Coastline extraction method and device

Abstract

The invention provides a coastline extraction method and device. The method provided by the invention comprises the following steps: acquiring remote sensing image data of a target coast; determining a target wave band based on the difference of the spectral characteristics of the mountain shadow region and the water body region under different wave bands; extracting a plurality of pixels of the remote sensing image data, and calculating saturation based on pixel values of the plurality of pixels; extracting pixels of the remote sensing image data under the target wave band, and calculating a mountain shadow index of each pixel based on a difference value of pixel values of the pixels of different wave bands, a saturation compensation factor and saturation; determining a first threshold value based on the mountain shadow index, and adjusting pixel values of pixels in different areas in the remote sensing image data based on the first threshold value; and calculating a difference value between the normalized water body index and the mountain shadow index to obtain a water body shoreline extraction index of each pixel, identifying a region type corresponding to each pixel based on a relationship between the water body shoreline extraction index and a second threshold, and extracting a coastline based on an identification result.

Description

Technical Field

[0001] This application relates to the field of remote sensing technology, and particularly to a method and device for coastline extraction. Background Art

[0002] The coastline is the boundary area between the ocean and the land, and its dynamic changes are affected by tides, storms, sea-level rise, and human activities (such as reclamation, port construction, etc.). Accurately extracting the coastline is of great significance for aspects such as marine environmental monitoring, coastal zone resource management, disaster prevention and mitigation, and waterway planning. For example, in the context of global climate change, sea-level rise may lead to coastal erosion and land loss. Therefore, precise coastline monitoring can provide a scientific basis for relevant departments. In addition, port construction, marine economic development, etc. require precise coastline data for reasonable planning. Therefore, high-precision and automated coastline extraction methods are crucial for marine scientific research and engineering applications.

[0003] Currently, coastline extraction mainly relies on remote sensing images and digital image processing technologies. Common methods include: methods based on optical remote sensing images, such as the Normalized Difference Water Index (NDWI) method, which uses the spectral characteristic differences between water bodies and land in different bands for water-land segmentation to extract the coastline. However, the NDWI method may produce misjudgments when facing high-turbidity water bodies or shadow areas near the coast, resulting in inaccurate water-land boundaries. Methods based on SAR (Synthetic Aperture Radar) images. SAR data can penetrate clouds and atmospheric interference and is suitable for all-weather monitoring of the coastline. However, due to the strong speckle noise in radar images, the coastline boundary is blurred and the extraction accuracy is limited. Methods based on deep learning. In recent years, deep learning (such as CNN, U-Net) has been used for coastline extraction and has achieved certain results. However, deep learning methods rely heavily on data and require a large amount of high-quality labeled data for training. In addition, in complex terrains (such as areas with heavy mountain shadows), deep learning models may be affected by the imbalance of the dataset, resulting in unstable extraction results.

[0004] Therefore, there is an urgent need for a method to solve the problem of inaccurate coastline extraction by existing methods in complex geographical environments (such as mountain shadows, high-turbidity water bodies, etc.) and improve the accuracy of coastline detection. Summary of the Invention

[0005] In view of this, this application provides a method and device for coastline extraction to solve the problem of inaccurate coastline extraction by existing methods in complex geographical environments (such as mountain shadows, high-turbidity water bodies, etc.) and improve the accuracy of coastline detection.

[0006] Specifically, this application is implemented through the following technical solutions:

[0007] The first aspect of the present application provides a coastline extraction method, and the method includes:

[0008] Obtain remote sensing image data of the target coast; the remote sensing image data includes remote sensing images of different regions in different bands; the different regions at least include a mountain shadow region and a water body region, and the radiation brightness difference between the remote sensing image of the mountain shadow region and the remote sensing image of the water body region is less than a preset threshold;

[0009] Determine a target band based on the spectral characteristic differences between the mountain shadow region and the water body region in different bands;

[0010] Extract a plurality of pixels of the remote sensing image data, and calculate the saturation based on the pixel values of the plurality of pixels; the saturation characterizes the degree of mountain shadow;

[0011] Extract the pixels of the remote sensing image data in the target band, and calculate the mountain shadow index of each pixel based on the difference between the pixel values of the pixels in different bands, the saturation compensation factor, and the saturation;

[0012] Determine a first threshold based on the mountain shadow index, and adjust the pixel values of the pixels in different regions in the remote sensing image data based on the first threshold;

[0013] Calculate the difference between the normalized water index and the mountain shadow index to obtain the water body coastline extraction index of each pixel, identify the region type corresponding to each pixel based on the relationship between the water body coastline extraction index and the second threshold, and extract the coastline based on the identification result.

[0014] The second aspect of the present application provides a coastline extraction device, and the device includes an acquisition module, a determination module, a calculation module, an adjustment module, and an extraction module;

[0015] Among them, the acquisition module is used to obtain remote sensing image data of the target coast; the remote sensing image data includes remote sensing images of different regions in different bands; the different regions at least include a mountain shadow region and a water body region, and the radiation brightness difference between the remote sensing image of the mountain shadow region and the remote sensing image of the water body region is less than a preset threshold;

[0016] The determination module is used to determine a target band based on the spectral characteristic differences between the mountain shadow region and the water body region in different bands;

[0017] The calculation module is used to extract a plurality of pixels of the remote sensing image data, and calculate the saturation based on the pixel values of the plurality of pixels; the saturation characterizes the degree of mountain shadow;

[0018] The calculation module is further configured to extract pixels of the remote sensing image data in the target band, and calculate the mountain shadow index of each pixel based on the difference between pixel values of different bands, the saturation compensation factor, and the saturation.

[0019] The adjustment module is configured to determine a first threshold based on the mountain shadow index, and adjust the pixel values of pixels in different regions of the remote sensing image data based on the first threshold.

[0020] The extraction module is configured to calculate the difference between the normalized water index and the mountain shadow index to obtain the water body shoreline extraction index of each pixel, identify the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and a second threshold, and extract the coastline based on the identification result.

