Deep learning remote sensing data cloud detection method based on physical guidance

Abstract

The invention discloses a deep learning remote sensing data cloud detection method based on physical guidance, which belongs to the technical field of data processing, and comprises the following steps: step 1, preprocessing satellite remote sensing observation data and background field data, step 2, calculating cloud probability based on a radiation transmission physical process, and step 3, calculating cloud probability based on the radiation transmission physical process. And a physically-guided image segmentation deep learning large model cloud detection step. According to the satellite remote sensing data cloud detection method based on physical guidance, radiation transmission physical simulation is adopted, and Bayesian maximum posteriori optimization is used for estimating the probability of cloud appearance. On the basis, constraint information is introduced into a prompt encoder of an SAM large model in a point prompt and frame prompt mode, so that the deep learning image segmentation large model SAM has physical priori knowledge, more accurate cloud detection is achieved, and the method can flexibly adapt to different atmosphere and ocean conditions.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and specifically, relates to a deep learning remote sensing data cloud detection method based on physical guidance. Background Art

[0002] Clouds absorb and scatter the sun's infrared radiation, interfering with remote sensing inversion of surface parameters. Therefore, cloud detection in satellite remote sensing data is crucial to the analysis and application of satellite remote sensing data. Clouds usually exhibit spectral characteristics of high reflectivity and low brightness temperature. Therefore, a fixed threshold is generally set to classify observed pixel values ​​as cloud or non-cloud. However, due to the influence of environmental variables such as surface type and regional range, the threshold method cannot adapt to different environments and dynamic changes, resulting in a high false detection rate.

[0003] Therefore, recent mainstream research has mostly used deep neural networks to extract features from different scenarios, and used functions to map feature vectors to dynamic thresholds to determine clouds. The latest computer vision research shows that the feature extraction ability of neural networks mainly depends on the size of the learnable parameters in the network. When the network parameters reach hundreds of millions, its ability to extract and express features will be greatly improved. This type of network is called a "large model" because the number of parameters is much larger than that of a conventional deep network.

[0004] The cloud detection task belongs to image segmentation in computer vision, and the corresponding large model is usually called "Segment Anything Mode (SAM)". However, although the SAM large model has a strong ability to extract image information, it cannot be directly used for satellite remote sensing cloud detection tasks. This is because the model lacks the guidance of physical prior knowledge during reasoning and can only distinguish clouds by the color and texture of the image, making it difficult for the model to accurately detect clouds under complex weather conditions. However, the physical process of remote sensing, namely atmospheric radiation transmission, can provide additional prior knowledge for the reasoning of the SAM model. Summary of the invention

[0005] In view of the problems of limited model representation ability and lack of physical prior knowledge in current deep network cloud detection methods, the present invention proposes a physics-guided deep learning remote sensing data cloud detection method to solve the above problems.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions to achieve the above problems: A physics-guided deep learning remote sensing data cloud detection method, comprising: Step 1: Satellite observation data and background field data Preprocessing steps include: Obtain the reflectivity of the visible light channel in satellite observation data; Obtain the brightness temperature of the thermal infrared channel in the satellite observation data, which is the observed brightness temperature ; Background field data Preprocessing is used to convert the background field data The spatial resolution of the satellite observation data is processed to be consistent with the resolution of the satellite observation data.

[0007] Step 2, the cloud probability calculation step based on the radiation transfer physical process, includes: Given satellite observation data and background field data In the case of, according to Bayesian theory, calculate the probability of clear sky : in, For a given background field data Assuming that the sky is clear, the probability of the satellite observation value appearing is: For a given background field data And assuming that it is in the cloud state, the probability of the satellite observation value appearing is, is the prior probability of clear sky, The calculation method is: ; in, is the probability density function of the spectrum under clear sky conditions, is the probability density function of the texture part under clear sky conditions; The calculation method is: ; is the probability density function of the spectrum under cloud conditions, is the probability density function of the texture part under the cloud state, and They are all obtained using lookup tables based on empirical statistics or numerical simulation; The calculation method is: First, the cloud coverage ratio TCC in the numerical weather forecast data is used to determine the prior probability of cloudiness. , if TCC is between 0.5-0.95, assign TCC to ; If TCC is less than 0.5, then ; If TCC is greater than 0.95, , calculated ; Calculate the given satellite observation data and background field data The cloud probability under =1- .

