A method and device for suppressing stray light in geostationary satellite cloud images

Abstract

The present invention discloses a method and a device for suppressing stray light in geostationary satellite cloud images, belonging to the technical field of image processing in satellite remote sensing. The method includes the following steps: S1: Detect the edge of the earth in the geostationary satellite cloud image, separate the earth disk and the cold space image outside the disk, and eliminate foreign object targets; S2: Calculate the mean value and standard deviation of the noise observed in the earth disk; S3: Dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light. The present invention can suppress the stray light caused by the multiple reflections of the energy of the off-field target into the field of view.

Description

Technical Field

[0001] The present invention relates to the field of image processing for satellite remote sensing, and particularly to a stray light suppression method and device for geostationary satellite cloud images. Background Art

[0002] Stray light in an optical system refers to non-imaging light reaching the image plane of the optical system. The stray light energy and the target energy within the field of view are received and responded to by the detector together, resulting in a decrease in image contrast and signal-to-noise ratio, and ultimately leading to a decline in image quality. Stray light in an optical system is a difficult problem that must be faced in the quantitative application of remote sensing data. In severe cases, stray light can even cause the system to fail and the acquired data cannot be applied.

[0003] Due to the complex illumination of the geostationary orbit, stray light is a difficult point in the quantitative application of geostationary optical remote sensing satellite data. An algorithm needs to be used to correct it to improve the radiation measurement accuracy. Figure 1 The stray light in the full-disk remote sensing image of the geostationary satellite is shown in [reference], and the source is the FY-2F satellite. Generally, there are two main reasons for geostationary stray light: 1. Stray light caused by direct leakage of off-field targets, which refers to the energy of targets with the optical axis pointing outside the field of view directly entering the field of view; 2. Stray light caused by multiple reflections of off-field target energy into the field of view. This type of stray light is the energy of targets with the optical axis pointing outside the field of view that enters the field of view after multiple reflections by the optical system. Compared with the direct leakage stray light, it is very difficult to determine the position of the energy source of this part of the stray light energy after multiple reflections. However, the magnitude of the stray light energy is related to off-field targets and is not a stable DC background, which has a great impact on the quality of remote sensing data.

[0004] The leakage stray light has a relatively large impact on data quality. However, due to the clear mechanism, good suppression effects have been achieved for leakage stray light both in the optical system design stage and in the ground data processing stage. In the optical system design stage, generally, the Monte Carlo analysis method or the ray tracing method is used to analyze the light rays entering the remote sensing instrument, and then optical components such as diaphragms are used to block the non-field-of-view light rays reaching the focal plane to achieve the purpose of reducing stray light; the typical ground data processing algorithm for leakage stray light is mainly the template method, and the template method is based on the stray light template as shown in Figure 2 the following figure.

[0005] Figure 2It is a stray light template represented in polar coordinates. The center point is the direction where the optical axis points. Take a certain direction as the 0° direction. The angle from the 0° direction is represented by α, and the distance from the center point is represented by l. Then, the stray light template is divided into several small regions with the unit area of dl·dα. By determining the excitation coefficient of the energy of each region contributing to the stray light at the center point, the magnitude of the stray light at the optical axis pointing position can be calculated from the observed values around the optical axis pointing position.

[0006] Specifically, it is described by formula (1) Figure 1 The shown stray light template.

[0007] S Z = S O *M + N………………………………………………………………(1)

[0008] Among them, S Z is the stray light at the instantaneous pointing point of the optical axis, S O is a small region on the stray light template, M is the stray light effect matrix, and N is the observation error matrix.

[0009] The template method determines the position of the excitation source contributing to the stray light at the optical axis pointing position, that is Figure 2 the position of the gray area on the stray light template in [], and then determines the excitation coefficient of each different small region contributing to the stray light at the optical axis pointing position through the statistics of actual remote sensing images, so as to calculate the energy magnitude of the stray light at the optical axis pointing position through the surrounding observed values, and then correct the stray light.