[0021] The coastline extraction method and device provided in the present application use the mountain shadow index to correct the normalized water index, comprehensively consider the similarity of the spectral characteristics of the water area and the mountain shadow area, and optimize the accuracy of the coastline extraction through multi-dimensional adjustment. On the one hand, in remote sensing images, the water area and the mountain shadow area have similar spectral characteristics in some bands, especially in the near-infrared and short-wave infrared bands. The mountain shadow may be mistakenly identified as a water body due to insufficient light and low reflectivity, thereby affecting the accuracy of coastline extraction. The traditional normalized water index method mainly uses the reflectivity difference between the green light and short-wave infrared bands to identify water bodies, but in areas with more mountain shadows, its effect may be limited, resulting in a decrease in the accuracy of water body identification. By introducing the mountain shadow index, the mountain shadow and the real water body can be effectively distinguished, thereby reducing the situation where the mountain shadow is misjudged as a water body and improving the accuracy of water body extraction. Secondly, the correction effect of the mountain shadow index can enhance the contrast between the water body and the non-water body area, making the coastline boundary clearer. Normally, the water index calculated by the normalized water index method may have a transition area near the coastline, especially in shallow water areas or at the junction of water and land, which is affected by factors such as sediments, tides, and vegetation, making the coastline boundary unclear. By combining the hill shadow index for correction, the pixel values of these transition areas can be adjusted to make the boundary between water and land clearer, thereby optimizing the accuracy of coastline extraction. Secondly, the use of the hill shadow index for correction helps to optimize the threshold selection and improve the adaptability of the coastline extraction algorithm. Traditional coastline extraction methods usually rely on fixed thresholds to distinguish between water areas and mountain shadow areas, but due to changes in lighting conditions, sensor characteristics, and terrain environments in different regions, fixed thresholds are difficult to adapt to all scenarios, resulting in poor water extraction in some areas. By introducing the hill shadow index, the water extraction threshold can be dynamically adjusted according to the lighting and terrain characteristics of different regions, making water extraction more stable and improving the automation and robustness of coastline detection. In addition, this method can reduce misjudgments caused by factors such as cloud shadows, tidal changes, and sediment suspension, and improve the overall stability of coastline detection. In practical applications, the position of the coastline may be affected by tidal fluctuations, resulting in changes in the water boundary. The introduction of the hill shadow index can help eliminate the interference of some non-water factors, making the coastline detection results more stable and accurate. It is suitable for remote sensing image data at different times and in different regions, and improves the versatility of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart of the coastline extraction method provided in Example 1 of the present application;

[0023] Figure 2 This is a schematic diagram of the structure of the coastline extraction device provided in Example 2 of the present application. DETAILED DESCRIPTION

[0024] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application.

[0025] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit the present application. The singular forms "a", "the", and "said" used in this application are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0026] It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "upon" or "in response to determining".

[0027] Specific embodiments are given below to introduce the technical solutions of the present application in detail.

[0028] Figure 1 It is a flowchart of the coastline extraction method provided for the first embodiment of the present application. Please refer to Figure 1 , the method provided in this embodiment may include:

[0029] S101. Obtain remote sensing image data of the target coast.

[0030] Specifically, the target coast is set according to actual needs, and in this embodiment, it is not limited thereto. The remote sensing image data of the target coast includes remote sensing images of different regions in different bands, where the different regions at least include mountain shadow regions and water body regions.

[0031] It should be noted that in water areas, especially in clear waters with low suspended sediment content, due to the strong light absorption of water, the radiance of water is low in the visible light band (especially red and green light) and the short-wave infrared band, and may be slightly higher only in the short-wave blue light band. In the mountain shadow area, the mountain shadow is essentially caused by terrain occlusion, resulting in reduced illumination, so its radiance is low, especially in areas with severe occlusion. Although the original surface (such as rocks and vegetation) may have a high reflectivity in the visible light band and the short-wave infrared band, due to the influence of the shadow, its radiance is greatly weakened. In the green and red light bands, the radiance of the remote sensing image in the mountain shadow area is often close to that in the water area, making it difficult to distinguish between the mountain shadow area and the water area, that is, the difference in radiance between the remote sensing image in the mountain shadow area and the remote sensing image in the water area is less than a preset threshold. The preset threshold is set according to actual needs and is not limited in this embodiment.

[0032] When specifically implemented, satellite image data provided by relevant satellites is consulted to select remote sensing image data of the target coast. The relevant satellites can be Landsat 5TM, Landsat 8OLI or other remote sensing satellites. The obtained remote sensing image data contains multiple bands. The mountain shadow area and the water area in the remote sensing image data are labeled by image segmentation methods (such as threshold segmentation, supervised classification or unsupervised classification), and the labeling results are adjusted by using reference data or manual labeling methods to ensure the accurate division of the mountain shadow area and the water area.

[0033] Optionally, after obtaining the remote sensing image data of the target coast, the method further includes: calculating the radiance based on the pixel value of each pixel in the remote sensing image data; calculating the difference in radiance between different pixels based on the radiance; and matching the corresponding coastline extraction method based on the difference in radiance.

[0034] Specifically, the pixel value (i.e., DN value, digital number) of each pixel in the remote sensing image data is read, and the pixel value is converted into radiance by using the radiometric calibration formula:

[0035] L r =G rescale ×DN+B rescale ;

[0036] wherein, the L r is the radiance; the G rescale is the gain coefficient; the DN is the pixel value; and the B rescale is the offset.

[0037] Further, after calculating the radiance of each pixel, key bands (such as visible light, near-infrared, shortwave infrared) are selected to calculate the average radiance of the mountain shadow area and the water area, calculate the radiance difference between the mountain shadow area and the water area, and compare the magnitude relationship between the radiance difference and a preset threshold. If the radiance difference is less than the preset threshold, it indicates that the spectral characteristics of the mountain shadow area and the water area are relatively close, and the method of this application needs to be used for further processing to divide different regions. If the radiance difference is not less than the preset threshold, it indicates that the spectral characteristics of the mountain shadow area and the water area are quite different, and the traditional water body index extraction method can be directly used to divide different regions.

