Remote sensing image cloud distribution dispersion degree detection method, device and equipment

Abstract

This disclosure provides a method, apparatus, and device for detecting the dispersion of cloud distribution in remote sensing images, applicable to the field of remote sensing image processing technology. The method includes: calculating composite band pixel gradient values ​​based on the blue-green band grayscale values ​​of the remote sensing image to be detected; calculating normalized composite band gradient values ​​based on the Sobel operator; generating a cloud mask binary image based on the remote sensing image to be detected; removing cloud patches with areas smaller than a first area threshold from the cloud mask binary image; extracting and calculating the mean of the normalized composite band gradient values ​​of the edge pixels of cloud patches with areas larger than a second area threshold; and calculating the total mean of the mean values; determining whether the total mean is less than the first threshold; if so, determining that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered. Based on this, accurate characterization of the dispersion of thin clouds can be achieved, thereby enabling the detection of cloud dispersion in images.

Description

Technical Field

[0001] This disclosure relates to the field of image processing, and more particularly to the field of remote sensing image processing technology, specifically to a method, apparatus, and device for detecting the dispersion of cloud distribution in remote sensing images. Background Technology

[0002] Clouds are one of the most significant interference factors when acquiring images using optical remote sensing satellites. Their presence significantly impacts the stability and accuracy of tasks such as ground feature extraction, time-series analysis, and multi-source image fusion. For a long time, cloud cover has been widely used as an important indicator of image usability. However, cloud cover only reflects the overall degree of cloud coverage and is insufficient to reveal the spatial distribution patterns and structural characteristics of clouds within an image. In practical applications, even if two images have the same cloud cover, their spatial cloud distribution may differ considerably, thus affecting the analysis and use of remote sensing images differently. The spatial distribution characteristics of clouds are highly complex, mainly manifested in two aspects:

[0003] Firstly, thin clouds have blurred boundaries and spectral reflectance close to the background of ground objects, often attached to the edges of large cloud patches; due to the large number of thin clouds, they may form local clusters or spatially connected areas, resulting in a complex aggregate structure in the overall distribution of clouds.

[0004] Secondly, the spatial structure of clouds is diverse and uneven. Clouds in images may present various spatial patterns such as clusters, stripes, and isolation. Different distribution patterns have different effects on image analysis. Cloud clusters can obscure ground features over large areas and easily form connected regions, affecting local pixel statistics and cloud dispersion analysis results.

[0005] Therefore, relying solely on cloud cover indicators is insufficient to reflect the true spatial structure of clouds in images, and it is difficult to provide accurate spatial distribution references for remote sensing analysis. How to quantify the aggregation or dispersion characteristics of clouds under different spatial distribution patterns has become an urgent problem to be solved in the application of remote sensing imagery. Summary of the Invention

[0006] This disclosure provides a method, apparatus, device, and storage medium for detecting the dispersion of cloud distribution in remote sensing images.

[0007] According to a first aspect of this disclosure, a method for detecting the dispersion of cloud distribution in remote sensing images is provided. The method includes:

[0008] Based on the blue-green band gray values ​​of the remote sensing image to be detected, the gradient value of the composite band pixel is calculated.

[0009] Based on the Sobel operator, the normalized composite band gradient value is calculated according to the composite band pixel gradient value.

[0010] Based on the remote sensing image to be detected, a cloud mask binary image is generated;

[0011] Remove cloud spots with an area smaller than the first area threshold from the binary image of the cloud mask, and extract cloud spots with an area larger than the second area threshold;

[0012] Calculate the mean of the normalized composite band gradient values ​​of the edge pixels of the cloud spots whose area is greater than the second area threshold, and calculate the total mean of the mean values;

[0013] Determine whether the total mean is less than a first threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered.

[0014] In addition to the aspects described above and any possible implementations, a further implementation is provided, wherein the method further includes:

[0015] When the total mean is greater than or equal to the first threshold, a scene-level cloud mask is generated based on the cloud mask binary image;

[0016] Determine whether the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected is greater than a second threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0017] If not, after removing the two cloud spots with the largest area in the scene-level cloud mask, determine whether the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image is greater than or equal to the third threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0018] If not, determine whether the ratio of the sum of cloud pixel values ​​in the binary cloud mask image to the sum of each pixel value in the cloud computing unit of the scene-level cloud mask is greater than the fourth threshold. If yes, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered. If no, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0019] As described above and in any possible implementation, a further implementation is provided, wherein calculating the synthetic band pixel gradient value based on the blue-green band grayscale value of the remote sensing image to be detected includes:

[0020] S i =2×b i -0.95×g i

[0021] Among them, S i b represents the composite band pixel feature value of the i-th pixel in the remote sensing image to be detected; i g represents the blue band grayscale value of the i-th pixel in the remote sensing image to be detected; iThis represents the green band gray value of the i-th pixel in the remote sensing image to be detected.