[0010] The principle of the template method is based on the fact that the energy source of the stray light at the optical axis pointing position is known, and what needs to be solved is only the excitation coefficient of the stray light excitation source to the stray light at the optical axis pointing position. This has a significant effect on the stray light caused by the direct light leakage of the optical system (i.e., reason 1); however, when the stray light is mainly the stray light caused by multiple reflections (i.e., reason 2), since no obvious excitation source can be found, that is, it is impossible to determine where the excitation source is in the matrix in formula (1), and without an accurate stray light template, there is no way to accurately calculate the energy magnitude of the stray light at the optical axis pointing position.

[0011] The information disclosed in this background art section is only intended to enhance the overall understanding of the present invention and should not be regarded as an admission or any form of implication that this information constitutes the prior art already known to those of ordinary skill in the art. Summary of the Invention

[0012] The purpose of the present invention is to provide a method and device for suppressing stray light in the cloud image of a geostationary satellite, which can suppress the stray light that the energy of off-field targets is reflected into the field of view multiple times.

[0013] To achieve the above object, the present invention provides a method for suppressing stray light in the geostationary satellite cloud image, which is used to suppress the stray light mainly caused by the multiple reflections of the energy of the off-field target into the field of view, including the following steps:

[0014] S1: Detect the edge of the earth in the geostationary satellite cloud image, separate the earth disk and the cold space image outside the disk, and eliminate foreign object targets;

[0015] S2: Calculate the noise mean value and standard deviation observed in the earth disk appearance;

[0016] S3: Dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light.

[0017] In an embodiment of the present invention, the foreign object targets include the shadow of the secondary mirror support part and the moon image.

[0018] In an embodiment of the present invention, step S2 includes the following sub-steps:

[0019] S201: Segment the disk image after separating the foreign object targets, and the size of each processing unit after segmentation is 5×5 pixels;

[0020] S202: Select a processing unit in the upper left area of the segmented disk image and in the observation area outside the earth disk, and calculate its mean value and uncertainty;

[0021] Among them, the calculation formulas for the mean value and uncertainty of the processing unit are as follows:

[0022]

[0023] Among them, x i is the actual observed value of each pixel in the 5×5 area of the processing unit.

[0024] In an embodiment of the present invention, in step S202, if a foreign object target appears in the processing unit in the upper left area of the segmented disk image, then avoid this processing unit and select a processing unit in other areas where the foreign object target has appeared.

[0025] In an embodiment of the present invention, step S3 includes the following sub-steps:

[0026] S301: After observing stray light (i.e., there is a numerical range) at the processing unit numbered k-1, calculate the predicted value of the adjacent processing unit

[0027]

[0028] Among them, the processing unit number selected in step S202 is 0, and the processing unit number next to it is 1, and so on; is the observed value of the processing unit numbered k; bias k-1 is the deviation caused by stray light in the processing unit numbered k - 1. When estimating, bias K = bias k-1 , bias0 = V mean ;

[0029] S302: Calculate the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ;

[0030] The uncertainty of the processing unit numbered k and the Kalman filter gain K pk are:

[0031]

[0032] Among them, n is the satellite observation noise,

[0033] S303: After obtaining the Kalman filter gain, calculate the true value x k of the processing unit numbered k after filtering:

[0034]

[0035] S304: Update the uncertainty P k of the processing unit numbered k and the true deviation bias k caused by stray light;

[0036] The uncertainty P k of the processing unit numbered k and the true deviation bias k caused by stray light are:

[0037]

[0038] S305: Repeat steps S301 to S304 until the true values of all processing units are obtained, and output the true value image.

[0039] The present invention also provides a stray light suppression device for geostationary satellite cloud images, which is used to suppress the stray light mainly caused by the multiple reflections of the energy of off - field targets into the field of view, including: an image processing module, an initialization module, and a stray light suppression module;

[0040] The image processing module is used to detect the earth's edge in the geostationary satellite cloud image, separate the earth disk and the cold space image outside the disk, and eliminate foreign object targets;

[0041] The initialization module is used to calculate the mean value and standard deviation of the noise observed in the earth disk appearance.