[0038] The method provided in this embodiment can improve the adaptability, calculation efficiency, and extraction accuracy of the algorithm by calculating the radiance difference of each pixel in the remote sensing image and matching the corresponding coastline extraction method. Since the radiance difference between the water area and the mountain shadow area is different under different coastal environments, directly using a fixed method may lead to misjudgment, while adaptively selecting a processing strategy based on the brightness difference can optimize the extraction effect. In areas where the water-land boundary is obvious and the radiance difference is large, the traditional Normalized Difference Water Index (NDWI) can be directly used for water body extraction, while for areas with a small brightness difference and similar spectral characteristics between the mountain shadow and the water body, the matching WSEM method of this application is adopted to correct the misclassification problem through the mountain shadow index and avoid the shadow being misidentified as water. In addition, the brightness characteristics in different bands may show complex non-linear relationships, such as special geomorphic areas like the intertidal zone and wetlands. At this time, deep learning methods can be combined to further optimize the boundary extraction. This method not only improves the extraction accuracy but also reduces unnecessary computational overhead, realizes the generalization adaptation to different types of remote sensing images, and makes the coastline extraction more intelligent and precise.

[0039] S102. Determine the target band based on the spectral characteristic differences between the mountain shadow area and the water area in different bands.

[0040] Specifically, the target band refers to the band that can effectively distinguish the spectral characteristics of the mountain shadow area and the water area, that is, the target band shows an obvious reflectance or radiance difference between the mountain shadow area and the water area. The target band usually includes the red light band and the near-infrared band. Since the reflectance of the water area in the near-infrared (NIR) band is extremely low, while the mountain shadow area, although having a low overall reflectance, usually shows a slightly higher value in the near-infrared band, the reflectance difference in the near-infrared band can be used for distinction. And in the red light (Red) band, the reflectance in the water area is low, while in the mountain shadow area, it may be different due to the weak reflection of the underlying surface. The red light band can cooperate with the near-infrared band to further optimize the segmentation of the mountain shadow area and water body extraction.

[0041] In specific implementation, based on the remote sensing image data of the target coast collected, for each pixel in the remote sensing image data, spectral reflectance data of different bands are extracted, and the reflectance distributions of the mountain shadow area and the water area in each band are calculated respectively. By comparing the reflectance differences between the mountain shadow area and the water area in each band, characteristic values (such as reflectance mean, standard deviation, normalized index, etc.) are calculated. Based on the results of spectral characteristic analysis, bands with significant differences in characteristic values that can significantly distinguish the mountain shadow area and the water area (such as the near-infrared band and the red light band) are selected as the target bands for subsequent processing.

[0042] S103. Extract a plurality of pixels of the remote sensing image data, and calculate the saturation based on the pixel values of the plurality of pixels.

[0043] Specifically, saturation represents the intensity or vividness of a color. In this embodiment, saturation is used to characterize the degree of mountain shadow.

[0044] It should be noted that the blue-green-red bands are selected to calculate the saturation in this embodiment for the following considerations. First, the blue, green, and red bands are visible light bands, which conform to the human eye's color perception model (RGB), facilitating the calculation of color saturation and reflecting the purity and depth of colors. Moreover, due to insufficient light in the mountain shadow area, the overall reflectance of the BGR bands is usually reduced, but its color distribution still retains a certain pattern. Calculating the saturation of the BGR bands can measure the depth of the mountain shadow because in the mountain shadow area, the pixel values of all three bands are low, resulting in a high saturation, while in the bright area, the color distribution is relatively uniform and the saturation is relatively low. Also, since the near-infrared (NIR) and short-wave infrared (SWIR) bands have strong absorption of water, and at the same time have a high reflectance on ground objects such as vegetation. If these bands are used to calculate the saturation, it may be affected by other ground objects, which is not conducive to accurately measuring the characteristics of the mountain shadow. In contrast, the BGR bands mainly reflect the color information within the visible light range and are more suitable for calculating color saturation.

[0045] In specific implementation, the step of extracting a plurality of pixels of the remote sensing image data and calculating the saturation based on the pixel values of the plurality of pixels includes: determining the maximum pixel value and the minimum pixel value based on the pixel values of the plurality of pixels of the remote sensing image data in the blue-green-red bands; calculating the difference between the maximum pixel value and the minimum pixel value; and determining the quotient of the difference and the maximum pixel value as the saturation.

[0046] In specific implementation, all pixels of the remote sensing image data in the blue, green, and red bands are extracted, and the pixel values of all these pixels are traversed to determine the maximum pixel value and the minimum pixel value among all pixel values. Subtract the minimum pixel value from the maximum pixel value, and divide the difference by the maximum pixel value to obtain the saturation. The saturation can be expressed as:

[0047]

[0048] Wherein, S is the saturation; max(B1, B2, B3) is the maximum pixel value; min(B1, B2, B3) is the minimum pixel value; B1, B2, and B3 are the blue, green, and red light wavebands.

[0049] S104. Extract the pixels of the remote sensing image data in the target waveband, and calculate the mountain shadow index of each pixel based on the difference between the pixel values of different wavebands, the saturation compensation factor, and the saturation.

[0050] Specifically, the mountain shadow index is used to quantify the intensity of the mountain shadow area in the remote sensing image. Since the illumination in the mountain shadow area is weak and the overall reflectivity is low, especially in the visible light wavebands (blue, green, red), its spectral characteristics are similar to those of water bodies, so it is difficult to directly distinguish. The calculation of the mountain shadow index combines the spectral characteristics of the near-infrared (NIR) and red (Red) wavebands, and combines the saturation information to enhance the recognition effect of the mountain shadow area and reduce the confusion with water bodies. The saturation compensation factor is used to adjust the difference between the pixel values of the pixels in the target waveband to reduce the saturation of the water body area and improve the separation degree of the mountain shadow area.

