A method and system for testing and correcting the antenna pattern of spaceborne SAR

Abstract

The present invention proposes a method and system for testing and correcting the antenna pattern of a spaceborne SAR. The method involves marking the pixel ranges of multiple marked water areas in each SAR image, training a PSPNet network model, and obtaining a trained PSPNet network model. Each real-time Ka-band SAR image is input into the trained PSPNet network model for water area extraction. The extracted pixel ranges of the water area are eliminated from each real-time SAR image by a mean filtering method, obtaining each real-time Ka-band SAR image after water area elimination. The antenna direction pixel coordinate sequence of the Ka-band SAR image after water area elimination is obtained by mean calculation and fourth-order curve fitting. The Ka-band SAR image to be corrected is input, and the antenna pattern correction is performed in combination with the antenna direction pixel coordinate sequence to obtain a corrected Ka-band SAR image. The present invention solves the problem of automatic extraction and correction of the range antenna pattern of a spaceborne SAR in the Ka-band strip mode.

Description

Technical Field

[0001] The invention belongs to the field of radar technology, and in particular relates to a method for testing and correcting the radiation pattern of a space-borne SAR antenna. Background Art

[0002] Synthetic Aperture Radar (SAR) is an active microwave sensor that offers all-day, all-weather, and high-resolution observation capabilities. Spaceborne SAR, powered by satellites, leverages its ability to observe Earth over large areas and has played a significant role in resource surveys, urban planning, and other fields.

[0003] The antenna pattern describes the relationship between the SAR antenna's radiated energy and its spatial orientation. When radiation deviates from the antenna's direction, the radiated energy gradually decreases, resulting in darker edges in the range direction of spaceborne SAR images, ultimately affecting the imaging radiation quality. Therefore, accurately correcting for the uneven image radiation caused by the antenna pattern during imaging processing is crucial for improving SAR image radiation quality.

[0004] The prerequisite for correcting the antenna pattern is to accurately extract the antenna pattern. Currently, the methods for extracting the range antenna pattern of spaceborne SAR mainly include the standard reflector extraction method, the ground receiver extraction method, and the distributed target method. Among them, the distributed target method has the advantages of low cost and high accuracy, and is a commonly used method for extracting the antenna pattern of spaceborne SAR. The Amazon rainforest has a large area and very stable scattering characteristics, and has gradually become an ideal distributed target for in-orbit extraction of the range antenna pattern of spaceborne SAR. Patent CN 117148289A proposes a polarization calibration method based on the tropical rainforest. The polarization calibration parameters are estimated through non-iterative calculations, and good results have been achieved in the multi-polarization calibration of C-band spaceborne SAR. The paper "Radiometric Calibration Algorithm for Spaceborne SAR in the Amazon Rainforest" (Yun Risheng et al., Radar Science and Technology, Issue 2, 2007, pp. 139-143+148) proposes an algorithm for extracting antenna patterns from spaceborne SAR in the Amazon rainforest. Through L-band experimental simulations, the algorithm demonstrates the high accuracy of extracting two-way elevation antenna patterns in this band. Patent CN 117392549 A proposes a method and system for extracting target characteristics based on high-resolution SAR images. This method uses multi-level SAR products from the rainforest region to perform radiometric calibration parameter correction operations, obtaining calibration parameter correction values, which are then used to perform radiometric correction on other multi-level high-resolution SAR target images. The aforementioned patents and literature demonstrate the high accuracy of extracting antenna patterns based on Amazon rainforest imagery in different bands. However, existing methods for extracting and accurately correcting SAR antenna range patterns still have the following problems:

[0005] Automating the accurate removal of non-uniform areas is difficult. The Amazon rainforest contains non-uniform targets, primarily rivers, which can affect the accuracy of antenna pattern extraction. Furthermore, water exhibits different characteristics in SAR imagery at different wavelengths. Traditional SAR image water extraction methods typically assume that water is a "pure black" area, making accurate removal difficult.

[0006] The influence of the range pattern on SAR imagery is time-varying. Strip-mode imaging takes a long time. During the SAR imaging process, the weighting of the range pattern on the echoes from the imaged scene can vary over time due to factors such as the undulating terrain of the illuminated area and orbital characteristics. This characteristic causes the darker areas at the edges and the brighter areas in the center of the entire image strip to vary over time, making it difficult to uniformly correct the range pattern across the entire image. Summary of the Invention

[0007] In order to solve the problem of automatic extraction and correction of spaceborne SAR range antenna pattern in strip mode, the present invention proposes a spaceborne SAR antenna pattern testing and correction method and system.