[0042] The stray light suppression module is used to dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light.

[0043] In an embodiment of the present invention, the foreign object target includes the shadow of the secondary mirror support part and the lunar image.

[0044] In an embodiment of the present invention, the initialization module includes the following sub-modules:

[0045] The processing unit segmentation module is used to segment the disk image after separating the foreign object target, and the size of each processing unit after segmentation is 5×5 pixels.

[0046] The initial unit calculation module is used to select a processing unit in the upper left area of the disk image after segmentation and in the observation area of the earth disk appearance, and calculate its mean value and uncertainty.

[0047] Among them, the calculation formulas for the mean value and uncertainty of the processing unit are as follows:

[0048]

[0049] Among them, x i is the actual observation value of each pixel in the 5×5 area of the processing unit.

[0050] In an embodiment of the present invention, in the initial unit calculation module, if a foreign object target appears in the processing unit in the upper left area of the disk image after segmentation, then avoid this processing unit and select a nearby adjacent processing unit.

[0051] In an embodiment of the present invention, the stray light suppression module includes the following sub-modules:

[0052] The estimated value calculation module is used to calculate the estimated value of the adjacent processing unit after observing stray light (that is there is a numerical range) at the processing unit numbered k-1

[0053]

[0054] Among them, the number of the processing unit selected in the initial unit calculation module is 0, the number of the processing unit next to it is 1, and so on; is the observed value of the processing unit numbered k;; bias k-1 is the deviation caused by stray light in the processing unit numbered k-1, and bias is used in the estimationK = bias k-1 , bias0 = V mean ;

[0055] A Kalman filter gain calculation module, configured to calculate the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ;

[0056] The uncertainty of the processing unit numbered k and the Kalman filter gain K pk are:

[0057]

[0058] where n is the satellite observation noise,

[0059] A true value calculation module, configured to calculate the true value x of the processing unit numbered k after filtering k :

[0060]

[0061] An update module, configured to update the uncertainty P of the processing unit numbered k k and the true bias bias caused by stray light k , until the true values of all processing units are obtained;

[0062] The uncertainty P of the processing unit numbered k k and the true bias bias caused by stray light k are:

[0063]

[0064] Compared with the prior art, a stray light suppression method and a suppression device for a geostationary satellite cloud map according to the present invention propose a stray light correction method for a geostationary satellite cloud map based on data filtering for the problem of energy reflection stray light of off-field targets in a full-disk remote sensing image of a geostationary satellite. The energy response of the optical axis pointing to the target within the field of view is regarded as the true value that should be obtained by observation, and the stray light is regarded as the noise at the observation moment. At this time, "actual observation value = target observation true value + noise (stray light)", and then the characteristics of the noise (stray light) itself are used to construct highly adaptable filtering parameters by statistical analysis, and then the data filter is used to effectively estimate and suppress the noise, and the stray light superimposed on the target is removed by the noise suppression method to achieve the purpose of stray light correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is a schematic diagram of the stray light of a full-disk remote sensing image of a geostationary satellite;

[0066] Figure 2 It is a schematic diagram of a stray light template used in the template method of the prior art;

[0067] Figure 3 It is a flowchart of the steps of a method for suppressing stray light in a geostationary satellite cloud image according to an embodiment of the present invention;

[0068] Figure 4 It is a schematic diagram of a lunar infrared image;

[0069] Figure 5 It is a schematic diagram of the shadow image of the secondary mirror support rod;

[0070] Figure 6 It is a schematic diagram of a device for suppressing stray light in a geostationary satellite cloud image according to an embodiment of the present invention. Detailed Embodiment

[0071] The following combines the accompanying drawings to describe in detail the specific embodiments of the present invention, but it should be understood that the protection scope of the present invention is not limited by the specific embodiments.