[0008] Step three, the physically guided image segmentation model cloud detection step, includes the following steps.

[0009] (31) Load the pre-trained SAM image segmentation model; (32), inputting the preprocessed reflectivity TR into the SAM image segmentation model to generate a feature representation of the image; (33) As a point cue input into the cue encoder of the SAM image segmentation model, the SAM image segmentation model outputs the predicted cloud probability of each pixel as the first predicted cloud probability; (34) Find out The pixels whose cloud probability is not less than the set threshold constitute a point set, and all possible squares consisting of four pixels are generated from the point set; (35) Establish an integer linear programming model with the goal of maximizing the sum of the vertex cloud probabilities of the selected square; (36) Using an integer linear programming solver or a heuristic algorithm to obtain an optimal or approximately optimal square combination; (37), inputting the square combination as a box prompt into the prompt encoder of the SAM image segmentation model, and the SAM image segmentation model outputs the predicted cloud probability of each pixel as the second predicted cloud probability; (38) Calculate the average value of the first predicted cloud probability and the second predicted cloud probability, compare the average value with the set threshold, determine whether the corresponding pixel point is a cloud pixel or a non-cloud pixel, and complete cloud detection.

[0010] In some embodiments, the conversion method of the reflectivity TR in step 1 is: ; Among them, ER is the reflectivity of the earth's visual angle, and SZA is the solar zenith angle.

[0011] In some embodiments, step 1 uses data from numerical weather forecasts as background field data. The preprocessing steps include: (11) Identify the nearest satellite observation data pixel based on the latitude and longitude information of the numerical weather forecast data; (12) Assigning the satellite zenith angle of the satellite observation data pixel to the corresponding numerical weather forecast data, and the numerical weather forecast data with the satellite zenith angle assignment is the preprocessed background field data .

[0012] In some embodiments, the probability density function of the spectral portion under clear sky conditions is The calculation method is: ; in, To observe brightness temperature Simulated brightness temperature The vector difference of Jacobian matrix for brightness temperature simulations for atmospheric radiative transfer; is the covariance matrix of the background field data; is the sum of the covariance of the simulated brightness temperature and the covariance of the observed brightness temperature under the assumption that the background field data error is ignored; The method for obtaining the brightness temperature is as follows: input the atmospheric profile parameters and sea surface information into the atmospheric radiation transfer model MODTRAN, and the output brightness temperature of the top of the atmosphere is the simulated brightness temperature. .

[0013] In some embodiments, the probability density function of the texture part in the clear sky state is Obtained through historical experience lookup table.

[0014] In some embodiments, the method for generating the square in step (34) includes: For each pair of pixels in the point set , calculate the positions of the other two vertices that can form a square; It is determined whether there are pixels in the point set at the other two vertices. If there are pixels, the four pixels generate a square. Otherwise, a square cannot be generated.

[0015] In some embodiments, the objective function of the integer linear programming model in step (35) is: ; in is a square set, For square The vertex set of Vertex The probability of existence of The constraint conditions of the integer linear programming model include: each pixel point is used to construct at most one square.

[0016] In some embodiments, the visible light channel includes visible light in wavelength bands of 0.436 μm-0.454 μm, 0.545 μm-0.565 μm, and 0.662 μm-0.682 μm.

[0017] In some embodiments, the thermal infrared channel includes wavelengths of 10.763 μm and 12.013 μm.

[0018] In some embodiments, after step three, the process also includes using morphological operations or connected domain analysis on the data for completed cloud detection to remove small noise areas and isolated pixels. For cloud pixel areas larger than a preset area threshold, a connected domain analysis method is used to merge adjacent cloud pixel areas or separate misdetected non-cloud pixel areas. The processed cloud detection data is converted into a standard format and combined with the original remote sensing data image to generate a final cloud detection result map.

[0019] Compared with the prior art, the advantages and positive effects of the present invention are: 1. Improve the accuracy of cloud detection: The physically guided process provides additional prior knowledge, which can effectively constrain the mask decoding results of the SAM large model to a physically feasible space through point prompts and box prompts, thereby improving the accuracy of cloud detection.