[0008] The technical solution of the method of the present invention is a method for testing and correcting the antenna pattern of a spaceborne SAR, comprising the following steps:

[0009] Step 1: Acquire multiple SAR images containing land and water areas, and mark the pixel ranges of multiple marked water areas in each SAR image.

[0010] Step 2: Construct a PSPNet network model and input each SAR image into the PSPNet network model to extract the water area. Multiple extracted water area pixel ranges are obtained for each SAR image. A loss function model is constructed based on the multiple labeled water area pixel ranges for each SAR image. The trained PSPNet network model is obtained by training using the gradient descent method.

[0011] Step 3: Obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model to extract the water area, and obtain multiple extracted water area pixel ranges of each real-time SAR image;

[0012] Step 4: Combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image by using the mean filtering method to obtain each real-time Ka-band SAR image after water area removal;

[0013] Step 5: Calculate the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and horizontal coordinate calculation;

[0014] Step 6: Input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal to obtain the corrected Ka-band SAR image;

[0015] Preferably, the loss function model in step 2 is specifically defined as follows:

[0016]

[0017] Where N is the total number of pixels in each SAR image, is the pixel value of the ith marker in each SAR image, o i is the i-th extracted pixel value of each SAR image. The default log function uses the natural constant e as the base.

[0018] Preferably, the mean value calculation in step 5 is as follows:

[0019] Define the row pixels of each real-time Ka-band SAR image after water body area removal as the range direction and the column pixels as the azimuth direction, calculate the average value of each row of pixels in each real-time Ka-band SAR image after water body area removal, and obtain the average pixel value of each row of each real-time Ka-band SAR image after water body area removal;

[0020] The fourth-order curve fitting described in step 5 is as follows:

[0021] The fourth-order curve fitting is performed on the multi-row mean pixels of each real-time Ka-band SAR image after the water body area is removed, and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after the water body area is removed is obtained;

[0022] The horizontal coordinates described in step 5 are calculated as follows:

[0023] The antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after the water body area is removed is calculated based on the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after the water body area is removed;

[0024] Furthermore, the horizontal axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal is the row number of the image, and the vertical axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal is the pixel after the fourth-order curve fitting per mean pixel;

[0025] The calculation is performed by combining the horizontal coordinates of the antenna direction pixel coordinates of each real-time Ka-band SAR image after water area removal. The specific formula is as follows:

[0026]

[0027] Where θ is the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal, R s is the slant distance between the satellite and the center of each real-time Ka-band SAR image after the water area is removed; R0 is the distance between the satellite and the sub-satellite point in the imaging parameters of each real-time Ka-band SAR image after the water area is removed; θ roll The satellite side swing angle in the imaging parameters of each real-time Ka-band SAR image after the water area is removed;

[0028] Preferably, the antenna pattern correction is performed in step 6, specifically as follows:

[0029] Step 6.1: After removing the water area to be corrected, the Ka-band SAR image is divided into multiple image blocks to be corrected along the columns. The mean of each row of pixels in the divided image blocks to be corrected is calculated to obtain the mean pixel value of each row of the divided image blocks to be corrected. A second-order curve is fitted on the multi-row mean pixels of the divided image blocks to be corrected to obtain the antenna direction pixel coordinate sequence of the Ka-band SAR image to be corrected;

[0030] Step 6.2: Calculate the offset of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water area removal and the antenna direction pixel coordinate sequence of the Ka-band SAR image to be corrected using the offset model of the antenna pattern changing with imaging time;

[0031] The calculation is performed using the offset model of the antenna pattern changing with imaging time. The specific process is as follows:

[0032] y=a n x n +a n-1 x n-1 +…+a0x+a0

[0033] Among them, a mis the m-order offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected, where m represents the order of the model polynomial, m∈[1,n], x is the serial number of the divided multiple image blocks to be corrected; y is the position of the row pixel pointed to by the pixel coordinate sequence in the antenna direction;

[0034] Step 6.3: Input the Ka-band SAR image to be corrected and combine it with the offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected to perform antenna pattern correction to obtain the corrected Ka-band SAR image;

[0035] The antenna pattern correction is performed in combination with the offset coefficient. The specific process is as follows:

[0036] Input the pixel at row x and column y in the Ka-band SAR image to be corrected. The corresponding pixel value is g. before (x,y), x∈[1,X], y∈[1,Y], X represents the number of rows in the Ka-band SAR image to be corrected, and Y represents the number of columns in the Ka-band SAR image to be corrected;

