SAR (Synthetic Aperture Radar) historical data recalibration method and system based on different-satellite crossing

Abstract

The invention relates to an SAR (Synthetic Aperture Radar) historical data recalibration method and an SAR historical data recalibration system based on cross of different satellites. The method comprises the following steps: acquiring calibrated satellite-borne SAR historical image data and to-be-calibrated satellite-borne SAR historical image data of different satellites with the same wave band, and performing preprocessing, terrain correction, longitude and latitude cutting and registration to generate a re-calibration historical data pair; through resolution resampling processing and incident angle correction, a recalibration cross data pair is constructed and input into a Gaussian mixture model, and a calibration coefficient of an SAR historical image is obtained through calculation. By collecting satellite-borne SAR historical image data of different satellites in the same wave band, the imaging resolution difference between different loads is solved; after processing such as preprocessing, topographic correction and cutting registration is performed on satellite-borne SAR recalibration data, clustering fitting is performed by adopting a Gaussian mixture model, and weighting fitting is performed on different data pairs to solve a calibration coefficient of to-be-calibrated historical data, so that the recalibration precision of the satellite-borne SAR historical data can be improved.

Description

Technical Field

[0001] The invention relates to the technical field of synthetic aperture radar radiation calibration, and in particular to a SAR historical data recalibration method and system based on alien satellite crossing. Background Art

[0002] Synthetic aperture radar (SAR) absolute radiometric calibration is the process of eliminating system errors and accurately converting the digital quantization value (Digital Number, DN) output by the radar system with the backscattering characteristics of the observed target (i.e., backscatter coefficient or radar cross section). This technology is the basis for quantitative processing and application of SAR data, which can ensure the accuracy, consistency and comparability of SAR data and meet the needs of quantitative remote sensing. Existing SAR radiometric calibration technologies are mainly divided into radiometric calibration technology based on manual calibrators and radiometric cross-calibration technology. Radiometric calibration technology based on manual calibrators calculates the absolute radiometric calibration coefficient based on the relationship between the digital quantization value of the image and the radar cross-section area by deploying a certain number of calibrators with known radar cross-section (RCS) at the calibration site. However, this technology faces limitations such as the deployment, maintenance and number of calibrators, making large-scale calibration difficult to achieve. In contrast, radiometric cross-calibration technology uses ground targets with stable characteristics as common reference objects, and uses the SAR sensor data that has been calibrated with high precision as a reference or standard to cross-calibrate the SAR sensor to be calibrated, eliminating the need to deploy manual calibrators on the ground.

[0003] Historical data recalibration technology refers to the recalibration of remote sensing data that have been collected. In the field of optical remote sensing, historical data recalibration technology has been widely used due to its important role in improving the accuracy of historical remote sensing data, ensuring data consistency, and supporting long-term research. Historical data recalibration technology can ensure the comparability of data between different times and different sensors, and can more accurately describe environmental changes and evaluate their impact. Traditional historical data recalibration technology is often based on the cross-calibration technology of the same satellite's historical data. Since the errors caused by payload performance fluctuations are not considered, when the calibrated data is used to calibrate the data to be calibrated with a long time interval, the ground targets as common reference objects change within this long time span (such as rainfall), which will introduce large calibration errors; and the stable statistical characteristics of different uniform objects are not accurately considered when calculating the recalibration coefficient, resulting in low radiation calibration accuracy, which in turn affects the quantitative inversion accuracy of SAR satellite remote sensing products.

[0004] Therefore, traditional historical data recalibration technology often has the problem of low calibration accuracy. Summary of the invention

[0005] Based on this, in order to solve the above technical problems, a SAR historical data recalibration method and system based on alien satellite crossing are provided, which can improve the accuracy of historical data recalibration.

[0006] A SAR historical data recalibration method based on alien satellite crossover, the method comprising:

[0007] Collect calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract data using the same polarization mode from the calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data as spaceborne SAR recalibration data pairs;

[0008] Preprocessing the spaceborne SAR recalibration data pair, and performing terrain correction on the preprocessed spaceborne SAR recalibration data pair, performing longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, and performing image registration using a SAR image registration method to generate a recalibration historical data pair;

[0009] Performing resolution resampling processing on the recalibrated historical data pairs, and performing incident angle correction on the recalibrated historical data pairs with incident angle differences, to construct recalibrated cross data pairs;

[0010] Inputting the recalibrated cross data pair into a Gaussian mixture model, fitting the recalibrated cross data pair through the Gaussian mixture model, and calculating the calibration coefficient of the SAR historical image;

[0011] The calibration coefficients of the SAR historical images are verified, and the historical data recalibration results are output.