[0020] 2. Flexible adaptation to different weather scenarios: While using the image encoder of the SAM large model to capture texture features, it also provides physical prior features for radiation transfer calculations. The two types of features can complement each other, thereby flexibly adapting to different weather scenarios.

[0021] Other features and advantages of the present invention will become more apparent after reading the detailed description of the embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a principle block diagram of an embodiment of the physical guided deep learning satellite remote sensing cloud detection method proposed by the present invention; Figure 2 It is a SAM image segmentation model diagram in one embodiment of the physical-guided deep learning satellite remote sensing cloud detection method proposed in the present invention. DETAILED DESCRIPTION

[0023] The specific implementation modes of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] It should be noted that in the description of the present invention, the terms "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings, which are merely for the convenience of description and do not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance. In the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0026] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0027] Example 1, see Figure 1 As shown, the present invention proposes a physical-guided deep learning satellite remote sensing cloud detection method, comprising: Step 1: Satellite observation data and background field data Preprocessing steps include: Since clouds usually show spectral characteristics of high reflectivity and low brightness temperature in remote sensing data, it is necessary to obtain reflectivity and brightness temperature data observed by satellite in this embodiment.

[0028] For the visible light channel (0.445μm, 0.555μm and 0.673μm bands), the earth view reflectance ER provided in the original data product is converted into the true reflectance TR: ; Among them, ER is the reflectivity of the earth's visual angle, and SZA is the solar zenith angle.

[0029] For the thermal infrared channels (10.763μm and 12.013μm), considering that brightness temperature has a larger dynamic range and can reflect more detailed features in the image, the radiance data provided in the original data product is converted into brightness temperature according to the "radiance-brightness temperature" lookup table, which is recorded as the observed brightness temperature. .

[0030] Background field data Preprocessing is used to convert background field data The spatial resolution of the remote sensing data is processed to be consistent with that of the remote sensing data.

[0031] Step 2, the cloud probability calculation step based on the radiation transfer physical process, includes: Given remote sensing data and background field data In the case of, according to Bayesian theory, calculate the probability of clear sky : .

[0032] in, It is the conditional likelihood probability under the Bayesian framework, defined as given background field data And assuming that it is a clear sky, the probability of the satellite observation value appearing. For a given background field data And assuming that it is a cloud layer, the probability of the satellite observation value appearing. Prior clear sky probability in the Bayesian formula It can be obtained by calculating the proportion of cloud pixels in the background field data. Through the Bayesian framework, combined with the physical radiation transfer process, and supplemented by the empirical lookup table, the probability of clear sky can be obtained , and the probability of cloud is The calculation results will be used to guide the next large-model cloud detection. Through the universality of physical laws, the large model will be constrained so that the results can meet the requirements of cloud detection accuracy while ensuring generalization, and no additional model training is required.

[0033] The following are respectively introduced , and The specific calculation method of .

[0034] (1) The calculation method is: .

[0035] in, is the probability density function of the spectrum under clear sky conditions, It is the probability density function of the texture part under clear sky condition.

[0036] (2) The calculation method is: .

[0037] is the probability density function of the spectrum under cloud conditions, is the probability density function of the texture part under the cloud state, and They are all obtained using a lookup table based on empirical statistics or numerical simulation.

[0038] (3) The calculation method is: First, the cloud coverage ratio TCC in the numerical weather forecast data is used to determine the prior probability of cloudiness. If the TCC value is between 0.5 and 0.95, the TCC value is assigned to ; If TCC is less than 0.5, then ; If TCC is greater than 0.95, Further calculations yield .

[0039] Step 3, the physical guided image segmentation model cloud detection step, includes: (31) From Download the SAM code and pre-trained model, and load the pre-trained SAM image segmentation model through the pytorch library.

[0040] (32) Input the preprocessed reflectivity TR into the image encoding module of the SAM model to generate a feature representation of the image. The image encoder codes used are all provided by the SAM model.

[0041] (33) As a point hint, it is input into the hint encoder of the SAM image segmentation model, and the SAM image segmentation model outputs the predicted cloud probability of each pixel, which is the first predicted cloud probability.

[0042] (34) Find out The pixel points whose cloud probability is not less than the set threshold constitute a point set, and all possible squares consisting of four pixels are generated from the point set.