[0051] It should be noted that the saturation compensation factor is a positive value, and its value range is usually between 0.01 and 0.1, and the saturation compensation factor is inversely proportional to the absolute value of the difference between the pixel values of the pixels in the target waveband. When the absolute value of the difference between the pixel values of the pixels in the target waveband is large, a smaller value can be selected for the saturation compensation factor to reasonably reduce the saturation. Since the difference between the pixel values of the pixels in the water body area in the target waveband is a small negative value and smaller than the difference between the pixel values of the pixels in the mountain shadow area in the target waveband, therefore, after multiplying the difference between the pixel values of the pixels in the water body area in the target waveband by the saturation compensation factor and then adding the saturation, the saturation of the water body area can be significantly reduced. In this way, there is a significant difference between the saturation of the water body area and the saturation of the mountain shadow area.

[0052] In specific implementation, the determination process of the saturation compensation factor includes: calculating the absolute value of the difference between the pixel value of the first pixel and the pixel value of the second pixel in the target waveband for different regions; determining an interval based on the inverse ratio of the absolute value of the difference; determining the difference margin of the mountain shadow area and the water body area in terms of the mountain shadow degree based on the recognition accuracy; and determining any value in the interval as the saturation compensation factor based on the difference margin.

[0053] Specifically, for different regions, the first pixel and the second pixel are selected in the target band, the pixel values of the first pixel and the second pixel are calculated respectively, and the absolute value of the difference between the two pixel values is calculated based on the calculated pixel values. An initial interval range of the saturation compensation factor is set, and according to the magnitude of the calculated absolute value of the difference, the interval range is adjusted by using the inverse relationship between the two. When the absolute value of the difference is small, a relatively large saturation compensation factor is corresponding, and when the absolute value of the difference is large, a relatively small saturation compensation factor is corresponding.

[0054]

[0055] Among them, the σ range is the said interval; the k is the adjustment coefficient; the D is the absolute value of the difference; the ∈ is a very small number to prevent the denominator from being zero.

[0056] Furthermore, the existing data is statistically analyzed, and the average mountain shadow index of the mountain shadow area and the water area is calculated. The difference margin between the average mountain shadow index of the mountain shadow area and the average mountain shadow index of the water area is calculated, and a threshold range is set based on the difference margin. In the determined saturation compensation factor interval, any value is selected. If the calculated mountain shadow index under this value meets the set threshold range, then this value is determined as the saturation compensation factor. If the calculated mountain shadow index does not meet the set threshold range, then a value is reselected within the interval.

[0057] The method provided in this embodiment, firstly, by calculating the absolute value of the difference between the pixel values of the mountain shadow area and the water area in the target band, it can dynamically adapt to the changes of different image data, ensure that the saturation compensation factor will not fail due to environmental changes, but can be adjusted according to the specific image characteristics, enhancing the applicability of the method. At the same time, this method avoids the misclassification problem that may be caused by the traditional fixed saturation compensation factor, enabling the effective distinction between the mountain shadow area and the water area under different remote sensing image conditions. Secondly, determining the interval based on the inverse ratio of the absolute value of the difference can make the saturation compensation factor increase appropriately when the pixel value difference is small and decrease appropriately when the difference is large. This adjustment mechanism can smooth the change of the saturation compensation factor, reduce the influence of outliers or local noise in the remote sensing image on the classification accuracy, and improve the robustness. In addition, this method optimizes the discrimination ability of the mountain shadow index by combining the difference margin, making the distinction between the water area and the mountain shadow area in terms of the degree of mountain shadow more obvious, thereby optimizing the accuracy of remote sensing image classification and avoiding misclassification caused by improper threshold setting. At the same time, since this method is not limited to a single band, but comprehensively calculates the saturation compensation factor based on the information of different bands, it can effectively play a role in multi-band images, enhancing the utilization rate of spectral information and improving the accuracy of coastline extraction.

[0058] In specific implementation, the pixels of the remote sensing image data in the target band are extracted, and the mountain shadow index of each pixel is calculated based on the difference between the pixel values of different bands, the saturation compensation factor, and the saturation, including: extracting the first pixel and the second pixel of the remote sensing image data in the target band, and calculating the difference between the pixel value of the first pixel and the pixel value of the second pixel; establishing a matching relationship between the difference and the saturation compensation factor, and determining the saturation compensation factor based on the matching relationship and the difference; determining the product of the difference and the saturation compensation factor as an intermediate value, and determining the sum value of the intermediate value and the saturation as the mountain shadow index.

[0059] Specifically, in the target band, the first pixel and the second pixel are extracted from the remote sensing image data, and their pixel values are respectively obtained. The difference between the pixel value of the first pixel and the pixel value of the second pixel is calculated to obtain the spectral feature change between different regions. A matching relationship between the pixel value difference and the saturation compensation factor is constructed, so that different differences correspond to different saturation compensation factors to adapt to different remote sensing image conditions. Based on the determined matching relationship, the corresponding saturation compensation factor is found through the calculated difference. Further, the pixel value difference is multiplied by the saturation compensation factor to obtain an intermediate value, and the intermediate value represents the influence of the saturation compensation factor on the pixel value difference. The sum value of the intermediate value and the saturation is determined as the final mountain shadow index. The mountain shadow index can be expressed as:

[0060] MSI = (NIR - Red) × σ + S;

[0061] Wherein, the MSI is the mountain shadow index; the NIR is the pixel value of the near-infrared band; the Red is the pixel value of the red light band; the σ is the saturation compensation factor; the S is the saturation.

[0062] S105. Determine a first threshold based on the mountain shadow index, and adjust the pixel values of the pixels in different regions of the remote sensing image data based on the first threshold.