[0012] In one embodiment, collecting calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data of different satellites in the same band includes:

[0013] The calibrated spaceborne SAR historical image data and the spaceborne SAR historical image data to be calibrated are collected from different satellites in the same band with the same coverage area, the same imaging ascending and descending orbits, and imaging time intervals within the threshold range.

[0014] In one embodiment, preprocessing the spaceborne SAR recalibration data pair includes:

[0015] Performing the same preprocessing operation on the data in the spaceborne SAR recalibration data pair;

[0016] The preprocessing operations include thermal noise removal, multi-view processing, speckle filtering, and radiation correction.

[0017] In one embodiment, terrain correction is performed on the pre-processed spaceborne SAR recalibration data, including:

[0018] The pre-processed spaceborne SAR recalibration data are subjected to terrain correction processing using a geocoding method under the same coordinate system, and the pre-processed spaceborne SAR recalibration data are centered, so that image data of different payloads have the same geographic information at corresponding pixel points.

[0019] In one embodiment, performing resolution resampling processing on the recalibrated historical data pair includes:

[0020] Using a lower resolution of the recalibrated historical data pair as a resampling benchmark;

[0021] Based on the resampling benchmark, a weighted calculation is performed on each pixel point of the image in the recalibrated historical data pair through a downsampling function to obtain a new pixel value.

[0022] In one embodiment, performing incident angle correction on the recalibrated historical data pair with incident angle differences includes:

[0023] Calculating an incident angle corresponding to the recalibration historical data pair, if the difference between the incident angle and the reference incident angle is within a threshold range, then there is no incident angle difference;

[0024] If the difference between the incident angle and the reference incident angle exceeds the threshold range, it is determined that there is an incident angle difference, and an incident angle correction model is used for correction.

[0025] In one embodiment, constructing a recalibrated cross data pair includes:

[0026] Extracting a first pixel value corresponding to the historical data to be calibrated in the recalibrated historical data pair, and using the first pixel value as an independent variable of the recalibrated cross data pair;

[0027] Extracting a second pixel value corresponding to the calibrated standard historical data after downsampling in the recalibrated historical data pair, and using the second pixel value as a dependent variable of the recalibrated cross data pair;

[0028] The independent variables and the dependent variables are combined to form a two-dimensional data pair, and the two-dimensional data pair is used as data corresponding to the recalibrated cross data pair to construct the recalibrated cross data pair.

[0029] In one embodiment, the recalibrated cross data pair is input into a Gaussian mixture model, and the recalibrated cross data pair is fitted by the Gaussian mixture model to calculate the calibration coefficient of the SAR historical image, including:

[0030] Inputting the recalibrated cross data pairs into a Gaussian mixture model, extracting the high-frequency region in each of the two-dimensional data pairs, and calculating the interquartile range of the two-dimensional data pairs;

[0031] Based on the high-frequency area, the bin width is calculated using the Freedman-Diaconis rule, and the joint distribution of the two-dimensional histogram statistical variables is used to obtain the high-frequency area data;

[0032] If the recalibrated cross data pairs are data pairs of different types of regions, the high-frequency region data are used to fit Gaussian mixture models of each two-dimensional Gaussian distribution, and the initial parameters of the Gaussian mixture model are initialized using the K-means method;

[0033] A Gaussian mixture model is fitted on the recalibrated cross data pair to calculate the calibration coefficient of the SAR historical image.

[0034] In one embodiment, performing Gaussian mixture model fitting on the recalibrated cross data pair to calculate the calibration coefficient of the SAR historical image includes:

[0035] Fitting the recalibrated cross data pair with a Gaussian mixture model to extract the corresponding expectation and Gaussian weight of the Gaussian distribution in the high-frequency region;

[0036] The Gaussian weight and the expected ratio corresponding to the Gaussian weight are multiplied to obtain the calibration coefficient of the SAR historical image.