[0022] As described above and in any possible implementation, a further implementation is provided, wherein calculating the normalized synthetic band gradient value based on the Sobel operator and according to the synthetic band pixel gradient value includes:

[0023]

[0024] Among them, s i (k) represents the pixel matrix within the range of k pixels surrounding the i-th effective pixel in the synthetic band; G x (i) represents the pixel gradient of the i-th effective pixel in the synthesized band in the x-direction; G y (i) represents the pixel gradient of the i-th effective pixel in the y-direction in the synthetic band; This represents the horizontal weight matrix of the Sobel operator; G represents the vertical weight matrix of the Sobel operator; min(G) and max(G) represent the minimum and maximum gradient magnitudes of the synthesized band, respectively; G N (i) represents the normalized synthetic band gradient value of the i-th effective pixel in the synthetic band.

[0025] In addition to the aspects described above and any possible implementations, a further implementation is provided, wherein calculating the mean of the normalized synthetic band gradient values ​​of the edge pixels of the cloud spots with areas greater than the second area threshold, and calculating the total mean of the mean, comprises:

[0026]

[0027] Where m represents the total number of edge pixels of the cloud patch, G N(i) represents the normalized composite band gradient value of the i-th edge pixel; n represents the total number of cloud spots with an area greater than the second area threshold; This indicates the overall level of gradient intensity at the cloud edge.

[0028] In addition to the aspects and any possible implementations described above, a further implementation is provided in which the pixel value of the cloud region is set to 1 and the cloud shadow and other regions are set to 0 in the binary cloud mask image.

[0029] The step of generating a scene-level cloud mask based on the cloud mask binary image includes:

[0030] Using a preset sliding window as the calculation unit, if the calculation unit contains cloud pixels, the entire calculation unit is determined to have clouds and is recorded as 1; otherwise, the calculation unit is cloudless and is recorded as 0. The cloud mask binary image is traversed one by one according to the preset sliding step size to determine the calculation unit and generate a scene-level cloud mask.

[0031] As described above and in any possible implementation, a further implementation is provided, wherein the calculation process for the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected, the calculation process for the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image, and the calculation process for the ratio of the sum of cloud-containing pixel values ​​in the cloud mask binary image to the sum of each pixel value of the cloud-containing cloud unit in the scene-level cloud mask include:

[0032]

[0033] Among them, M s M represents the value of cloud pixels in the scene-level cloud mask; p M represents the value of a cloud-containing pixel in a binary image with a cloud mask. t q1 represents the value of cloud-containing pixels in the remaining area of ​​the scene-level cloud mask after removing the two largest cloud patches; q2 represents the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected; q3 represents the ratio of the total number of cloud-containing pixels in the cloud mask binary image to the total number of cloud-containing pixels in the scene-level cloud mask binary image; H and W represent the length and width of the image, respectively.

[0034] According to a second aspect of this disclosure, an apparatus for detecting the dispersion of cloud distribution in remote sensing images is provided. The apparatus includes:

[0035] The calculation module is used to calculate the gradient value of the composite band pixels based on the blue-green band gray values ​​of the remote sensing image to be detected.

[0036] The calculation module is also used to calculate the normalized synthetic band gradient value of the pixel based on the Sobel operator and the synthetic band pixel gradient value.

[0037] The generation module is used to generate a cloud mask binary image based on the remote sensing image to be detected;

[0038] The extraction module is used to remove cloud spots with an area smaller than a first area threshold on the binary image of the cloud mask, and to extract cloud spots with an area larger than a second area threshold.

[0039] The calculation module is also used to calculate the mean of the normalized synthetic band gradient values ​​of the edge pixels of the cloud spots whose area is greater than the second area threshold, and to calculate the total mean of the mean values.

[0040] The judgment module is used to determine whether the total mean is less than a first threshold. If so, the cloud distribution dispersion of the remote sensing image to be detected is determined to be relatively clustered.

[0041] According to a third aspect of this disclosure, an electronic device is provided. The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described above.