[0072] Unless otherwise clearly stated, in the whole specification and claims, the term "comprising" or its variations such as "including" or "having" etc. will be understood to include the stated elements or components, without excluding other elements or other components.

[0073] As Figures 3 to 5 shown, a method for suppressing stray light in a geostationary satellite cloud image according to a preferred embodiment of the present invention is used to suppress stray light mainly caused by multiple reflections of off-field target energy into the field of view. In the present invention, the state x in the time series refers to the actual observed value of the true value of the target observation of each pixel point on the cloud image (manifested as a spatial relationship on the image) is the true cloud image count value including the observed true value of the observation together with the stray light.

[0074] As Figure 3 shown, a method for suppressing stray light in a geostationary satellite cloud image of the present invention includes the following steps:

[0075] S1: Detect the edge of the earth in the geostationary satellite cloud image, separate the earth disk and the cold space image outside the disk, and eliminate foreign object targets to avoid errors in the statistical analysis of stray light caused by foreign object targets.

[0076] The geostationary satellite cloud image is the entire image including the earth target and the cosmic background space observation.

[0077] Completing the separation of the Earth disk and the cold space is the basis for observing the noise covariance matrix, determining the stray light mean value in a small range, determining the true value of the Earth's edge observation, and calculating the estimated error covariance matrix. Therefore, the first step S1 of the present invention is the separation of the Earth disk and the cold space images.

[0078] The foreign object targets to be removed mainly include the following two types: 1. The shadow of the secondary mirror support part. As Figure 5 shown, when approaching midnight, the shadow of the secondary mirror support part may appear on the cloud image due to direct sunlight. The appearance of the shadow of the secondary mirror support part has a time pattern, generally appearing within 12 ± 2 hours at midnight. Therefore, the images during this period are removed (not processed) by time; 2. The moon image. As Figure 4 shown, the moon can be observed in the cold space area outside the Earth disk in the cloud image of the geostationary satellite. Therefore, the moon position calculation software can be used to calculate its position, and the area containing the moon is avoided as the starting area, and no special processing is required when performing stray light processing (refer to the Earth image).

[0079] These targets will affect the observation noise statistics. Therefore, it is necessary to focus on studying the use of various filtering technologies, image processing technologies, etc. to remove the interference of foreign object targets in the cold space such as the moon and the shadow of the support rod, so as to ensure the accuracy and availability of the cold space background data obtained by separation.

[0080] S2: Calculate the noise mean value and standard deviation observed on the Earth disk appearance.

[0081] Specifically, this step S2 includes the following sub-steps:

[0082] S201: Segment the disk image after separating the foreign object targets, and the size of each processing unit after segmentation is 5 × 5 pixels.

[0083] S202: Select a processing unit (5 × 5) in the upper left area of the segmented disk image and in the observation area outside the Earth disk, and calculate its mean value and uncertainty (standard deviation).

[0084] If a foreign object target has appeared in the processing unit in the upper left area of the segmented disk image, then avoid this processing unit and select a processing unit in other areas where foreign object targets have appeared.

[0085] The calculation formulas for the mean value and uncertainty of the processing unit are as follows:

[0086]

[0087] Among them, x i is the actual observed value of each pixel in the 5 × 5 area within the processing unit.

[0088] Since the natural target outside the earth disk is cold air, the output response of the remote sensing instrument is 0, that is, the image shows pure black. Therefore, the mean value and uncertainty calculated in the processing unit both come from stray light.

[0089] S3: Utilize the characteristic that stray light has low-frequency variation to dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light.

[0090] For the convenience of description and calculation, the processing unit selected in the above step S202 is numbered 0, the adjacent processing unit is numbered 1, and so on. In the following calculation process, k starts from 1.