[0037] A SAR historical data recalibration system based on alien satellite crossover, the system comprising:

[0038] The data acquisition module is used to collect the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract the data using the same polarization mode from the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data as the spaceborne SAR recalibration data pair;

[0039] A data preprocessing module is used to preprocess the spaceborne SAR recalibration data pair, perform terrain correction on the preprocessed spaceborne SAR recalibration data pair, perform longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, and perform image registration using a SAR image registration method to generate a recalibration historical data pair;

[0040] A data pair construction module, used for performing resolution resampling processing on the recalibration historical data pairs, and performing incident angle correction on the recalibration historical data pairs with incident angle differences, to construct recalibration cross data pairs;

[0041] A calibration coefficient calculation module, used for inputting the recalibration cross data pair into a Gaussian mixture model, fitting the recalibration cross data pair through the Gaussian mixture model, and calculating the calibration coefficient of the SAR historical image;

[0042] The result verification module is used to verify the calibration coefficients of the SAR historical images and output the historical data recalibration results.

[0043] The above-mentioned SAR historical data recalibration method and system based on heterosatellite crossover solves the imaging resolution difference between different payloads by collecting spaceborne SAR historical image data from different satellites in the same band; constructs recalibration cross data pairs after preprocessing, terrain correction, cropping and registration of spaceborne SAR recalibration data pairs, thereby expanding the data source and application scope of historical data radiation recalibration; adopts Gaussian mixture model for cluster fitting, performs weighted fitting on different data pairs to solve the calibration coefficient of the historical data to be calibrated, which can improve the accuracy of spaceborne SAR historical data recalibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A diagram of the application environment of a SAR historical data recalibration method based on alien satellite crossover in one embodiment;

[0045] Figure 2 It is a flowchart of a SAR historical data recalibration method based on alien satellite crossover in one embodiment;

[0046] Figure 3 A schematic diagram of a region of recalibrated historical data pairs selected after clipping in one embodiment;

[0047] Figure 4 A specific flow chart of solving historical data calibration coefficients based on a Gaussian mixture model in one embodiment;

[0048] Figure 5 A result diagram of a high-frequency region of recalibrated data based on Gaussian mixture model fitting in one embodiment;

[0049] Figure 6 It is a flowchart of a SAR historical data recalibration method based on alien satellite crossing in another embodiment;

[0050] Figure 7 It is a structural block diagram of a SAR historical data recalibration system based on alien satellite crossover in one embodiment;

[0051] Figure 8 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0053] The SAR historical data recalibration method based on heterogeneous satellite crossover provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Figure 1 As shown, the application environment includes a computer device 110. The computer device 110 can collect calibrated satellite-borne SAR historical image data and to-be-calibrated satellite-borne SAR historical image data of different satellites in the same band, and extract data using the same polarization mode from the calibrated satellite-borne SAR historical image data and to-be-calibrated satellite-borne SAR historical image data as satellite-borne SAR recalibration data pairs; the computer device 110 can pre-process the satellite-borne SAR recalibration data pairs, and perform terrain correction on the pre-processed satellite-borne SAR recalibration data pairs, and perform latitude and longitude clipping on the calibrated satellite-borne SAR historical image data and to-be-calibrated satellite-borne SAR historical image data after terrain correction. , use the SAR image registration method to perform image registration, and generate a recalibrated historical data pair; the computer device 110 can perform resolution resampling processing on the recalibrated historical data pair, and perform incident angle correction on the recalibrated historical data pair with incident angle difference, and construct a recalibrated cross data pair; the computer device 110 can input the recalibrated cross data pair into the Gaussian mixture model, fit the recalibrated cross data pair through the Gaussian mixture model, and calculate the calibration coefficient of the SAR historical image; the computer device 110 can verify the calibration coefficient of the SAR historical image and output the historical data recalibration result. Among them, the computer device 110 can be, but not limited to, various personal computers, laptops, smart phones, robots, unmanned aerial vehicles and other devices.

[0054] In one embodiment, Figure 2 As shown, a SAR historical data recalibration method based on alien satellite crossing is provided, comprising the following steps:

[0055] Step 202, collect calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract data using the same polarization mode from the calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data as spaceborne SAR recalibration data pairs.

[0056] The computer device can extract recalibration data pairs with similar SAR payload times of different satellites in the same band, and use them to obtain uniform and stable reference target data used for SAR radiation cross recalibration. Specifically, in this embodiment, the computer device can obtain standard calibrated and uncalibrated satellite-borne SAR historical image data, so as to screen out imaging areas and cross-image data pairs that meet the requirements of two satellites with different payloads in the same band.