[0042] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the method described above.

[0043] This application provides a method for detecting the dispersion of cloud distribution in remote sensing images. The method calculates the gradient value of synthetic band pixels based on the blue-green band grayscale values ​​of the remote sensing image to be detected; then, based on the Sobel operator, it calculates the normalized synthetic band gradient value of pixels according to the gradient value of synthetic band pixels; based on the remote sensing image to be detected, it generates a cloud mask binary image; then, it removes cloud patches with an area smaller than a first area threshold from the cloud mask binary image and extracts cloud patches with an area larger than a second area threshold; it calculates the mean of the normalized synthetic band gradient values ​​of the edge pixels of cloud patches with an area larger than the second area threshold, and calculates the total mean of the mean values; it determines whether the total mean is less than the first threshold. If so, it determines that the dispersion of cloud distribution in the remote sensing image to be detected is relatively clustered. Based on this, it can address the problem of blurred thin cloud boundaries by calculating the gradient value of synthetic bands based on the optical characteristics of the remote sensing image, normalizing it, calculating the mean of cloud edge pixel features, and distinguishing them by thresholds, thus achieving an accurate characterization of the dispersion of thin clouds and enabling the detection of cloud dispersion in images.

[0044] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0045] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0046] Figure 1 A flowchart is shown for a method for detecting the dispersion of cloud distribution in remote sensing images according to an embodiment of the present disclosure;

[0047] Figure 2 A schematic diagram illustrating the traversal of computing units used according to embodiments of the present disclosure is shown;

[0048] Figure 3 A schematic diagram of a scene-level cloud mask used according to an embodiment of the present disclosure is shown;

[0049] Figure 4 A flowchart is shown for another method for detecting the dispersion of cloud distribution in remote sensing images according to an embodiment of the present disclosure;

[0050] Figure 5 A block diagram of a device for detecting the dispersion of cloud distribution in remote sensing images according to an embodiment of the present disclosure is shown;

[0051] Figure 6 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0053] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0054] In this disclosure, the gradient value of the synthetic band can be calculated based on the optical characteristics of the remote sensing image to address the problem of blurred thin cloud boundaries. After normalization, the mean value of the cloud edge pixel features is calculated, and the difference is made by thresholding. This achieves an accurate characterization of the dispersion of thin clouds, thereby enabling the detection of the dispersion of clouds in the image.

[0055] Figure 1 A flowchart of a method 100 for detecting the dispersion of cloud distribution in remote sensing images according to an embodiment of the present disclosure is shown.

[0056] In box 110, the gradient value of the composite band pixel is calculated based on the blue-green band gray values ​​of the remote sensing image to be detected.

[0057] In some embodiments, the remote sensing image to be detected is an optical satellite remote sensing image with a cloud cover of 5% to 90%.

[0058] In some embodiments, the study is based on the dispersion of clouds in remote sensing images, aiming to quantitatively analyze the spatial distribution characteristics of clouds in remote sensing images. When the cloud cover is too low or too high, the analysis of the spatial distribution of clouds in remote sensing images has limited significance. The analysis of cloud distribution characteristics is only valuable when the cloud cover of remote sensing images is between 5% and 90%. Therefore, it is necessary to ensure that the cloud cover of the selected images, i.e., the remote sensing images to be detected, is within this effective range to ensure the reliability and scientific nature of the cloud dispersion analysis.

[0059] It is worth noting that this disclosure introduces the concept of "cloud dispersion" for the first time, filling the gap in current cloud volume indicators for describing cloud spatial structure.

[0060] In some embodiments, the composite band features of the remote sensing image, i.e., the composite band pixel gradient value, can be calculated using the gray values ​​of the blue and green bands of the remote sensing image.

[0061] In some embodiments, the above-mentioned calculation of the synthetic band pixel gradient value based on the blue-green band gray values ​​of the remote sensing image to be detected includes:

[0062] S i =2×b i -0.95×g i

[0063] Among them, S i b represents the composite band pixel feature value of the i-th pixel in the remote sensing image to be detected; i g represents the blue band grayscale value of the i-th pixel in the remote sensing image to be detected; i This represents the green band gray value of the i-th pixel in the remote sensing image to be detected.

[0064] In some embodiments, a synthetic band gradient feature map can be generated by calculating the synthetic band pixel gradient values ​​of remote sensing images.

[0065] In box 120, based on the Sobel operator, the normalized synthetic band gradient value is calculated according to the synthetic band pixel gradient value.