[0037] The pixel value corresponding to the pixel position (x, y) of the corrected Ka-band SAR image is g after (x,y):

[0038] g after (x,y)=g before (x,y)÷r(x)

[0039] Wherein, r(x) is the offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected;

[0040] The present invention also proposes a spaceborne SAR antenna pattern testing and correction system, comprising:

[0041] Image acquisition and marking module, used to acquire multiple SAR images containing land and water areas, and mark the pixel range of multiple marked water areas in each SAR image

[0042] The network training module is used to build a PSPNet network model. Each SAR image is input into the PSPNet network model to extract the water area, obtain multiple extracted water area pixel ranges of each SAR image, and construct a loss function model based on the multiple labeled water area pixel ranges of each SAR image. The trained PSPNet network model is obtained by training using the gradient descent method.

[0043] A real-time SAR image extraction module is used to obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model to extract the water area, and obtain multiple extracted water area pixel ranges for each real-time SAR image;

[0044] A mean filtering module is used to combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image through a mean filtering method to obtain a Ka-band SAR image after the water area is removed from each real-time SAR image;

[0045] The antenna beam angle calculation module is used to calculate the antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and abscissa calculation;

[0046] The antenna pattern correction module is used to input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal, so as to obtain the corrected Ka-band SAR image;

[0047] The advantages of the present invention are:

[0048] This paper introduces a deep learning model into antenna pattern extraction technology and establishes a deep learning-based water extraction model using a Ka-band SAR image slice dataset. This model provides good support for the removal of non-uniform media in Amazon rainforest images in this band, solves the problem that the threshold method is difficult to effectively remove water bodies in this band, and improves the accuracy of antenna pattern extraction.

[0049] This method divides the SAR image to be corrected into blocks along the azimuth direction and uses the peak position of the range-direction pixel value curve within each block to determine how the weighted range antenna pattern for the imaged scene echo changes over imaging time. Using antenna pattern curves extracted from tropical rainforests to correct the image, this method solves the problem of antenna pattern variation with azimuth in strip mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 : A flow chart of a method according to an embodiment of the present invention.

[0051] Figure 2 : Water extraction effect of the embodiment of the present invention.

[0052] Figure 3 : Schematic diagram of mean filtering adopted in an embodiment of the present invention.

[0053] Figure 4 : The embodiment of the present invention is a fitting curve from the tropical rainforest image distance to the pixel value and the corresponding antenna radiation pattern.

[0054] Figure 5 : Schematic diagram of determining the time-varying relationship of the antenna pattern according to an embodiment of the present invention.

[0055] Figure 6 : Image comparison results before and after correction of the image antenna pattern according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0057] In specific implementation, the method proposed in the technical solution of the present invention can be automatically run by those skilled in the art using computer software technology. System devices that implement the method, such as computer-readable storage media that store the corresponding computer program of the technical solution of the present invention and computer equipment that runs the corresponding computer program, should also be within the scope of protection of the present invention.

[0058] The experimental scenario for this embodiment of the present invention involved using stripe-mode imagery of the Amazon rainforest captured by the Luojia-2 01 Ka-band SAR satellite for water extraction. Antenna pattern correction was then performed using SAR satellite imagery of the Amazon basin after water removal. The imagery had a range width of 6.873 km and an azimuth width of 208.931 km. The terrain types included mountains, farmland, buildings, and water bodies. The imagery was divided into five regions along the azimuth direction.

[0059] The following is combined with Figure 1-6 The specific implementation of the method of the present invention is a method for extracting and correcting the range antenna pattern of a SAR strip mode, which is as follows:

[0060] like Figure 1 FIG. 1 is a flow chart of a method according to an embodiment of the present invention.

[0061] Step 1: Acquire multiple SAR images containing land and water areas, and mark the pixel ranges of multiple marked water areas in each SAR image. In this embodiment, the size of each SAR image to be marked is 1024×1024 pixels.

[0062] Step 2: Construct a PSPNet network model, input each SAR image into the PSPNet network model to extract the water area, obtain multiple extracted water area pixel ranges of each SAR image, and construct a loss function model based on the multiple labeled water area pixel ranges of each SAR image. Train the model using the gradient descent method to obtain a trained PSPNet network model. The maximum number of iterations set in the training process is i, the initialization learning rate is p, and the number of samples transferred in a single training is set to q. In this embodiment, i = 20, p = 0.001, and q = 20, until the number of iterations is completed.