[0057] In one embodiment, a SAR historical data recalibration method based on heterosatellite crossing is provided, which may also include a process of collecting satellite-borne SAR historical image data, and the specific process includes: collecting calibrated satellite-borne SAR historical image data and satellite-borne SAR historical image data to be calibrated under different satellites of the same band with the same coverage area, the same imaging ascending and descending orbits, and imaging time intervals within a threshold range.

[0058] The computer device can obtain a set of SAR image history data, including calibrated spaceborne SAR historical image data and uncalibrated spaceborne SAR historical image data. The calibrated spaceborne SAR historical image data and uncalibrated spaceborne SAR historical image data come from different SAR satellites in the same band, such as Sentinel-1 and Gaofen-3 in the C band. The calibrated spaceborne SAR historical image data needs to be calibrated with high precision radiation to reduce the subsequent error introduced by the standard data observation value error.

[0059] In this embodiment, the collected calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data cover the same area and contain evenly and stably distributed targets, such as urban areas, bare land, Gobi or rain forests, etc., have the same imaging ascending and descending orbits, and the imaging time interval is less than 60 days. If the time intervals are very close, the difference does not exceed one week, and the selected target objects are relatively stable, such as saline-alkali land, forests, etc., they can also be used as spaceborne SAR recalibration data pairs.

[0060] After the computer equipment collects the calibrated spaceborne SAR historical image data and the uncalibrated spaceborne SAR historical image data, it can extract the same polarization mode data for SAR radiation cross recalibration. The calibrated spaceborne SAR historical image data and the uncalibrated spaceborne SAR historical image data must be in the same polarization mode to reduce the errors caused by factors such as different polarization channels and polarization modes in the subsequent radiation cross calibration, and ensure that the recalibration results are closer to the true value of the ground object.

[0061] For example, in this embodiment, the main data used by the computer equipment is the historical image data obtained by the C-band spaceborne SAR satellite Sentinel-1A and GF-3A. Both satellites are on-orbit satellites with high C-band payload performance and imaging quality. They are similar in sensor parameters and comparable, and have certain intercommunication capabilities in data. The two satellites irradiate most of the earth's objects and areas while orbiting the earth, and their detection areas have many overlapping parts. At the same time, the revisit period is less than 1 month, and the data collection time is also feasible. It is possible to screen multiple groups of historical cross data pairs with small time difference, the same irradiation area, and the same polarization mode in the same ascending and descending orbit. In terms of historical data, the two satellites have good data conditions. Sentinel-1A will monitor radiation performance every year and publish quality reports regularly. Its absolute radiation accuracy is 1dB. As my country's first C-band multi-polarization, multi-mode SAR satellite, GF-3A also has high radiation performance. The China Resources Satellite Center conducts field calibration experiments every year, and its absolute radiation accuracy is 1.5dB. The specific parameters of the two satellite sensors are shown in the following table:

[0062]

[0063] Among them, in terms of data selection, it is necessary to select data with the same polarization mode. Since the dual polarization modes of Sentinel-1A and GF-3A are completely different, the data selection is Sentinel-1A's dual polarization GRD data (GRD-VV+VH) and GF-3A's L1A-level full polarization (QPSI mode) data (SLC-HH+VV+VH+HV). Computer equipment can extract the data collected corresponding to the common co-polarization (VV) mode of the two for subsequent radiation recalibration processing. Among them, the Sentinel's GRD data does not contain phase information, so the GF-3A's L1A-level SLC data also needs to be pre-processed by multi-viewing, filtering, etc. to obtain the intensity value. The corresponding area schematic diagram is as follows Figure 3 The specific parameters of the selected experimental area are shown in the following table:

[0064]

[0065] Step 204, preprocessing the spaceborne SAR recalibration data pair, and performing terrain correction on the preprocessed spaceborne SAR recalibration data pair, after performing longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, perform image registration using a SAR image registration method to generate a recalibration historical data pair.

[0066] The computer device can pre-process the spaceborne SAR recalibration data pairs to achieve the registration and alignment of different satellite SAR data. Specifically, the target area selected in step 202 is used to obtain the historical data radiation recalibration data area production process, and the necessary pre-processing is performed for the subsequent production of the recalibration cross data pairs, so as to achieve the registration and cropping of the historical image data of two spaceborne SAR satellites with the same band and different payloads.