[0066] In some embodiments, in order to obtain edge information of remote sensing images, the Sobel operator can be used to calculate the synthetic band gradient feature map. That is, based on the gradient value of the synthetic band pixel, the pixel normalized synthetic band gradient value is calculated, and finally the normalized synthetic band gradient magnitude map of the remote sensing image is generated.

[0067] In some embodiments, the above-mentioned calculation of the normalized synthetic band gradient value based on the Sobel operator and the synthetic band pixel gradient value includes:

[0068]

[0069] Among them, si (k) represents the pixel matrix within the range of k pixels surrounding the i-th effective pixel in the synthetic band; G x (i) represents the pixel gradient of the i-th effective pixel in the synthesized band in the x-direction; G y (i) represents the pixel gradient of the i-th effective pixel in the y-direction in the synthetic band; This represents the horizontal weight matrix of the Sobel operator; G represents the vertical weight matrix of the Sobel operator; min(G) and max(G) represent the minimum and maximum gradient magnitudes of the synthesized band, respectively; G N (i) represents the normalized synthetic band gradient value of the i-th effective pixel in the synthetic band.

[0070] In box 130, a cloud mask binary map is generated based on the remote sensing image to be detected.

[0071] In some embodiments, a cloud masking algorithm can be used to generate a corresponding cloud mask image based on the remote sensing image to be detected, so as to separate the cloud cover from the remote sensing image, and then convert the cloud mask image into a binary image, i.e., a cloud mask binary image.

[0072] In some embodiments, values ​​can be assigned based on the binary image of the cloud mask. For example, the pixel value of the cloud region can be set to 1, and the cloud shadow and other regions can be set to 0, which facilitates the subsequent generation of scene-level cloud masks.

[0073] In box 140, cloud spots with an area smaller than the first area threshold are removed from the binary image of the cloud mask, and cloud spots with an area greater than the second area threshold are extracted.

[0074] In some embodiments, the first area threshold 'a' and the second area threshold 'b' can be set according to the user's actual needs. For example, 'a' can be set to 100 pixels and 'b' to 400 pixels.

[0075] In some embodiments, during cloud detection, some ground features with similar spectral characteristics are easily misidentified as clouds. These ground features typically have a small spatial scale, while cloud patches often exhibit a larger spatial scale. Based on this, cloud patches with an area smaller than the area threshold a (i.e., less than 100 pixels) can be removed from the binary cloud mask image. Simultaneously, for cloud patches with an area greater than the area threshold b (i.e., greater than 400 pixels), the mean of the normalized composite band gradient values ​​of their edge pixels is calculated, thereby further calculating the overall mean of the mean of the mean of the normalized composite band gradient values ​​of the edge pixels of the cloud patch.

[0076] In box 150, calculate the mean of the normalized composite band gradient values ​​of the edge pixels of cloud spots with an area greater than the second area threshold, and calculate the total mean of the mean values.

[0077] In some embodiments, the above-mentioned calculation of the mean of the normalized composite band gradient values ​​of edge pixels of cloud spots with an area greater than the second area threshold, and the calculation of the total mean of the mean, includes:

[0078]

[0079] Where m represents the total number of edge pixels of the cloud patch, G N(i) represents the normalized composite band gradient value of the i-th edge pixel; n represents the total number of cloud spots with an area greater than the second area threshold; This indicates the overall level of gradient intensity at the cloud edge.

[0080] In box 160, determine whether the total mean is less than the first threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered.

[0081] In some embodiments, the first threshold, i.e., threshold 1, can be set according to the user's actual needs and experimental experience. For example, threshold 1 can be set to 0.09.

[0082] In some embodiments, when A value less than 0.09 indicates that the cloud patch edges change gently and the cloud distribution is relatively clustered. In other words, the dispersion of the cloud distribution in the remote sensing image to be detected is determined to be relatively clustered, and the dispersion level is marked as 1. If... If the value is greater than or equal to 0.09, the next step is to determine the discrete distribution characteristics of the cloud.

