi-temporal Landsat surface reflectance image set covering the target area, after radiometric calibration and unified band naming. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and shortwave infrared band, as well as the code of each pixel in the quality assessment band. Based on the code of the quality assessment band and the preset blue band reflectivity threshold, cloud and invalid pixel masks are applied to the image set to obtain a clean image set. Based on the surface reflectance of each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, the bare soil index (BSI) and normalized water index (NDWI) are calculated pixel by pixel, and a binary beach mask that satisfies the condition BSI>0 and NDWI≤0 is generated. Perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; The continuous mask is vectorized and geometrically simplified to obtain the initial beach surface vector; Post-processing operations are performed on the initial beach surface vector to obtain the final beach boundary vector; wherein, the post-processing operations include one or more of the following: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

[0112] According to embodiments of this disclosure, the radiometric calibration and unified band naming process includes: uniformly converting the reflectance values ​​of optical bands corresponding to different Landsat satellite sensors to the surface reflectance range of 0-1, and uniformly naming the surface reflectances of optical bands corresponding to different Landsat satellite sensors as Blue, Green, Red, NIR, and SWIR according to their physical meaning; wherein, Blue is the surface reflectance of the blue band, Green is the surface reflectance of the green band, Red is the surface reflectance of the red band, NIR is the surface reflectance of the near-infrared band, and SWIR is the surface reflectance of the shortwave infrared band.

[0113] According to embodiments of this disclosure, the cloud and invalid pixel masking of the image set based on the code of the quality assessment band and a preset blue band reflectance threshold includes: Mask the bits that identify clouds and cloud shadows in the decoding quality assessment band; Pixels with a surface reflectance of less than 0.1 in the blue band are identified as invalid pixels and masked.

[0114] According to an embodiment of this disclosure, the morphological closing operation uses a 3×3 pixel circular structure kernel to perform a first expansion followed by erosion operation to fill holes with an area of ​​less than 9 pixels.

[0115] According to embodiments of this disclosure, the vectorization is implemented using the reduceToVectors function on the Google Earth Engine platform, where the scale parameter is set to 30 meters and the maxPixels parameter is set to 1×10. 8 .

[0116] According to embodiments of this disclosure, the geometric simplification employs the Douglas-Peucker algorithm.

[0117] According to embodiments of this disclosure, the adjacent surface fusion operation in the post-processing operation is performed according to the image acquisition year and quarter field.

[0118] Figure 6 A schematic diagram of the structure of a computer system suitable for implementing the method according to embodiments of the present disclosure is shown.

[0119] like Figure 6 As shown, the computer system includes a processing unit that can execute various methods described above based on a program stored in a read-only memory (ROM) or a program loaded from a storage portion into a random access memory (RAM). The RAM also stores various programs and data required for the operation of the computer system. The processing unit, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0120] The following components are connected to the I / O interface: input sections including keyboards, mice, etc.; output sections including cathode ray tubes (CRTs), liquid crystal displays (LCDs), and speakers; storage sections including hard disks; and communication sections including network interface cards such as LAN cards and modems. The communication section performs communication processes via a network such as the Internet. Drives are also connected to the I / O interface as needed. Removable media, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive as needed so that computer programs read from them can be installed into the storage section as required. The processing unit can be implemented as a CPU, GPU, TPU, FPGA, NPU, etc.

[0121] In particular, according to embodiments of this disclosure, the methods described above can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program containing program code for performing the methods described above. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium.

[0122] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0123] The units or modules described in the embodiments of this disclosure can be implemented in software or programmable hardware. The described units or modules can also be located in a processor, and the names of these units or modules do not necessarily constitute a limitation on the unit or module itself.

[0124] In another aspect, this disclosure also provides a computer-readable storage medium, which may be a computer-readable storage medium included in the electronic device or computer system described above; or it may be a standalone computer-readable storage medium not assembled into a device. The computer-readable storage medium stores one or more programs, which are used by one or more processors to perform the methods described in this disclosure.