[0091] Specifically, this step S3 includes the following sub-steps:

[0092] S301: Based on the fact that stray light is a low-frequency signal and changes slowly, after observing stray light (i.e., there is a numerical range) at the processing unit numbered k - 1, calculate the predicted value of the adjacent processing unit

[0093]

[0094] where, is the observed value of the processing unit numbered k. The observed value is the value of the captured image, and its physical meaning represents the energy of the target; bias k-1 is the deviation caused by stray light in the previous processing unit. Because stray light changes slowly, bias K is set to bias k-1 , bias0 = V mean .

[0095] Among them, the observation state transition matrix is the set matrix of the predicted values of each processing unit.

[0096] S302: Calculate the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ;

[0097] The uncertainty of the processing unit numbered k is:

[0098]

[0099] where, n is the satellite observation noise, obtained through on-orbit testing, and is a constant value in the present invention, P0 = std.

[0100] The Kalman filter gain K pk of the processing unit numbered k is

[0101]

[0102] S303: After obtaining the Kalman filter gain, calculate the true value x of the processing unit numbered k after filtering k :

[0103]

[0104] S304: Update the uncertainty P of the processing unit numbered k k and the actual bias caused by stray light k , used to calculate the estimated value of the processing unit numbered k+1.

[0105] The uncertainty P of the processing unit number k k for:

[0106]

[0107] The actual bias caused by stray light of the processing unit number k k for:

[0108]

[0109] S305: Repeat steps S301 to S304 until the true values of all processing units are obtained and a true value image is output. The true value is the image with noise (stray light) removed, and the true value image is the image with stray light interference suppressed.

[0110] like Figure 6 As shown, a stray light suppression device for geostationary satellite cloud images in a preferred embodiment of the present invention is used to suppress stray light mainly caused by multiple reflections of target energy outside the field of view into the field of view. The device includes: an image processing module 1, an initialization module 2 and a stray light suppression module 3.

[0111] The image processing module 1 is used to detect the edge of the earth in the geostationary orbit satellite cloud image, separate the earth disk and the cold space image outside the disk, and remove foreign objects to avoid errors caused by foreign objects in the statistical analysis of stray light.

[0112] The geostationary satellite cloud image is a complete image that includes space observations of Earth targets and cosmic background.

[0113] Completing the separation of the earth disk and cold space is the basis for observing the noise covariance matrix, determining the mean of stray light in a small range, determining the true value of the earth edge observation, and calculating the estimation error covariance matrix. Therefore, the image processing module 1 of the present invention is used to separate the earth disk and cold space images.

[0114] The foreign body targets to be removed mainly include the following two types: 1. The shadow of the secondary mirror supporting parts, such asFigure 4 As shown, near midnight, due to direct sunlight, the shadow of the secondary mirror support part may appear on the cloud image. The appearance of the shadow of the secondary mirror support part follows a time pattern, generally appearing within 12 ± 2 hours in the middle of the night. Therefore, the images during this period are removed (not processed) by time; 2. Lunar images, as Figure 4 shown, the moon can be observed in the cold space area outside the earth disk in the cloud image of a geostationary satellite. Therefore, the lunar position calculation software can be used to calculate its position, and the area containing the moon is avoided as the starting area, and no special treatment is required during stray light processing (refer to the earth image).

[0115] These targets will affect the observation noise statistics. Therefore, it is necessary to focus on studying the use of various filtering technologies, image processing technologies, etc. to remove the interference of cold space foreign object targets such as the moon and the shadow of the support rod, and ensure that the separated cold space background data is accurate and available.

[0116] The initialization module 2 is used to calculate the mean value and standard deviation of the noise observed outside the earth disk.

[0117] Specifically, the initialization module 2 includes the following sub-modules:

[0118] The processing unit segmentation module 201 is used to segment the disk image after separating the foreign object targets. The size of each processing unit after segmentation is 5×5 pixels.

[0119] The initial unit calculation module 202 is used to select a processing unit (5×5) in the upper left area of the disk image after segmentation and in the observation area outside the earth disk, and calculate its mean value and uncertainty (standard deviation).