[0083] According to the embodiments of this disclosure, the following technical effects are achieved:

[0084] This system can calculate the composite band pixel gradient value based on the blue-green band grayscale values ​​of the remote sensing image to be detected; then, based on the Sobel operator, it calculates the pixel-normalized composite band gradient value; based on the remote sensing image to be detected, it generates a cloud mask binary image; then, it removes cloud patches with an area smaller than a first area threshold from the cloud mask binary image, and extracts cloud patches with an area larger than a second area threshold; it calculates the mean of the normalized composite band gradient values ​​of the edge pixels of the cloud patches with an area larger than the second area threshold, and calculates the total mean of the mean values; it determines whether the total mean is less than the first threshold, and if so, it determines that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered; based on this, it can address the problem of blurred thin cloud boundaries by calculating the composite band gradient value based on the optical characteristics of the remote sensing image, normalizing it, calculating the mean of cloud edge pixel features, and distinguishing it by a threshold, thus achieving accurate characterization of the dispersion of thin clouds and enabling the detection of cloud dispersion in the image.

[0085] In some embodiments, the above method further includes:

[0086] When the total mean is greater than or equal to the first threshold, a scene-level cloud mask is generated based on the cloud mask binary image.

[0087] Determine whether the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected is greater than the second threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0088] If not, after removing the two largest cloud spots in the scene-level cloud mask, determine whether the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image is greater than or equal to the third threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0089] If not, determine whether the ratio of the sum of cloud pixel values ​​in the binary cloud mask image to the sum of each cloud cell value in the scene-level cloud mask is greater than the fourth threshold. If yes, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered. If no, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

[0090] In some embodiments, the second threshold, the third threshold, and the fourth threshold, namely threshold 2, threshold 3, and threshold 4, can be set according to the user's actual needs and experimental experience. For example, threshold 2 = 0.5, threshold 3 = 0.2, and threshold 4 = 0.35 can be set.

[0091] In some embodiments, generating a scene-level cloud mask based on a cloud mask binary image includes:

[0092] Using a preset sliding window as the calculation unit, if the calculation unit contains cloud pixels, the entire calculation unit is determined to have clouds and is recorded as 1; otherwise, the calculation unit is cloudless and is recorded as 0. The cloud mask binary image is traversed one by one according to the preset sliding step size to judge the calculation unit and generate a scene-level cloud mask.

[0093] In some embodiments, the specific settings for the preset sliding window and the preset sliding step size can be configured according to the user's actual needs.

[0094] In some embodiments, a square sliding window of a certain size can be used as a calculation unit, such as a square sliding window with a side length of 256 pixels. If the calculation unit contains cloud pixels, the entire calculation unit is determined to have clouds and recorded as 1; otherwise, the calculation unit is cloudless and recorded as 0. The calculation unit is determined by traversing the cloud mask binary image with a certain sliding step size, such as a sliding step size of 128 pixels, one by one. Figure 2 As shown; if adjacent computing units in an overlapping area have clouds, the overlapping area is determined to have clouds, and a scene-level cloud mask is obtained, such as... Figure 3As shown; where, if there are no complete computing units at the edge of the cloud mask binary map, the actual data range can be used to determine whether there are cloud computing units.

[0095] In some embodiments, based on scene-level cloud masks, the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the optical satellite remote sensing image, the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image, and the ratio of the sum of cloud pixel values ​​in the cloud mask binary image to the sum of each pixel value in the scene-level cloud mask cloud cloud unit can be calculated respectively. Different thresholds are applied to different ratios for classification, thereby judging the degree of cloud dispersion and further realizing the detection of the degree of cloud dispersion in the image.

[0096] In some embodiments, the calculation process for the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected, the calculation process for the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image, and the calculation process for the ratio of the sum of cloud pixel values ​​in the cloud mask binary image to the sum of each pixel value in the scene-level cloud mask cloud computing unit include:

[0097]

[0098] Among them, M s M represents the value of cloud pixels in the scene-level cloud mask; p M represents the value of a cloud-containing pixel in a binary image with a cloud mask. t q1 represents the value of cloud-containing pixels in the remaining area of ​​the scene-level cloud mask after removing the two largest cloud patches; q2 represents the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected; q3 represents the ratio of the total number of cloud-containing pixels in the cloud mask binary image to the total number of cloud-containing pixels in the scene-level cloud mask binary image; H and W represent the length and width of the image, respectively.