[0063] Specifically, the first threshold is used to screen out the mountain shadow area in the remote sensing image data and remove the water area in the remote sensing image data. The first threshold is determined according to the mountain shadow index, and the first threshold is usually between the pixel value of the pixel in the water area in the remote sensing image data and the pixel value of the pixel in the mountain shadow area in the remote sensing image data.

[0064] It should be noted that the pixel value of the pixel in the adjusted mountain shadow area in the remote sensing image data is a high value, and the pixel value of the pixel in the adjusted water area in the remote sensing image data is 0, that is, the mountain shadow indexes of the mountain shadow area and the water area are quite different and are relatively easy to distinguish.

[0065] In specific implementation, determining the first threshold based on the mountain shadow index includes: determining a third pixel value and a fourth pixel value based on the pixel values of pixels in the remote sensing image data in different regions; the third pixel value being the maximum pixel value of the pixels in the water area in the remote sensing image data, and the fourth pixel value being the minimum pixel value of the pixels in the mountain shadow area in the remote sensing image data; selecting any pixel value between the third pixel value and the fourth pixel value and determining it as the initial threshold; and adjusting the initial threshold based on the mountain shadow index to obtain the first threshold.

[0066] Specifically, in the remote sensing image data, through the divided water area and mountain shadow area, extract the pixels of these areas and calculate the mountain shadow index of each pixel. Among all the pixels in the water area, traverse and find the pixel with the highest pixel value, denoted as the third pixel value. Among all the pixels in the mountain shadow area, traverse and find the pixel with the lowest pixel value, denoted as the fourth pixel value. Between the third pixel value (the maximum water pixel value) and the fourth pixel value (the minimum mountain shadow pixel value), select a suitable pixel value as the initial threshold, and when selecting, it can be based on the mean, median or some empirical rules. Further, calculate the average mountain shadow index of the water area and the average mountain shadow index of the mountain shadow area in the remote sensing image data obtained by calculation, and calculate the difference in the mountain shadow index between the water area and the mountain shadow area. Based on the magnitude of the difference in the mountain shadow index, determine the adjustment coefficient for adjusting the initial threshold, and adjust the initial threshold based on the adjustment coefficient to obtain the first threshold. Among them, the adjustment coefficient is negatively correlated with the difference in the mountain shadow index.

[0067] In specific implementation, adjusting the pixel values of pixels in different regions in the remote sensing image data based on the first threshold includes: comparing the mountain shadow index of each pixel with the first threshold, and determining the pixel value of the pixel lower than the first threshold as 0; comparing the mountain shadow index of each pixel with the first threshold, and retaining the pixel values of the pixels not lower than the first threshold.

[0068] Specifically, traverse the mountain shadow index of each pixel. If the mountain shadow index of a certain pixel is lower than the first threshold, then this pixel is a non-mountain shadow area, and set the pixel value of this pixel to 0. If the mountain shadow index of a certain pixel is not lower than the first threshold, then this pixel is a mountain shadow area, and keep the pixel value of this pixel unchanged.

[0069] The method provided in this embodiment can more accurately distinguish the mountain shadow area and the water area and improve the processing accuracy of remote sensing image data by determining the first threshold based on the mountain shadow index and adjusting the pixel values of pixels. First, in different spectral bands, the spectral characteristics of the water area and the mountain shadow area are similar, and directly using a fixed threshold may lead to classification errors. Therefore, the maximum pixel value of the water area and the minimum pixel value of the mountain shadow area in the remote sensing image data are used to dynamically determine the initial threshold, making the method adaptive. Second, since the saturation of the remote sensing image reflects the contrast and color intensity of the image, and environmental illumination and imaging noise may cause pixel value fluctuations, the initial threshold is adjusted based on the mountain shadow index to make the finally determined first threshold more stable, which helps to reduce misclassification. After that, the mountain shadow index is screened using the first threshold, and the pixel values lower than the first threshold are set to 0, while the pixel values of the pixels not lower than the threshold are retained, thereby enhancing the contrast of the target area, highlighting the mountain shadow or water area, reducing background interference, and improving the accuracy of target extraction. In addition, compared with complex classification algorithms, this method can quickly complete classification through simple threshold calculation and pixel value adjustment, improving the calculation efficiency and being applicable to the processing of large-scale remote sensing image data. At the same time, since this method is dynamically adjusted according to the spectral characteristics of different regions, it can adapt to different lighting, terrain, and environmental conditions, enhancing the generalization ability of the classification method and making the remote sensing image processing more stable and reliable.

[0070] S106. Calculate the difference between the normalized water index and the mountain shadow index to obtain the water body shoreline extraction index for each pixel, identify the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and the second threshold, and extract the coastline based on the identification result.

[0071] Specifically, the water body shoreline extraction index is used to distinguish the water area and the mountain shadow area. The water body shoreline extraction index inherits the sensitivity of the normalized water index to water and the sensitivity of the mountain shadow index to mountain shadows, corrects the defect of the water index method in distinguishing mountain shadows, and after calculation by the water body shoreline extraction index, the water area shows high values, while the land area and the mountain shadow area both show low values. By setting the second threshold, the water and land can be separated.

[0072] When specifically implemented, the step of calculating the difference between the normalized water index and the mountain shadow index to obtain the water body shoreline extraction index for each pixel includes: determining the detection band based on the reflectance of the water area in different bands; extracting the third pixel and the fourth pixel of the remote sensing image data in the detection band, and calculating the normalized water index based on the third pixel and the fourth pixel; and determining the difference between the normalized water index and the mountain shadow index as the water body shoreline extraction index.