[0063] The loss function model described in step 2 is specifically defined as follows:

[0064]

[0065] Where N is the total number of pixels in each SAR image, is the pixel value of the i-th marker in each SAR image, o i is the i-th extracted pixel value of each SAR image. The default log function uses the natural constant e as the base.

[0066] Step 3: Obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model for water area extraction, and obtain multiple extracted water area pixel ranges for each real-time SAR image. Figure 2 The results before and after water extraction from Ka-band SAR images are shown;

[0067] Step 4: Combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image through the mean filtering method to obtain each real-time Ka-band SAR image after water area removal. In this example, the filter window size is set to 9. The filtering process diagram is as follows: Figure 3 As shown;

[0068] Step 5: Calculate the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and horizontal coordinate calculation;

[0069] The mean value calculation described in step 5 is as follows:

[0070] Define the row pixels of each real-time Ka-band SAR image after water body area removal as the range direction and the column pixels as the azimuth direction, calculate the average value of each row of pixels in each real-time Ka-band SAR image after water body area removal, and obtain the average pixel value of each row of each real-time Ka-band SAR image after water body area removal;

[0071] The fourth-order curve fitting described in step 5 is as follows:

[0072] The fourth-order curve fitting is performed on the multi-row mean pixels of each real-time Ka-band SAR image after the water body area is removed, and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after the water body area is removed is obtained, as shown in the figure: Figure 4 As shown, the antenna direction pixel coordinate sequence is represented by a fitted curve;

[0073] The horizontal coordinates described in step 5 are calculated as follows:

[0074] The antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after the water body area is removed is calculated based on the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after the water body area is removed;

[0075] The horizontal axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal is the row number of the image, and the vertical axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal is the pixel after the fourth-order curve fitting per mean pixel;

[0076] The calculation is performed by combining the horizontal coordinates of the antenna direction pixel coordinates of each real-time Ka-band SAR image after water area removal. The specific formula is as follows:

[0077]

[0078] Where θ is the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal, R s is the slant distance between the satellite and the center of each real-time Ka-band SAR image after the water area is removed; R0 is the distance between the satellite and the sub-satellite point in the imaging parameters of each real-time Ka-band SAR image after the water area is removed; θ roll The satellite side swing angle in the imaging parameters of each real-time Ka-band SAR image after the water area is removed;

[0079] Step 6: Input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal to obtain the corrected Ka-band SAR image;

[0080] Perform antenna pattern correction as described in step 6, as follows:

[0081] Step 6.1: After removing the water area to be corrected, the Ka-band SAR image is divided into a plurality of image blocks to be corrected along the columns. In this embodiment, Figure 5 As shown in FIG, the SAR image is divided into 5 image blocks. The mean of each row of pixels in the divided image blocks to be corrected is calculated to obtain the mean pixel of each row in the divided image blocks to be corrected. The mean pixel of each row in the divided image blocks to be corrected is fitted with a second-order curve to obtain the pixel coordinate sequence in the antenna direction of the Ka-band SAR image to be corrected.

[0082] Step 6.2: The antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water area removal and the antenna direction pixel coordinate sequence of the Ka-band SAR image to be corrected are calculated using the offset model of the antenna pattern changing with imaging time to calculate the offset, such as Figure 5 As shown in Figure 2, the deviation of the antenna pattern of the Ka-band SAR image to be corrected changes with the imaging time.

[0083] The calculation is performed using the offset model of the antenna pattern changing with imaging time. The specific process is as follows:

[0084] y=a n x n +a n-1 x n-1 +…+a0x+a0

[0085] Among them, a m is the m-order offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected, where m represents the order of the model polynomial, m∈[1,n], x is the serial number of the divided multiple image blocks to be corrected; y is the position of the row pixel pointed to by the pixel coordinate sequence in the antenna direction;

[0086] Step 6.3: Input the Ka-band SAR image to be corrected and combine it with the offset coefficient of the mean pixel of each row of the divided multiple image blocks to be corrected to perform antenna pattern correction to obtain the corrected Ka-band SAR image, as shown in the figure. Figure 6 The following figure shows the comparison of image effects before and after correction;

[0087] like Figure 6As shown, the SAR image on the left is a Ka-band SAR image to be corrected. It has the characteristics of dark sides and bright center in the range direction, and the bright part in the middle is distributed in a curve in the azimuth direction. After the antenna pattern correction is performed by the method proposed in the present invention, the water part in the image is first removed, which reduces the error effect on the antenna pattern calculation. In addition, the polynomial curve fitting can better fit the Ka-band SAR antenna pattern. Therefore, the overall brightness of the corrected SAR image is uniform. The method proposed in the present invention effectively solves the problem of uneven SAR image brightness caused by the change of the antenna pattern with azimuth in the strip mode.