[0099] In some embodiments, considering the optical characteristics of clouds in the image and the spatial distribution characteristics of pixel-level cloud masks and scene-level cloud masks, a multi-level threshold segmentation is adopted to determine the degree of cloud dispersion. The specific judgment process for accurately determining the degree of cloud dispersion may include:

[0100] If q1 is greater than 0.5 (threshold 2), the dispersion level is marked as 0, indicating that the cloud distribution is relatively scattered; otherwise, it is judged as q2. If q2 is greater than 0.2 (threshold 3), the dispersion level is marked as 0, indicating that the cloud distribution is relatively scattered; otherwise, it is judged as q3. If q3 is greater than 0.35 (threshold 4), the dispersion level is marked as 1, indicating that the cloud distribution is relatively clustered; otherwise, the dispersion level is marked as 0, indicating that the cloud distribution is relatively scattered.

[0101] As can be seen, considering the complexity of cloud distribution in real space, this disclosure calculates the ratio between pixel-level cloud masks and scene-level cloud masks in images. The pixel-level cloud mask reflects the cloud state of each pixel in the image, capturing local detail information; the scene-level cloud mask aggregates local information through fixed-size computational units, reflecting the overall cloud distribution pattern. The main differences between the two are: in areas of cloud aggregation, the pixel-level and scene-level masks are highly consistent; in areas of cloud dispersion, the pixel-level mask shows scattered cloud points, while the scene-level mask does not provide complete coverage. Finally, by applying different thresholds to different feature values, the dispersion of clouds can be quantified, enabling accurate characterization of cloud distribution in images under different cloud amounts, cloud structures, and ground feature backgrounds. This achieves a stable characterization of the discrete distribution characteristics of clouds under different cloud amounts, cloud structures, and ground feature backgrounds.

[0102] Figure 4 A flowchart illustrating another method for detecting the dispersion of cloud distribution in remote sensing images according to embodiments of the present disclosure is shown, such as... Figure 4 As shown, this can be used as an optical remote sensing image cloud dispersion assessment algorithm based on synthetic band gradient analysis and pixel scene mask analysis, specifically including:

[0103] Step 1: Load optical satellite remote sensing images with cloud cover between 5% and 90%; based on the optical satellite remote sensing images, the pixel gradient values ​​of the blue-green composite bands of the optical satellite remote sensing image data can be calculated, and the normalized pixel gradient values ​​can be calculated using the Sobel operator; based on the optical satellite remote sensing images, a cloud mask binary map can also be generated and assigned values, cloud spots with an area less than 100 pixels on the cloud mask binary map are removed, and cloud spots with an area greater than 400 pixels are extracted; determine whether the total mean of the normalized pixel gradient values ​​of the composite band cloud spots is less than 0.09. If so, it indicates that the cloud distribution is relatively clustered; otherwise, combine with Step 2 to further determine the distribution characteristics of the clouds;

[0104] Step 2: Based on the cloud mask binary image, using a square sliding window of a certain size as the calculation unit, determine whether each calculation unit has clouds and assign values ​​to each pixel to generate a scene-level cloud mask; calculate the cloud amount in the scene-level cloud mask and determine whether it is greater than 0.5. If yes, it indicates that the cloud distribution is relatively dispersed; if not, remove the two cloud spots with the largest area, calculate and determine whether the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud pixel values ​​in the cloud mask binary image is not less than 0.2. If yes, it indicates that the cloud distribution is relatively dispersed; if not, calculate and determine whether the ratio of the sum of cloud pixel values ​​in the cloud mask binary image to the sum of cloud pixel values ​​in the scene-level cloud mask is greater than 0.35. If not, it indicates that the cloud distribution is relatively dispersed; if yes, it indicates that the cloud distribution is relatively clustered.

[0105] In summary, the above approach is based on the fact that thin clouds have blurred boundaries and their spectral reflectance characteristics are close to the background of ground objects. They typically have low gradients and their boundaries tend to expand and converge. Considering that misjudged ground objects often appear as small, isolated patches, this disclosure removes isolated cloud patches with areas smaller than the area threshold 'a' to ensure the stability of the results. Only the edge pixels of cloud patches with areas larger than the area threshold 'b' are selected to calculate the normalized composite band gradient average. That is, by calculating the normalized composite band gradient average of the edge pixels of cloud patches, the degree of expansion of the cloud patch edges is shown. The total mean of the normalized composite band gradient average of cloud patches is also calculated. When the total mean is less than the threshold 1, it indicates that the cloud patch edges are gentle, the boundaries show an expanding trend, and the cloud distribution is relatively clustered. Otherwise, further analysis of the cloud distribution characteristics will be conducted, that is, the first-level judgment of cloud dispersion will be completed by using the threshold 1. Subsequently, scene-level cloud masks are calculated. By analyzing the distribution characteristics of pixel-level and scene-level cloud masks, the differences between local and overall cloud distribution are quantified. Different thresholds are used to classify different features, achieving the classification of image cloud dispersion under different cloud amounts, cloud structures, and ground background conditions. Specifically, a square is used as a calculation unit. If there are cloud pixels within the unit, the entire unit is determined to be a cloud. If the edge of the cloud patch does not meet the condition, it is judged according to the actual area. The calculation unit slides according to a set step size to generate scene-level cloud masks. Based on the cloud amount in the scene-level cloud mask, a second-level judgment is performed according to threshold 2. Further, the two cloud patches with the largest area are removed from the scene-level cloud mask. The ratio of the sum of cloud patch pixel values ​​in the remaining range of the pixel-level cloud mask to the sum of cloud pixel values ​​on the pixel-level cloud mask is calculated, and a third-level judgment is completed based on threshold 3. Finally, the ratio of the pixel-level cloud mask to the scene-level cloud mask is calculated, and a fourth-level judgment is performed based on threshold 4.