[0043] (35) An integer linear programming model is established with the goal of maximizing the sum of the vertex cloud probabilities of the selected square.

[0044] (36) Using an integer linear programming solver or a heuristic algorithm, obtain the optimal or nearly optimal square combination.

[0045] (37) The square combination is input as a box prompt into the prompt encoder of the SAM image segmentation model, and the SAM image segmentation model outputs the predicted cloud probability of each pixel, which is the second predicted cloud probability.

[0046] (38) Calculate the average of the first predicted cloud probability and the second predicted cloud probability, compare the average with the set threshold, determine whether the corresponding pixel is a cloud pixel or a non-cloud pixel, and complete cloud detection.

[0047] In this embodiment, the SAM image segmentation model is used for cloud detection. The SAM image segmentation model has hundreds of millions of parameters, which far exceeds the conventional cloud detection neural network. Therefore, it has a strong feature extraction capability, which provides a prerequisite for high-precision cloud detection.

[0048] When implementing this solution, you need to ensure that the working environment has an NVIDIA GPU and CUDA and cuDNN are correctly installed to support deep learning acceleration. Install Python 3.6 or later and use pip to install the necessary libraries. Visit the PyTorch official website, get the appropriate installation command based on the CUDA version, and install PyTorch and torchvision. Then, install Hugging Face's Transformers library. When verifying the installation, check the versions of PyTorch and Transformers, and the availability of CUDA is True. Depending on the requirements of the solution, you may need to install other dependent libraries such as numpy and pandas. Finally, set CUDA-related environment variables to ensure that the GPU can be used normally. These steps will ensure that the environment is ready to accelerate the model training and inference process.

[0049] Start with the pre-trained SAM (SegmentAnything Model) image segmentation model and load the basic model parameters. Download the SAM model code and pre-trained model, and load the pre-trained model through pytorch. Input the pre-processed image into the image encoder of the SAM model to generate the feature representation of the image. The structure of the SAM image segmentation model is as follows Figure 2 shown.

[0050] The cloud detection method of this embodiment uses a physics-guided image segmentation model for cloud detection, uses radiation transmission physical simulation, and combines Bayesian theorem to physically constrain the probability of cloud appearance, and introduces the constraint information into the prompt encoder of the SAM large model in the form of point prompts and box prompts, generating a high-dimensional feature representation rich in physical information for subsequent pixel-by-pixel cloud detection. Due to the additional physical prior knowledge, this method improves the accuracy of cloud detection.

[0051] In some embodiments, the conversion of radiance into reflectivity in step 1 is based on the radiation transmission principle. Specifically, the conversion method of reflectivity TR is as follows.

[0052] ; Among them, ER is the reflectivity of the earth's visual angle, and SZA is the solar zenith angle.

[0053] In some embodiments, the brightness temperature conversion in step one utilizes the radiance-brightness temperature conversion lookup table provided in the VIIRS data product to convert the radiance of the thermal infrared channel into brightness temperature.

[0054] In some embodiments, in step 1, data in numerical weather forecast NWP (Numerical Weather Prediction) is used as background field data. Since the spatial resolutions of satellite remote sensing data and NWP data are different, background field data preprocessing is required.

[0055] Background field data The preprocessing steps include: (11) According to the latitude and longitude information of the numerical weather forecast data, identify the nearest remote sensing data pixel.

[0056] (12) Assign the satellite zenith angle of the remote sensing data pixel to the corresponding numerical weather forecast data. The numerical weather forecast data with the satellite zenith angle assignment is the preprocessed background field data. .

[0057] In some embodiments, the probability density function of the spectral portion under clear sky conditions is The calculation of can be derived from the Gaussian distribution assumption, and the calculation method is: .

[0058] in, To observe brightness temperature Simulated brightness temperature The vector difference of , The Jacobian matrix for brightness temperature simulations for atmospheric radiative transfer, is the covariance matrix of the background field data, is the sum of the covariance of the simulated brightness temperature and the covariance of the observed brightness temperature under the assumption that the background field data error is ignored, The method for obtaining the brightness temperature is as follows: input the atmospheric profile parameters and sea surface information into the atmospheric radiation transfer model MODTRAN, and the output brightness temperature of the top of the atmosphere is the simulated brightness temperature. .