[0120] If a foreign object target has ever appeared in the processing unit in the upper left area of the disk image after segmentation, then avoid this processing unit and select the processing unit in other areas where the foreign object target has appeared.

[0121] The mean value V mean and the uncertainty std of the processing unit are calculated as follows:

[0122]

[0123] where x i is the actual observed value of each pixel in the 5×5 area within the processing unit.

[0124] Since the natural target outside the earth disk is cold space, the output response of the remote sensing instrument is 0, that is, the image shows pure black. Therefore, the mean value and uncertainty calculated in the processing unit are all from stray light.

[0125] The stray light suppression module 3 is used to dynamically calculate the observation state transition matrix based on the characteristic that the stray light is a low-frequency change, perform Kalman filtering on the geostationary satellite cloud image, and suppress the stray light.

[0126] For the convenience of description and calculation, the processing unit selected in the above initial unit calculation module 202 is numbered 0, the processing unit next to it is numbered 1, and so on. In the following calculation process, k starts from 1.

[0127] Specifically, the stray light suppression module 3 includes the following sub-modules:

[0128] The predicted value calculation module 301 is used to calculate the predicted value of the adjacent processing unit after observing the stray light (i.e., there is a numerical range) at the processing unit numbered k - 1 based on the fact that the stray light is a low-frequency signal and changes slowly.

[0129]

[0130] Among them, is the observed value of the processing unit numbered k. The observed value is the value of the captured image, and its physical meaning represents the energy of the target; bias k-1 is the deviation caused by the stray light in the previous processing unit. Because the stray light changes slowly, bias K is set to bias k-1 , bias0 = V mean .

[0131] The observation state transition matrix is the set matrix of the predicted values of each processing unit.

[0132] The Kalman filter gain calculation module 302 is used to calculate the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ;

[0133] The uncertainty of the processing unit numbered k is:

[0134]

[0135] Among them, n is the satellite observation noise, which is obtained through on-orbit testing and is a constant value in the present invention.

[0136] The Kalman filter gain K pk of the processing unit numbered k is

[0137]

[0138] The true value calculation module 303 is used to calculate the true value x of the processing unit with number k after filtering, after obtaining the Kalman filter gain. k :

[0139]

[0140] The update module 304 is used to update the uncertainty P of the processing unit with number k k and the true deviation bias caused by stray light k , and is used to calculate the predicted value of the processing unit with number k + 1 until the true values of all processing units are obtained, and output the true value image. The true value is the one after removing the noise (stray light), and the true value image is the image that suppresses the interference of stray light.

[0141] The uncertainty P of the processing unit with number k k is:

[0142]

[0143] The true deviation bias caused by stray light of the processing unit with number k k is:

[0144]

[0145] The foregoing description of the specific exemplary embodiments of the present invention is for the purposes of illustration and exemplification. These descriptions are not intended to limit the invention to the precise forms disclosed, and it is apparent that many changes and variations are possible in light of the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical applications, so that those skilled in the art can implement and utilize various different exemplary embodiments of the present invention, as well as various different selections and changes. The scope of the present invention is intended to be defined by the claims and their equivalents.