[0106] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.

[0107] The above is an introduction to the method embodiments. The following describes the solution described in this disclosure further through device embodiments.

[0108] Figure 5 A block diagram of a remote sensing image cloud distribution dispersion detection device 500 according to an embodiment of the present disclosure is shown. Figure 5 As shown, the device 500 includes:

[0109] The calculation module 510 is used to calculate the gradient value of the composite band pixels based on the blue-green band gray values ​​of the remote sensing image to be detected.

[0110] The calculation module 510 is also used to calculate the normalized synthetic band gradient value of the pixel based on the Sobel operator and the synthetic band pixel gradient value.

[0111] The generation module 520 is used to generate a cloud mask binary image based on the remote sensing image to be detected.

[0112] The extraction module 530 is used to remove cloud spots with an area smaller than the first area threshold on the binary image of the cloud mask, and to extract cloud spots with an area greater than the second area threshold.

[0113] The calculation module 510 is also used to calculate the mean of the normalized synthetic band gradient values ​​of the edge pixels of cloud spots with an area greater than the second area threshold, and to calculate the total mean of the mean values.

[0114] The judgment module 540 is used to determine whether the total mean is less than the first threshold. If so, the cloud distribution dispersion of the remote sensing image to be detected is determined to be relatively clustered.

[0115] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0116] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0117] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0118] Figure 6 A block diagram of an exemplary electronic device 600 capable of implementing embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0119] Electronic device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in ROM 602 or a computer program loaded into RAM 603 from storage unit 608. RAM 603 may also store various programs and data required for the operation of electronic device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. I / O interface 605 is also connected to bus 604.

[0120] Multiple components in electronic device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of displays, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows electronic device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0121] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608.

[0122] In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by computing unit 601, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, computing unit 601 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).

[0123] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0124] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0125] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0126] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).

[0127] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0128] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0129] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0130] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for detecting the dispersion of cloud distribution in remote sensing images, characterized in that, include: Based on the blue-green band gray values ​​of the remote sensing image to be detected, the gradient value of the composite band pixel is calculated. The calculation of the synthetic band pixel gradient value based on the blue-green band grayscale values ​​of the remote sensing image to be detected includes: ;in, This represents the synthetic band pixel feature value of the i-th pixel in the remote sensing image to be detected; This represents the blue band grayscale value of the i-th pixel in the remote sensing image to be detected; This represents the green band gray value of the i-th pixel in the remote sensing image to be detected; Based on the Sobel operator, the normalized composite band gradient value is calculated according to the composite band pixel gradient value. Based on the remote sensing image to be detected, a cloud mask binary image is generated; Remove cloud spots with an area smaller than the first area threshold from the binary image of the cloud mask, and extract cloud spots with an area larger than the second area threshold; Calculate the mean of the normalized composite band gradient values ​​of the edge pixels of the cloud spots whose area is greater than the second area threshold, and calculate the total mean of the mean values; Determine whether the total mean is less than a first threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered.