[0073] Specifically, the detection band refers to a specific spectral band used to calculate the Normalized Difference Water Index (NDWI) and distinguish water areas from mountain shadow areas. The spectral characteristics of the detection band in water areas and mountain shadow areas are significantly different. The detection band usually includes the green band and the short-wave infrared band. Since the reflectivity of the green band in water is relatively high, especially in clean water bodies, strong scattering will occur, so it can effectively enhance the brightness characteristics of water bodies. Moreover, the green band can also suppress the influence of land, making the contrast between water bodies and the surrounding environment more obvious, which helps to improve the accuracy of water body identification. Water bodies have a very high absorption rate in the short-wave infrared band and basically no reflection, resulting in extremely low brightness values in water areas at this band. The reflectivity of land (such as vegetation, soil, rocks, etc.) in the short-wave infrared band is relatively high, and the difference between the two is obvious. Therefore, the short-wave infrared band can be used to enhance the contrast between water bodies and land.

[0074] In specific implementation, based on the reflectivities of water areas and mountain shadow areas in different bands, select the bands with significant spectral characteristics in water areas and the bands significantly suppressed in mountain shadow areas, such as the green band and the short-wave infrared band. The green band is used to enhance water body characteristics, and the short-wave infrared band is used to suppress the influence of land areas. From the remote sensing image data, obtain the pixel values corresponding to the third pixel and the fourth pixel in the detection band respectively. Calculate the Normalized Difference Water Index based on the pixel values:

[0075]

[0076] where, the NDWI is the Normalized Difference Water Index; the Green is the pixel value in the green band; the SWIR1 is the pixel value in the short-wave infrared band.

[0077] After calculating the Normalized Difference Water Index, use the mountain shadow index to correct the Normalized Difference Water Index, and determine the difference between the Normalized Difference Water Index and the mountain shadow index as the water body shoreline extraction index:

[0078]

[0079] where, the WSEM is the water body shoreline extraction index; the Green is the pixel value in the green band; the SWIR1 is the pixel value in the short-wave infrared band; the NIR is the pixel value in the near-infrared band; the Red is the pixel value in the red band; the σ is the saturation compensation factor; the max(B1, B2, B3) is the maximum pixel value; the min(B1, B2, B3) is the minimum pixel value; the B1, B2, B3 are the blue, green, and red bands.

[0080] In specific implementation, identifying the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and the second threshold includes: determining the minimum water body shoreline extraction index and the maximum water body shoreline extraction index based on the water body shoreline extraction indices of multiple pixels in the remote sensing image data for different regions; selecting any value between the minimum water body shoreline extraction index and the maximum water body shoreline extraction index as the second threshold; comparing the water body shoreline extraction index of each pixel with the second threshold, setting the pixel value of the pixel lower than the second threshold to 0, setting the pixel value of the pixel not lower than the second threshold to 1, and determining the region type corresponding to each pixel based on the adjusted pixels; repeating the operation of determining the second threshold until the number of second thresholds reaches the maximum value, and comparing the recognition results under each second threshold to determine the optimal recognition result.

[0081] Specifically, count the water body shoreline extraction index values of all pixels, and determine the minimum value (representing the mountain shadow area) and the maximum value (representing the water body area) among them. Select a value between the minimum value and the maximum value as the initial second threshold to distinguish the water body area and the mountain shadow area. Compare the water body shoreline extraction index of each pixel with the second threshold. If the water body shoreline extraction index of the pixel is lower than the second threshold, set the pixel value of the pixel to 0 (indicating the mountain shadow area). If the water body shoreline extraction index of the pixel is not lower than the second threshold, set the pixel value of the pixel to 1 (indicating the water body area). Through the division of 0 and 1, initially distinguish the water body area and the mountain shadow area to obtain a preliminary coastline extraction result. Select multiple different second thresholds in sequence and repeat the above steps to generate multiple different coastline extraction results. Evaluate the coastline extraction results under multiple second thresholds, and determine the optimal second threshold by comparing accuracy indicators (such as classification accuracy rate, boundary clarity, etc.), and finally output the optimal coastline extraction result.

[0082] Optionally, after adjusting the pixel values of the pixels in the remote sensing image based on the second threshold, use opening operation, closing operation, and boundary detection algorithm to process the remote sensing image to obtain the coastline extraction result. The opening operation first performs an erosion operation on the remote sensing image to remove small interfering objects or noises in the remote sensing image, such as non-water body structures like bridges and buildings, which often affect the extraction of the coastline in coastline detection. The erosion operation eliminates these non-water body structures, but at the same time may cause partial loss of information on the water body boundary. Further, perform a dilation operation to restore the main information of the water body boundary. Dilation expands the non-water body area to fill the gaps that may appear after erosion, helping to highlight the main area of the coastline and ensuring the continuity and integrity of the water-land boundary. This processing process of first eroding and then dilating can effectively remove fine noises, thereby accurately extracting the coastline.

[0083] The order of closing operation is exactly the opposite of opening operation. First, the dilation operation is performed. By expanding the boundaries of non-water body areas, small gaps or discontinuities in the image are filled, such as small rivers or small cracks within the land. These gaps may affect the continuity of the coastline during coastline detection. The dilation operation fills these gaps, thus ensuring the integrity of the water body area. Then, the erosion operation is carried out. Erosion can restore the smoothness of the water body boundary, eliminate the expansion generated during the dilation process, and ensure the accurate shape of the coastline. Through this way of dilation first and then erosion, the closing operation can effectively close the small gaps in the water body and at the same time restore the smooth coastline boundary.

[0084] Finally, a boundary detection algorithm is used to extract the coastline in the remote sensing image. The boundary detection algorithm can identify the places with large gray-scale changes in the remote sensing image. For the binary remote sensing image, the coastline is the boundary between 0 and 1, that is, the boundary of the coastline.