[0067] Specifically, in one embodiment, a SAR historical data recalibration method based on heterosatellite crossing is provided, which may also include a data preprocessing process, and the specific process includes: performing the same preprocessing operation on the data in the satellite-borne SAR recalibration data pair; wherein the preprocessing operation includes thermal noise removal, multi-look processing, speckle filtering, and radiation correction.

[0068] In this embodiment, the computer device pre-processes the calibrated space-borne SAR historical image data and the to-be-calibrated space-borne SAR historical image data to eliminate the differences between different satellites.

[0069] The computer equipment can perform the same preprocessing on the selected historical image data pairs that are uniform and stable, have the same ascending and descending orbits, similar imaging times, and the same polarization mode, and then obtain the backscatter coefficient value σ of the calibrated spaceborne SAR historical image data and the image digital quantization value DN of the spaceborne SAR historical image data to be calibrated. The preprocessing operations mainly include: thermal noise removal, multi-view processing, speckle filtering, radiation correction, etc., among which:

[0070] Thermal noise removal is to remove the noise caused by the sensor system and improve data accuracy;

[0071] Multi-view processing is to average the single-view complex (SLC) image in azimuth and range, and the result is the intensity data after multi-view;

[0072] Coherent speckle filtering is to eliminate the influence of coherent speckle in image data on the subsequent processing of SAR images;

[0073] Radiation correction is to convert the digital quantization value of the calibrated spaceborne SAR historical image data into the corresponding backscatter coefficient value to facilitate subsequent calibration processing.

[0074] Among them, when the computer equipment performs multi-view processing and speckle filtering, it mainly pre-processes the two sets of spaceborne SAR satellite historical data obtained. In order to ensure the uniformity of pre-processing, the pre-processing process uses a unified process and parameters to perform filtering pre-processing on the GRD data of Sentinel-1A, and multi-view processing and filtering pre-processing on the SLC data of GF-3A.

[0075] In one embodiment, a SAR historical data recalibration method based on heterosatellite crossing is provided, which may also include a terrain correction process, and the specific process includes: performing terrain correction processing on the pre-processed satellite-borne SAR recalibration data using a geographic coding method under the same coordinate system, centering the pre-processed satellite-borne SAR recalibration data, and the image data of different payloads have the same geographic information at corresponding pixel points.

[0076] The computer equipment can use the geocoding method in the same coordinate system to perform terrain correction processing on the calibrated historical data and the historical data to be calibrated in the spaceborne SAR recalibration data pair, so that the image data of different payloads have the same geographic information at the corresponding pixel points, eliminating the influence of distance deformation caused by the side-view imaging of the spaceborne SAR satellite. Specifically, the computer equipment can use the digital elevation model (DEM) data downloaded by USGS to perform unified terrain correction processing on the image data obtained by preprocessing the historical data of the two satellites, to ensure that the geographic data used is the same.

[0077] In one embodiment, a method for recalibrating SAR historical data based on alien satellite crossing is provided, which may also include the process of longitude and latitude clipping and calibration data pair registration. Specifically, the computer equipment performs longitude and latitude clipping on the calibrated and uncalibrated historical data after terrain correction to ensure the mapping accuracy of the data pair. The cropping area should select areas in the image that are uniform, stable, and have a large backscatter coefficient, mainly including: bare land, Gobi, rainforest, saline-alkali land or forest, etc. Subsequently, the SAR image registration method is used to perform image registration to achieve matching of the calibrated historical data and the uncalibrated historical data, and generate a recalibrated cross data pair containing multiple land object type data slices.

[0078] Specifically, the computer equipment can perform longitude and latitude clipping on the image data obtained after preprocessing and geocoding, perform the same longitude and latitude clipping on the two sets of calibration data, and then use the SAR scale-invariant feature transformation method to obtain the image pair matching feature points, and perform recalibration data pair registration. Two pairs of data slices of different ground object types are extracted from the overlapping areas of the two sets of images. The detailed parameters of each set of calibration data pairs are shown in the following table:

[0079]

[0080] In this embodiment, the selected recalibration historical data pair area diagram after clipping is as follows: Figure 3 As shown. The recalibrated data pairs 1 and 2 correspond to Figure 3 The two regions in (b) correspond to the recalibrated data pairs 3 and 4 respectively. Figure 3 In the two regions in (d), the cropped area is not obvious because the Sentinel-1A dual-polarization imaging area is large. Figure 3The corresponding data blocks are not marked in the schematic diagram.