2. The method according to claim 1, characterized in that, The method further includes: When the total mean is greater than or equal to the first threshold, a scene-level cloud mask is generated based on the cloud mask binary image; in the cloud mask binary image, the pixel value of the cloud region is set to 1, and the cloud shadow and other regions are set to 0; the generation of the scene-level cloud mask based on the cloud mask binary image includes: using a preset sliding window as a calculation unit, if the calculation unit has cloud pixels, the calculation unit is determined to have clouds as a whole and recorded as 1, otherwise the calculation unit is cloudless and recorded as 0, and the cloud mask binary image is traversed one by one according to a preset sliding step size to judge the calculation unit and generate a scene-level cloud mask; Determine whether the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected is greater than a second threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed. If not, after removing the two cloud spots with the largest area in the scene-level cloud mask, determine whether the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image is greater than or equal to the third threshold. If so, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed. If not, determine whether the ratio of the sum of cloud pixel values ​​in the binary cloud mask image to the sum of each pixel value in the cloud computing unit of the scene-level cloud mask is greater than the fourth threshold. If yes, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively clustered. If no, determine that the cloud distribution dispersion of the remote sensing image to be detected is relatively dispersed.

3. The method according to claim 2, characterized in that, The calculation of the normalized synthetic band gradient value based on the Sobel operator and the synthetic band pixel gradient value includes: in, Indicates the first in the synthetic band i Around each effective cell A pixel matrix within a pixel range; Indicates the first in the synthetic band i One effective pixel in Pixel gradient in the direction; Indicates the first in the synthetic band i One effective pixel in Pixel gradient in the direction; This represents the horizontal weight matrix of the Sobel operator; This represents the vertical weight matrix of the Sobel operator; and These represent the minimum and maximum gradient magnitudes of the synthetic band, respectively; Indicates the first in the synthetic band i The normalized composite band gradient value of each effective pixel.

4. The method according to claim 3, characterized in that, The calculation of the mean of the normalized composite band gradient values ​​of the edge pixels of cloud spots with an area greater than the second area threshold, and the calculation of the total mean of the mean, includes: Where m represents the total number of edge pixels of the cloud patch. represents the normalized composite band gradient value of the i-th edge pixel; n represents the total number of cloud spots with an area greater than the second area threshold; This indicates the overall level of gradient intensity at the cloud edge.

5. The method according to claim 2, characterized in that, The calculation process for the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected, the calculation process for the ratio of the sum of cloud spot pixel values ​​in the remaining range of the cloud mask binary image to the sum of cloud spot pixel values ​​in the cloud mask binary image, and the calculation process for the ratio of the sum of cloud pixel values ​​in the cloud mask binary image to the sum of each pixel value of the cloud computing unit in the scene-level cloud mask include: in, This represents the value of cloud pixels in the scene-level cloud mask; This represents the value of the cloud-containing pixels in the binary image of the cloud mask; This indicates the values ​​of cloud pixels in the remaining area of ​​the binary cloud mask image after removing the two largest cloud spots in the scene-level cloud mask. This represents the ratio of the total number of pixels in the scene-level cloud mask to the total number of pixels in the remote sensing image to be detected. This represents the ratio of the sum of cloud spot pixel values ​​in the remaining range of the binary cloud mask image to the sum of cloud spot pixel values ​​in the binary cloud mask image. This represents the ratio of the total cloud pixel value in the binary image of the cloud mask to the total cloud unit pixel value in the scene-level cloud mask; H and W represent the length and width of the image, respectively.

6. A device for detecting the dispersion of cloud distribution in remote sensing images, characterized in that, include: The calculation module is used to calculate the gradient value of the composite band pixels based on the blue-green band gray values ​​of the remote sensing image to be detected. The calculation of the synthetic band pixel gradient value based on the blue-green band grayscale values ​​of the remote sensing image to be detected includes: ;in, This represents the synthetic band pixel feature value of the i-th pixel in the remote sensing image to be detected; This represents the blue band grayscale value of the i-th pixel in the remote sensing image to be detected; This represents the green band gray value of the i-th pixel in the remote sensing image to be detected; The calculation module is also used to calculate the normalized synthetic band gradient value of the pixel based on the Sobel operator and the synthetic band pixel gradient value. The generation module is used to generate a cloud mask binary image based on the remote sensing image to be detected; The extraction module is used to remove cloud spots with an area smaller than a first area threshold on the binary image of the cloud mask, and to extract cloud spots with an area larger than a second area threshold. The calculation module is also used to calculate the mean of the normalized synthetic band gradient values ​​of the edge pixels of the cloud spots whose area is greater than the second area threshold, and to calculate the total mean of the mean values. The judgment module is used to determine whether the total mean is less than a first threshold. If so, the cloud distribution dispersion of the remote sensing image to be detected is determined to be relatively clustered.

7. An electronic device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-5.

8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-5.

Images