[0059] Atmospheric profile parameters and sea surface information are read from NWP data. Atmospheric profile parameters include air pressure, temperature, humidity, ozone content, and carbon dioxide content. Sea surface information includes sea surface temperature, wind speed at a height of 10m, and humidity at a height of 2m.

[0060] In some embodiments, the probability density function of the texture part in the clear sky state is It is obtained through an empirical lookup table. In this lookup table, the .

[0061] In some embodiments, the method for generating the square in step (34) includes: For each pair of pixels in the point set , calculate the positions of the other two vertices that can form a square.

[0062] In this embodiment, a data structure such as a KD tree or a quadtree may be used to accelerate neighbor point queries.

[0063] It is determined whether there are pixels in the point set at the other two vertices. If there are pixels, the four pixels generate a square. Otherwise, a square cannot be generated.

[0064] In some embodiments, the objective function of the integer linear programming model in step (35) is: ; in is a square set, is a square The vertex set of Vertex The probability of existence of The constraint conditions of the integer linear programming model include: each pixel point is used to construct at most one square.

[0065] In some embodiments, the visible light channel includes visible light in wavelength bands of 0.436 μm-0.454 μm, 0.545 μm-0.565 μm, and 0.662 μm-0.682 μm.

[0066] In some embodiments, the thermal infrared channel includes wavelengths of 10.763 μm and 12.013 μm.

[0067] In some embodiments, after step three, the process also includes using morphological operations or connected domain analysis on the data after cloud detection to remove small noise areas and isolated pixels. For cloud pixel areas larger than a preset area threshold, a connected domain analysis method is used to merge adjacent cloud pixel areas or separate misdetected non-cloud pixel areas.

[0068] The processed cloud detection data is converted into a standard format and combined with the original remote sensing data image to generate the final cloud detection result map.

[0069] In some embodiments, the method further includes evaluating the cloud detection results.

[0070] Verify the model performance on an independent test set and evaluate its actual effect in the cloud detection task. The calculated indicators include: IoU (Intersection over Union): measures the degree of overlap between the predicted segmentation area and the true segmentation area. The calculation formula is: .

[0071] Where TP (True Positive) is the number of pixels predicted to be clouds and are actually clouds. FP (False Positive) is the number of pixels predicted to be clouds but are not clouds. FN (False Negative) is the number of pixels predicted not to be clouds but are actually clouds.

[0072] Dice Coefficient: Measures the similarity between the predicted segmentation and the true segmentation area.

[0073] .

[0074] The definitions of TP, FP, and FN are the same as above.

[0075] Pixel Accuracy: The ratio of correctly predicted pixels to total pixels.

[0076] .

[0077] Where TN is the number of pixels that the model predicts are not clouds and are not actually clouds.

[0078] Precision: The ratio of correctly predicted positive samples to predicted positive samples.

[0079] .

[0080] Recall: The proportion of correct predictions in actual positive samples: .