[0081] Step 206 , performing resolution resampling processing on the recalibrated historical data pairs, and performing incident angle correction on the recalibrated historical data pairs with incident angle differences, to construct recalibrated cross data pairs.

[0082] The computer equipment can perform resolution resampling of the recalibrated historical data pairs to eliminate the impact of resolution differences. That is, the computer equipment uses the output preprocessed and registered historical image data to perform resolution resampling according to the imaging resolution of the two satellites, performs incident angle correction on the data pairs with incident angle differences, and reconstructs one or more groups of recalibrated cross data pairs that meet the requirements as input for solving the recalibration coefficients based on the Gaussian mixture model method; finally, the calibration coefficients for the radiation recalibration of the historical satellite-borne SAR data to be calibrated are fitted using a two-dimensional Gaussian mixture model, and compared with the standard values ​​to verify the accuracy of the recalibration results.

[0083] Specifically, in one embodiment, a SAR historical data recalibration method based on alien satellite crossing is provided, which may also include a resolution resampling processing process, and the specific process includes: using a lower resolution in the recalibrated historical data pair as a resampling benchmark; based on the resampling benchmark, performing weighted calculation on each pixel point of the image in the recalibrated historical data pair through a downsampling function to obtain a new pixel value.

[0084] The computer equipment can resample the resolution of each pair of calibrated historical image data and historical image data to be calibrated according to the results obtained after longitude and latitude clipping and then registration, to ensure that the resolution difference of historical data from different satellites will not have a direct impact on the calibration results during the historical data radiation cross-calibration process, thereby improving the accuracy of the calibration results. Among them, the computer equipment resamples each group of recalibrated data pairs obtained to keep the resolution of SAR historical data from different payloads consistent, eliminating the resolution impact for the next step of building data pairs. For example, each group of GF-3A corresponding data can be downsampled to 15m resolution to obtain image data with the same resolution as Sentinel-1A.

[0085] In the resampling process, the downsampling process is adopted, and the lower resolution of the two is used as the resampling benchmark to avoid introducing new values ​​and retain the original scattering characteristics of the ground object. For example, the resolution of the spaceborne SAR satellite 1 image is P1, and the corresponding pixel value is I(x, y). The resolution of the spaceborne SAR satellite 2 image is P2, and the corresponding pixel value is I1(x, y). Assuming that the P2 value is less than the P1 value (that is, P2 corresponds to high-resolution image data), the high-resolution image data P1 is downsampled to P2 using the downsampling interpolation method, and the downsampling is obtained by I′(x′, y′). The downsampling kernel function formula is expressed as:

[0086] For each pixel in the image, the computer device can weight the surrounding pixels through the downsampling kernel function to calculate the new pixel value. The specific steps include: positioning calculation, determining the position of the pixel in the target image in the source image. For the scaling operation, this involves the calculation of the scaling factor; weight calculation, calculating the weight of the target pixel and the neighboring pixel in the source image according to the downsampling kernel function; weighted summation: weighted summation of the values ​​of the neighboring pixels in the source image according to the weights to obtain the new value of the target pixel. The calculation formula is expressed as: Wherein, I(x, y) is the pixel value of the source image, I′(x′, y′) is the pixel value of the target image, L(x) is the downsampling kernel function, and a is the window size. In this embodiment, the value of a can be 3.

[0087] In this embodiment, the computer device makes the historical image data from different satellites in the same band have the same resolution through the resampling process, so as to eliminate the calibration error introduced by the inconsistent imaging resolution when constructing the recalibration cross data pair.

[0088] In one embodiment, a SAR historical data recalibration method based on alien satellite crossing is provided, which may also include a process of performing an incident angle correction, and the specific process includes: calculating the incident angle corresponding to the recalibrated historical data pair, if the difference between the incident angle and the reference incident angle is within a threshold range, then there is no incident angle difference; if the difference between the incident angle and the reference incident angle exceeds the threshold range, it is determined that there is an incident angle difference, and correction is performed using an incident angle correction model.

[0089] Among them, the computer equipment can determine whether there is an incident angle difference between the historical data and the recalibrated data. If there is an incident angle difference, it is handled in two cases: if the incident angle difference in the selected area is small, for example, less than 1 degree, the incident angle correction step is directly ignored; if it is large, an incident angle correction model is required for correction, wherein the incident angle correction model can be an Oh model, a Lambertian model, a cosine model, etc.