[0085] The method provided in this embodiment uses the mountain shadow index to correct the normalized water index, comprehensively considers the similarity of the spectral characteristics of the water area and the mountain shadow area, and optimizes the accuracy of coastline extraction through multi-dimensional adjustment. On the one hand, in remote sensing images, the water area and the mountain shadow area have similar spectral characteristics in some bands, especially in the near-infrared and short-wave infrared bands. The mountain shadow may be mistakenly identified as a water body due to insufficient light and low reflectivity, thereby affecting the accuracy of coastline extraction. The traditional normalized water index method mainly uses the reflectivity difference between the green light and short-wave infrared bands for water body identification, but in areas with more mountain shadows, its effect may be limited, resulting in a decrease in the accuracy of water body identification. By introducing the mountain shadow index, the mountain shadow and the real water body can be effectively distinguished, thereby reducing the situation where the mountain shadow is misjudged as a water body and improving the accuracy of water body extraction. Secondly, the correction effect of the mountain shadow index can enhance the contrast between the water body and the non-water body area, making the coastline boundary clearer. Normally, the water index calculated by the normalized water index method may have a transition area near the coastline, especially in shallow water areas or at the junction of water and land, which is affected by factors such as sediments, tides, and vegetation, making the coastline boundary unclear. By combining the hill shadow index for correction, the pixel values of these transition areas can be adjusted to make the boundary between water and land clearer, thereby optimizing the accuracy of coastline extraction. Secondly, the use of the hill shadow index for correction helps to optimize the threshold selection and improve the adaptability of the coastline extraction algorithm. Traditional coastline extraction methods usually rely on fixed thresholds to distinguish between water areas and mountain shadow areas, but due to changes in lighting conditions, sensor characteristics, and terrain environments in different regions, fixed thresholds are difficult to adapt to all scenarios, resulting in poor water extraction in some areas. By introducing the hill shadow index, the water extraction threshold can be dynamically adjusted according to the lighting and terrain characteristics of different regions, making water extraction more stable and improving the automation and robustness of coastline detection. In addition, this method can reduce misjudgments caused by factors such as cloud shadows, tidal changes, and sediment suspension, and improve the overall stability of coastline detection. In practical applications, the location of the coastline may be affected by tidal fluctuations, resulting in changes in the water body boundary. The introduction of the hill shadow index can help eliminate the interference of some non-water factors, making the coastline detection results more stable and accurate, and applicable to remote sensing image data of different times and regions, thus improving the versatility of the method. Thirdly, by determining the first threshold based on the hill shadow index and adjusting the pixel value of the pixel, this method can more accurately distinguish between the hill shadow area and the water body area, thereby improving the processing accuracy of remote sensing image data. First, in different spectral bands, the spectral characteristics of the water body and the hill shadow area are similar. Directly using a fixed threshold may lead to classification errors. Therefore, the maximum pixel value of the water body area and the minimum pixel value of the hill shadow area in the remote sensing image data are used to dynamically determine the initial threshold, making the method adaptive.Secondly, since the saturation of remote sensing images reflects the image contrast and color intensity, and environmental illumination and imaging noise may cause pixel value fluctuations, the initial threshold is adjusted based on the mountain shadow index to make the finally determined first threshold more stable, which helps reduce misclassification. Then, the mountain shadow index is screened using the first threshold, and pixel values lower than the first threshold are set to 0, while pixel values of pixels not lower than the threshold are retained, thereby enhancing the contrast of the target area, highlighting the mountain shadow or water body area, reducing background interference, and improving the accuracy of target extraction. In addition, compared with complex classification algorithms, this method can quickly complete classification through simple threshold calculation and pixel value adjustment, improving the calculation efficiency and being applicable to the processing of large-scale remote sensing image data. At the same time, since this method is dynamically adjusted according to the spectral characteristics of different regions, it can adapt to different illumination, terrain, and environmental conditions, enhancing the generalization ability of the classification method and making remote sensing image processing more stable and reliable.

[0086] Corresponding to the foregoing embodiment of a method for extracting a coastline, the present application also provides an embodiment of a device for extracting a coastline.

[0087] Figure 2 It is a schematic structural diagram of the coastline extraction device provided in the second embodiment of the present application. Please refer to Figure 2 The device provided in this embodiment includes an acquisition module 210, a determination module 220, a calculation module 230, an adjustment module 240, and an extraction module 250;

[0088] Among them, the acquisition module 210 is configured to acquire remote sensing image data of a target coast; the remote sensing image data includes remote sensing images of different regions in different bands; the different regions at least include a mountain shadow region and a water body region, and the radiation brightness difference between the remote sensing image of the mountain shadow region and the remote sensing image of the water body region is less than a preset threshold;

[0089] The determination module 220 is configured to determine a target band based on the spectral characteristic differences between the mountain shadow region and the water body region in different bands;

[0090] The calculation module 230 is configured to extract multiple pixels of the remote sensing image data and calculate the saturation based on the pixel values of the multiple pixels; the saturation characterizes the degree of mountain shadow;

[0091] The calculation module 230 is further configured to extract the pixels of the remote sensing image data in the target band and calculate the mountain shadow index of each pixel based on the difference between the pixel values of the pixels in different bands, the saturation compensation factor, and the saturation;

[0092] The adjustment module 240 is configured to determine a first threshold based on the mountain shadow index, and adjust the pixel values of pixels in different regions of the remote sensing image data based on the first threshold;

[0093] The extraction module 250 is configured to calculate the difference between the normalized water index and the mountain shadow index to obtain a water body shoreline extraction index for each pixel, identify the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and a second threshold, and extract the coastline based on the identification result.

[0094] The device of this embodiment can be used to execute Figure 1 the steps of the method embodiment shown. The specific implementation principle and process are similar, and will not be elaborated here.

[0095] For the implementation process of the functions and roles of each unit in the above device, please refer to the implementation process of the corresponding steps in the above method for details, which will not be elaborated here.

[0096] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0097] The above are only the preferred embodiments of this application, and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included in the scope of protection of this application.