[0125] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A method for accurately extracting the shoreline of small sandbars based on the synergistic effect of bare soil and water body dual indices, characterized in that, Includes the following steps: Acquire a multi-temporal Landsat surface reflectance image set covering the target area, after radiometric calibration and unified band naming. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and shortwave infrared band, as well as the code of each pixel in the quality assessment band. Based on the code of the quality assessment band and the preset blue band reflectivity threshold, cloud and invalid pixel masks are applied to the image set to obtain a clean image set. Based on the surface reflectance of each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, the bare soil index (BSI) and normalized water index (NDWI) are calculated pixel by pixel, and a binary beach mask that satisfies the condition BSI>0 and NDWI≤0 is generated. Perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; The continuous mask is vectorized and geometrically simplified to obtain the initial beach surface vector; Post-processing operations are performed on the initial beach surface vector to obtain the final beach boundary vector; wherein, the post-processing operations include one or more of the following: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

2. The method according to claim 1, characterized in that, The radiometric calibration and unified band naming process includes: uniformly converting the reflectance values ​​of optical bands corresponding to different Landsat satellite sensors to the surface reflectance range of 0-1, and uniformly naming the surface reflectances of optical bands corresponding to different Landsat satellite sensors as Blue, Green, Red, NIR, and SWIR according to their physical meaning; where Blue is the surface reflectance of the blue band, Green is the surface reflectance of the green band, Red is the surface reflectance of the red band, NIR is the surface reflectance of the near-infrared band, and SWIR is the surface reflectance of the shortwave infrared band.

3. The method according to claim 1, characterized in that, The process of masking clouds and invalid pixels in the image set based on the code of the quality assessment band and a preset blue band reflectance threshold includes: Mask the bits that identify clouds and cloud shadows in the decoding quality assessment band; Pixels with a surface reflectance of less than 0.1 in the blue band are identified as invalid pixels and masked.

4. The method according to any one of claims 1 to 3, characterized in that, The morphological closing operation uses a 3×3 pixel circular kernel to perform an expansion followed by erosion operation to fill holes with an area of ​​less than 9 pixels.

5. The method according to claim 4, characterized in that, The vectorization is implemented using the `reduceToVectors` function on the Google Earth Engine platform, where the `scale` parameter is set to 30 meters and the `maxPixels` parameter is set to 1×10. 8 .

6. The method according to claim 5, characterized in that, The geometric simplification employs the Douglas-Peucker algorithm.

7. The method according to claim 1, characterized in that, The adjacent surface fusion operation in the post-processing operation is performed according to the year and quarter fields of image acquisition.

8. A small-scale precise shoreline extraction system for sandbars based on the synergistic effect of bare soil and water body dual indices, characterized in that, include: The acquisition module is configured to acquire a multi-temporal Landsat surface reflectance image set covering the target area, which has been radiometrically calibrated and uniformly named by bands. The image set includes the surface reflectance of each pixel in the blue band, green band, red band, near-infrared band and shortwave infrared band, as well as the code of each pixel in the quality assessment band. The masking module is configured to perform cloud and invalid pixel masking on the image set based on the code of the quality assessment band and a preset blue band reflectivity threshold to obtain a clean image set; The binary beach mask generation module is configured to calculate the bare soil index (BSI) and normalized water index (NDWI) for each pixel in the clean image set in the blue band, green band, red band, near-infrared band and short-wave infrared band, and generate a binary beach mask that satisfies the conditions of BSI>0 and NDWI≤0. A continuous mask generation module is configured to perform a morphological closing operation on the binary beach mask to fill the small holes and obtain a continuous mask; The initial beach surface generation module is configured to vectorize and geometrically simplify the continuous mask to obtain the initial beach surface vector; The beach boundary generation module is configured to perform post-processing operations on the initial beach surface vector to obtain the final beach boundary vector; wherein the post-processing operations include one or more of the following operations: fragment removal, in-plane void elimination, bidirectional buffer smoothing, and adjacent surface merging.

9. An electronic device, characterized in that, It includes a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method of any one of claims 1-7.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by a processor, the computer instructions implement the method described in any one of claims 1-7.

Images