Claims

1. A method for suppressing stray light in geostationary satellite cloud images, which is used to suppress stray light mainly caused by multiple reflections of off-field target energy into the field of view, and is characterized in that, Including the following steps: S1: Detect the edge of the Earth in the geostationary satellite cloud image, separate the Earth disk and the cold space image outside the disk, and eliminate foreign object targets; S2: Calculate the mean value and standard deviation of the noise observed in the Earth disk; S3: Dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light; Step S2 includes the following sub-steps: S201: Segment the disk image after separating the foreign object targets, and the size of each processing unit after segmentation is 5×5 pixels; S202: Select a processing unit in the upper left area of the segmented disk image and in the observation area outside the Earth disk, and calculate its mean value and uncertainty; Among them, the calculation formulas for the mean value and uncertainty of the processing unit are as follows: where x i is the actual observation value of each pixel in the 5×5 area of the processing unit; Step S3 includes the following sub-steps: S301: After observing stray light at the processing unit numbered k - 1, calculate the predicted values of adjacent processing units. Among them, the processing unit number selected in step S202 is 0, and the processing unit number next to it is 1, and so on; is the observed value of the processing unit numbered k; bias k-1 is the deviation caused by stray light in the processing unit numbered k - 1. When estimating, bias K = bias k-1 , bias0 = V mean ; S302: Calculate the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ; Uncertainty of the processing unit numbered k and the Kalman filter gain K pk are as follows: where n is the satellite observation noise, S303: After obtaining the Kalman filter gain, calculate the true value x of the processing unit with number k after filtering k : S304: Update the uncertainty P of the processing unit with number k k and the true deviation bias caused by stray light k ; Uncertainty P of the processing unit with number k k and the true deviation bias caused by stray light k is as follows: S305: Repeat steps S301 to S304 until the true values of all processing units are obtained, and output the true value image.

2. The stray light suppression method for geostationary satellite cloud images according to claim 1, wherein The foreign object targets include the shadow of the secondary mirror support part and the lunar image.

3. The stray light suppression method for geostationary satellite cloud images according to claim 1, wherein, In step S202, if a foreign object target appears in the processing unit in the upper left area of the segmented disk image, avoid this processing unit and select a processing unit in other areas where foreign object targets have appeared.

4. A stray light suppression device for geostationary satellite cloud images, which is used to suppress the stray light mainly caused by the multiple reflections of the energy of off-field targets into the field of view. It is characterized in that, Including: An image processing module, an initialization module, and a stray light suppression module; The image processing module is used to detect the edge of the Earth in the geostationary satellite cloud image, separate the Earth disk and the cold space image outside the disk, and eliminate foreign object targets; The initialization module is used to calculate the mean value and standard deviation of the noise observed in the Earth disk; The stray light suppression module is used to dynamically calculate the observation state transition matrix, perform Kalman filtering on the geostationary satellite cloud image, and suppress stray light; The initialization module includes the following sub-modules: A processing unit segmentation module, which is used to segment the disk image after separating the foreign object targets, and the size of each processing unit after segmentation is 5×5 pixels; An initial unit calculation module, which is used to select a processing unit in the upper left area of the segmented disk image and in the observation area outside the Earth disk, and calculate its mean value and uncertainty; Among them, the calculation formulas for the mean value and uncertainty of the processing unit are as follows: where x i is the actual observation value of each pixel in the 5×5 area of the processing unit; The stray light suppression module includes the following sub-modules: The predicted value calculation module is used to calculate the predicted values of adjacent processing units after stray light is observed at the processing unit numbered k-1. Among them, the processing unit number selected in the initial unit calculation module is 0, the processing unit number next to it is 1, and so on; is the observation value of the processing unit with number k; bias k-1 is the deviation caused by stray light in the processing unit with number k - 1. When estimating, bias K = bias k-1 , bias0 = V mean ; A Kalman filter gain calculation module for calculating the uncertainty of the processing unit numbered k and the Kalman filter gain K pk ; Uncertainty of the processing unit with number k and the Kalman filter gain K pk are as follows: where n is the satellite observation noise, True value calculation module, used to calculate the true value x of the processing unit with the filtered number k k : An update module for updating the uncertainty P of the processing unit with number k k and the true deviation bias caused by stray light k , until the true values of all processing units are obtained and a true value image is output; Uncertainty P of the processing unit numbered k k and the true deviation bias caused by stray light k are as follows:

5. The stray light suppression device for geostationary satellite cloud images according to claim 4, wherein The foreign object targets include the shadow of the secondary mirror support part and the lunar image.

6. The stray light suppression device for geostationary satellite cloud images according to claim 4, wherein In the initial unit calculation module, if a foreign object target appears in the processing unit in the upper left area of the segmented disk image, avoid this processing unit and select a nearby adjacent processing unit.

Images