[0088] The antenna pattern correction is performed in combination with the offset coefficient. The specific process is as follows:

[0089] Input the pixel at row x and column y in the Ka-band SAR image to be corrected. The corresponding pixel value is g. before (x,y), x∈[1,X], y∈[1,Y], X represents the number of rows in the Ka-band SAR image to be corrected, and Y represents the number of columns in the Ka-band SAR image to be corrected;

[0090] The pixel value corresponding to the pixel position (x, y) of the corrected Ka-band SAR image is g after (x,y):

[0091] g after (x,y)=g before (x,y)÷r(x)

[0092] Wherein, r(x) is the offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected;

[0093] The technical solution of the system embodiment of the present invention is a spaceborne SAR antenna pattern testing and correction system, comprising:

[0094] An image acquisition and marking module is used to acquire multiple SAR images containing land and water areas, and mark the pixel ranges of multiple marked water areas in each SAR image;

[0095] The network training module is used to build a PSPNet network model. Each SAR image is input into the PSPNet network model to extract the water area, obtain multiple extracted water area pixel ranges of each SAR image, and construct a loss function model based on the multiple labeled water area pixel ranges of each SAR image. The trained PSPNet network model is obtained by training using the gradient descent method.

[0096] A real-time SAR image extraction module is used to obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model to extract the water area, and obtain multiple extracted water area pixel ranges for each real-time SAR image;

[0097] A mean filtering module is used to combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image through a mean filtering method to obtain a Ka-band SAR image after the water area is removed from each real-time SAR image;

[0098] The antenna beam angle calculation module is used to calculate the antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and abscissa calculation;

[0099] The antenna pattern correction module is used to input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal, so as to obtain the corrected Ka-band SAR image;

[0100] The image acquisition and marking module, network training module, real-time SAR image extraction module, mean filter module, antenna beam angle calculation module, and antenna pattern correction module are all deployed on the server;

[0101] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0102] It should be understood that the above description of the embodiments is relatively detailed and cannot be regarded as limiting the scope of protection of the patent of the present invention. Under the guidance of the present invention, ordinary technicians in this field can also make substitutions or modifications without departing from the scope of protection of the claims of the present invention, which all fall within the scope of protection of the present invention. The scope of protection requested by the present invention shall be based on the attached claims.

Claims

1. A method for testing and correcting the antenna pattern of a spaceborne SAR, characterized in that: The following steps are involved: Step 1: Acquire multiple SAR images containing land and water areas, and mark the pixel ranges of multiple marked water areas in each SAR image. Step 2: Construct a PSPNet network model and input each SAR image into the PSPNet network model to extract the water area. Multiple extracted water area pixel ranges are obtained for each SAR image. A loss function model is constructed based on the multiple labeled water area pixel ranges for each SAR image. The trained PSPNet network model is obtained by training using the gradient descent method. Step 3: Obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model to extract the water area, and obtain multiple extracted water area pixel ranges of each real-time SAR image; Step 4: Combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image by using the mean filtering method to obtain each real-time Ka-band SAR image after water area removal; Step 5: Calculate the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and horizontal coordinate calculation; Step 6: Input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal to obtain the corrected Ka-band SAR image.

2. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 1, wherein: The loss function model described in step 2 is specifically defined as follows: Where N is the total number of pixels in each SAR image, is the pixel value of the ith marker in each SAR image, o i is the i-th extracted pixel value of each SAR image. The default log function uses the natural constant e as the base.

3. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 2, wherein: The mean value calculation described in step 5 is as follows: The row pixels of each real-time Ka-band SAR image after water body removal are defined as the range direction, and the column pixels are defined as the azimuth direction. The mean of each row of pixels in each real-time Ka-band SAR image after water body removal is calculated to obtain the mean pixel of each row of each real-time Ka-band SAR image after water body removal.

4. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 3, wherein: The fourth-order curve fitting described in step 5 is as follows: The fourth-order curve fitting is performed on the multi-row mean pixels of each real-time Ka-band SAR image after water body area removal to obtain the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal.

5. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 4, wherein: The horizontal coordinates described in step 5 are calculated as follows: The antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after water body area removal is obtained by combining the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after water body area removal.

6. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 5, wherein: The horizontal axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after the water body area is removed is the row number of the image, and the vertical axis of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after the water body area is removed is the pixel after the fourth-order curve fitting per mean pixel.

7. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 6, wherein: The calculation is performed by combining the horizontal coordinates of the antenna direction pixel coordinates of each real-time Ka-band SAR image after water area removal. The specific formula is as follows: Where θ is the antenna beam angle of the horizontal coordinate of the pixel coordinate in the antenna direction of each real-time Ka-band SAR image after water area removal, R s is the slant distance between the satellite and the center of each real-time Ka-band SAR image after the water area is removed; R0 is the distance between the satellite and the sub-satellite point in the imaging parameters of each real-time Ka-band SAR image after the water area is removed; θ roll The satellite side swing angle is one of the imaging parameters in each real-time Ka-band SAR image after water area removal.

8. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 7, wherein: Perform antenna pattern correction as described in step 6, as follows: Step 6.1: After removing the water area to be corrected, the Ka-band SAR image is divided into multiple image blocks to be corrected along the columns. The mean of each row of pixels in the divided image blocks to be corrected is calculated to obtain the mean pixel value of each row of the divided image blocks to be corrected. A second-order curve is fitted on the multi-row mean pixels of the divided image blocks to be corrected to obtain the antenna direction pixel coordinate sequence of the Ka-band SAR image to be corrected; Step 6.2: Calculate the offset of the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water area removal and the antenna direction pixel coordinate sequence of the Ka-band SAR image to be corrected using the offset model of the antenna pattern changing with imaging time; Step 6.3: Input the Ka-band SAR image to be corrected and perform antenna pattern correction in combination with the offset coefficients of the mean pixels in each row of the divided multiple image blocks to be corrected to obtain the corrected Ka-band SAR image.

9. The method for testing and correcting the antenna pattern of a spaceborne SAR according to claim 8, wherein: The calculation is performed using the offset model of the antenna pattern changing with imaging time. The specific process is as follows: and now n x n +a n-1 x n-1 +…+a0x+a0 Among them, a m is the m-order offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected, where m represents the order of the model polynomial, m∈[1,n], x is the serial number of the divided multiple image blocks to be corrected; y is the position of the row pixel pointed to by the pixel coordinate sequence in the antenna direction; The antenna pattern correction is performed in combination with the offset coefficient. The specific process is as follows: Input the pixel at row x and column y in the Ka-band SAR image to be corrected. The corresponding pixel value is g. before (x,y), x∈[1,X], y∈[1,Y], X represents the number of rows in the Ka-band SAR image to be corrected, and Y represents the number of columns in the Ka-band SAR image to be corrected; The pixel value corresponding to the pixel position (x, y) of the corrected Ka-band SAR image is g after (x,y): g after (x,y)=g before (x,y)÷r(x) Wherein, r(x) is the offset coefficient of the mean pixel in each row of the divided multiple image blocks to be corrected.

10. A spaceborne SAR antenna pattern testing and correction system, characterized in that: include: Image acquisition and marking module, used to acquire multiple SAR images containing land and water areas, and mark the pixel range of multiple marked water areas in each SAR image The network training module is used to build a PSPNet network model. Each SAR image is input into the PSPNet network model to extract the water area, obtain multiple extracted water area pixel ranges of each SAR image, and construct a loss function model based on the multiple labeled water area pixel ranges of each SAR image. The trained PSPNet network model is obtained by training using the gradient descent method. A real-time SAR image extraction module is used to obtain each real-time Ka-band SAR image, input each real-time Ka-band SAR image into the trained PSPNet network model to extract the water area, and obtain multiple extracted water area pixel ranges for each real-time SAR image; A mean filtering module is used to combine the multiple extracted water area pixel ranges of each real-time SAR image and remove the water area in each real-time SAR image through a mean filtering method to obtain a Ka-band SAR image after the water area is removed from each real-time SAR image; The antenna beam angle calculation module is used to calculate the antenna beam angle of the abscissa of the pixel coordinates in the antenna direction of each real-time Ka-band SAR image after water area removal through mean calculation, fourth-order curve fitting, and abscissa calculation; The antenna pattern correction module is used to input the Ka-band SAR image to be corrected, and perform antenna pattern correction based on the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal and the antenna direction pixel coordinate sequence of each real-time Ka-band SAR image after water body area removal to obtain the corrected Ka-band SAR image.

Images