Claims

1. A coastline extraction method, characterized in that, The method includes: Obtaining remote sensing image data of a target coast; the remote sensing image data includes remote sensing images of different regions in different bands; the different regions at least include a mountain shadow region and a water body region, and the difference in radiance between the remote sensing image of the mountain shadow region and the remote sensing image of the water body region is less than a preset threshold; Determining a target band based on the spectral characteristics differences between the mountain shadow region and the water body region in different bands; Extracting multiple pixels of the remote sensing image data, and calculating the saturation based on the pixel values of the multiple pixels; the saturation characterizes the degree of mountain shadow; Extracting the pixels of the remote sensing image data in the target band, and calculating the mountain shadow index of each pixel based on the difference between the pixel values of the pixels in different bands, the saturation compensation factor, and the saturation; Determining a first threshold based on the mountain shadow index, and adjusting the pixel values of the pixels in different regions in the remote sensing image data based on the first threshold; Calculating the difference between the normalized difference water index and the mountain shadow index to obtain the water body shoreline extraction index of each pixel, identifying the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and a second threshold, and extracting the coastline based on the identification result.

2. The method according to claim 1, wherein The extracting multiple pixels of the remote sensing image data and calculating the saturation based on the pixel values of the multiple pixels includes: Determining the maximum pixel value and the minimum pixel value based on the pixel values of the multiple pixels of the remote sensing image data in the blue, green, and red bands; Calculating the difference between the maximum pixel value and the minimum pixel value; Determining the quotient of the difference and the maximum pixel value as the saturation.

3. The method according to claim 1, wherein The extracting the pixels of the remote sensing image data in the target band and calculating the mountain shadow index of each pixel based on the difference between the pixel values of the pixels in different bands, the saturation compensation factor, and the saturation includes: Extracting a first pixel and a second pixel of the remote sensing image data in the target band, and calculating the difference between the pixel value of the first pixel and the pixel value of the second pixel; Establishing a matching relationship between the difference and the saturation compensation factor, and determining the saturation compensation factor based on the matching relationship and the difference; Determining the product of the difference and the saturation compensation factor as an intermediate value, and determining the sum value of the intermediate value and the saturation as the mountain shadow index.

4. The method according to claim 1, wherein The process of determining the saturation compensation factor includes: Calculating the absolute value of the difference between the pixel value of the first pixel and the pixel value of the second pixel in the target band for different regions; Determining an interval based on the inverse ratio of the absolute value of the difference; Determining the difference margin in the degree of mountain shadow between the mountain shadow region and the water body region based on the recognition accuracy; Determining any value from the interval as the saturation compensation factor based on the difference margin.

5. The method according to claim 1, characterized in that The determining the first threshold based on the mountain shadow index includes: Determining a third pixel value and a fourth pixel value based on the pixel values of the pixels in the remote sensing image data for different regions; the third pixel value is the maximum pixel value of the pixels in the water body region in the remote sensing image data, and the fourth pixel value is the minimum pixel value of the pixels in the mountain shadow region in the remote sensing image data; Select any pixel value between the third pixel value and the fourth pixel value and determine it as the initial threshold; Adjust the initial threshold based on the mountain shadow index to obtain the first threshold.

6. The method according to claim 1, characterized in that The adjusting the pixel values of pixels in different regions in the remote sensing image data based on the first threshold includes: Compare the mountain shadow index of each pixel with the first threshold, and determine the pixel value of the pixel lower than the first threshold as 0; Compare the mountain shadow index of each pixel with the first threshold, and retain the pixel values of the pixels not lower than the first threshold.

7. The method according to claim 1, characterized in that, The calculating the difference between the normalized water index and the mountain shadow index to obtain the water body shoreline extraction index for each pixel includes: Determine the detection band based on the reflectance of the water body area in different bands; Extract the third pixel and the fourth pixel of the remote sensing image data in the detection band, and calculate the normalized water index based on the third pixel and the fourth pixel; Determine the difference between the normalized water index and the mountain shadow index as the water body shoreline extraction index.

8. The method according to claim 1, characterized in that, The identifying the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and the second threshold includes: Determine the minimum water body shoreline extraction index and the maximum water body shoreline extraction index based on the water body shoreline extraction indices of multiple pixels in different regions in the remote sensing image data; Select any value between the minimum water body shoreline extraction index and the maximum water body shoreline extraction index and determine it as the second threshold; Compare the water body shoreline extraction index of each pixel with the second threshold, determine the pixel value of the pixel lower than the second threshold as 0, determine the pixel value of the pixel not lower than the second threshold as 1, and determine the region type corresponding to each pixel based on the adjusted pixels; Repeat the operation of determining the second threshold until the number of second thresholds reaches the maximum value, and compare the recognition results under each second threshold to determine the optimal recognition result.

9. The method according to claim 1, wherein After obtaining the remote sensing image data of the target coast, the method further includes: Calculate the radiance based on the pixel values of each pixel in the remote sensing image data; Calculate the radiance difference between different pixels based on the radiance; Match the corresponding coastline extraction method based on the radiance difference.

10. A coastline extraction device, characterized in that, The device includes an acquisition module, a determination module, a calculation module, an adjustment module and an extraction module; Among them, the acquisition module is used to acquire the remote sensing image data of the target coast; the remote sensing image data includes remote sensing images of different regions in different bands; the different regions at least include a mountain shadow region and a water body region, and the radiance difference between the remote sensing image of the mountain shadow region and the remote sensing image of the water body region is less than a preset threshold; The determination module is used to determine the target band based on the spectral characteristics differences between the mountain shadow region and the water body region in different bands; The calculation module is used to extract multiple pixels of the remote sensing image data and calculate the saturation based on the pixel values of the multiple pixels; the saturation characterizes the mountain shadow degree; The calculation module is further configured to extract pixels of the remote sensing image data in the target band, and calculate the mountain shadow index of each pixel based on the difference between the pixel values of different bands, the saturation compensation factor, and the saturation; The adjustment module is configured to determine a first threshold based on the mountain shadow index, and adjust the pixel values of pixels in different regions of the remote sensing image data based on the first threshold; The extraction module is configured to calculate the difference between the normalized water index and the mountain shadow index to obtain the water body shoreline extraction index of each pixel, identify the region type corresponding to each pixel based on the relationship between the water body shoreline extraction index and the second threshold, and extract the coastline based on the identification result.

Images