[0081] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by ordinary technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A deep learning remote sensing data cloud detection method based on physical guidance, characterized in that: include: Step 1: Satellite observation data and background field data Preprocessing steps include: Obtain the reflectivity of the visible light channel in satellite observation data; Obtain the brightness temperature of the thermal infrared channel in the satellite observation data, which is the observed brightness temperature ; Background field data Preprocessing is used to convert the background field data The spatial resolution of the satellite observation data is processed to be consistent with the resolution of the satellite observation data; Step 2, the cloud probability calculation step based on the radiation transfer physical process, includes: Given satellite observation data and background field data In the case of, according to Bayesian theory, calculate the probability of clear sky : ; in, For a given background field data Assuming that the sky is clear, the probability of the satellite observation value appearing is: For a given background field data And assuming that it is in the cloud state, the probability of the satellite observation value appearing is, is the prior clear sky probability, The calculation method is: ； in, is the probability density function of the spectrum under clear sky conditions, is the probability density function of the texture part under clear sky conditions; The calculation method is: ； is the probability density function of the spectrum under cloud conditions, is the probability density function of the texture part under the cloud state, and They are all obtained using lookup tables based on empirical statistics or numerical simulation; The calculation method is: Determine the prior probability of cloud cover by using the cloud cover ratio TCC in numerical weather forecast data , if TCC is between 0.5-0.95, assign TCC to ; If TCC is less than 0.5, then ; If TCC is greater than 0.95, , calculated ; Calculate the given satellite observation data and background field data The cloud probability under =1- ; Step 3, the physical guided image segmentation model cloud detection step, includes: (31) Load the pre-trained SAM image segmentation model; (32), inputting the preprocessed reflectivity TR into the SAM image segmentation model to generate a feature representation of the image; (33) As a point cue input into the cue encoder of the SAM image segmentation model, the SAM image segmentation model outputs the predicted cloud probability of each pixel as the first predicted cloud probability; (34) Find out The pixels whose cloud probability is not less than the set threshold constitute a point set, and all possible squares consisting of four pixels are generated from the point set; (35) Establish an integer linear programming model with the goal of maximizing the sum of the vertex cloud probabilities of the selected square; (36) Using an integer linear programming solver or a heuristic algorithm to obtain an optimal or approximately optimal square combination; (37), inputting the square combination as a box prompt into the prompt encoder of the SAM image segmentation model, and the SAM image segmentation model outputs the predicted cloud probability of each pixel as the second predicted cloud probability; (38) Calculate the average value of the first predicted cloud probability and the second predicted cloud probability, compare the average value with the set threshold, determine whether the corresponding pixel point is a cloud pixel or a non-cloud pixel, and complete cloud detection.

2. The method according to claim 1, characterized in that The conversion method of reflectivity TR in step 1 is: ； Among them, ER is the reflectivity of the earth's visual angle, and SZA is the solar zenith angle.

3. The method according to claim 1, characterized in that In step 1, the data from numerical weather forecast is used as background field data. The preprocessing steps include: (11) Identify the nearest satellite observation data pixel based on the latitude and longitude information of the numerical weather forecast data; (12) Assigning the satellite zenith angle of the satellite observation data pixel to the corresponding numerical weather forecast data, and the numerical weather forecast data with the satellite zenith angle assignment is the preprocessed background field data .

4. The method according to claim 1, characterized in that: Probability density function of the spectrum under clear sky conditions The calculation method is: ； in, To observe brightness temperature Simulated brightness temperature The vector difference of Jacobian matrix for brightness temperature simulations for atmospheric radiative transfer; is the covariance matrix of the background field data; is the sum of the covariance of the simulated brightness temperature and the covariance of the observed brightness temperature under the assumption that the background field data error is ignored; The method for obtaining the brightness temperature is as follows: input the atmospheric profile parameters and sea surface information into the atmospheric radiation transfer model MODTRAN, and the output brightness temperature of the top of the atmosphere is the simulated brightness temperature. .

5. The method according to claim 1, characterized in that: Probability density function of texture part under clear sky condition Obtained through historical experience lookup table.

6. The method according to any one of claims 1 to 5, characterized in that: The method for generating the square in step (34) includes: For each pair of pixels in the point set , calculate the positions of the other two vertices that can form a square; It is determined whether there are pixels in the point set at the other two vertices. If there are pixels, the four pixels generate a square. Otherwise, a square cannot be generated.

7. The method according to any one of claims 1 to 5, characterized in that: The objective function of the integer linear programming model in step (35) is: ； in is a square set, is a square The vertex set of Vertex The probability of existence of The constraint conditions of the integer linear programming model include: each pixel point is used to construct at most one square.

8. The method according to any one of claims 1 to 5, characterized in that: The visible light channel includes wavelength bands of 0.436 μm-0.454 μm, 0.545 μm-0.565 μm and 0.662 μm-0.682 μm.

9. The method according to any one of claims 1 to 5, characterized in that: The thermal infrared channel includes wavelength bands of 11 μm and 12 μm.

10. The method according to any one of claims 1 to 5, characterized in that: After step three, it also includes using morphological operations or connected domain analysis on the data after cloud detection to remove small noise areas and isolated pixels. For cloud pixel areas larger than a preset area threshold, the connected domain analysis method is used to merge adjacent cloud pixel areas, or to separate the misdetected non-cloud pixel areas; the processed cloud detection data is converted into a standard format, and combined with the original remote sensing data image to generate the final cloud detection result map.

Images