[0090] In this embodiment, the selection of the incident angle model should comply with the data characteristic requirements, accuracy requirements, accuracy and computational complexity requirements, wherein the data characteristic requirements: different surface types (such as water bodies, vegetation, cities, bare soil) have different response characteristics to the incident angle, and the selected correction model should be able to accurately describe the reflection characteristics of the target surface; accuracy requirements: select an appropriate model according to the application requirements. If high-precision surface information (such as topographic measurement, fine classification) is required, a complex physical model (such as the Oh model) can be used. If only a rough correction is required for observation, a simple empirical model (such as the cosine model) can be used; accuracy and computational complexity requirements: for real-time processing requirements (such as real-time monitoring), the cosine model with simple calculation and fast speed is more suitable.

[0091] In this embodiment, the computer equipment can perform incident angle correction according to the incident angle difference of each group of recalibrated cross pairs. During the experiment, the incident angle deviation of four image pairs that can be selected does not differ by more than 0.2°, and the impact of the incident angle on the historical data recalibration process can be ignored. If a large incident angle difference value occurs, it is necessary to perform incident angle correction on the selection rule of the incident angle model according to the specific situation.

[0092] In one embodiment, a SAR historical data recalibration method based on alien satellite crossing is provided, which may also include a process of constructing a recalibration cross data pair, and the specific process includes: extracting a first pixel value corresponding to the historical data to be calibrated in the recalibration historical data pair, and using the first pixel value as an independent variable of the recalibration cross data pair; extracting a second pixel value corresponding to the calibrated standard historical data after downsampling in the recalibration historical data pair, and using the second pixel value as a dependent variable of the recalibration cross data pair; combining the independent variable and the dependent variable to form a two-dimensional data pair, and using the two-dimensional data pair as the data corresponding to the recalibration cross data pair to construct the recalibration cross data pair.

[0093] Among them, by resampling and correcting the incident angle of the recalibrated cross data pair after registration, we get a recalibrated historical data pair that is almost the same except for the influence of the load. Each data pair contains a backscatter coefficient data σ of a calibrated image. cal and the image digital quantization value data DN of an image to be calibrated uncal During the data pair construction process, the computer device can construct the obtained pair of image data into (DN uncal , σ cal ) in order to solve the calibration coefficient k of the historical image to be calibrated using the subsequent Gaussian mixture model.

[0094] Specifically, in this embodiment, the recalibration cross data pair construction process may include:

[0095] A pair of n×n uncalibrated historical data and calibrated historical data with the same resolution and pixel size is regarded as a cross-calibration data pair (referred to as cross-data pair);

[0096] Extract the historical data to be calibrated I1(x i ,y i ) corresponds to the first pixel value (DN uncal ) as the independent variable x of the recalibrated cross data pair;

[0097] Extract the calibrated standard historical data I′(x i ′,y i ′) corresponds to the second pixel value (σ cal ) as the dependent variable y of the recalibrated cross data pair;

[0098] The extracted first pixel value and the second pixel value are combined to construct a (x, y) two-dimensional array as the corresponding data at the (i, i)th position of the group of recalibrated cross data pairs, which serves as the input for the Gaussian mixture model to calculate the calibration coefficients of the historical data image to be calibrated in the next step.

[0099] In this embodiment, the computer device can extract the backscatter coefficient (σ) of the calibrated Sentinel-1A historical data in each group of data pairs after registration and resampling. cal ) image as the dependent variable, and extract the digital quantization value (DN) of the historical data of GF-3A to be calibrated. uncal ) as independent variables, construct a two-dimensional cross-data (DN) corresponding to the recalibration data pair uncal , σ cal ), forming the standard data input for solving the radiation recalibration coefficients using the subsequent Gaussian mixture model.

[0100] Step 208: input the recalibrated cross data pair into the Gaussian mixture model, fit the recalibrated cross data pair through the Gaussian mixture model, and calculate the calibration coefficient of the SAR historical image.

[0101] Computer equipment can use the Gaussian mixture model clustering algorithm instead of the traditional least squares method to fit the cross-data pairs, extract the scattering characteristics of the high-frequency area, eliminate the influence of other types of land objects and noise, and finally calculate the calibration coefficients of historical SAR images.

[0102] In one embodiment, a SAR historical data recalibration method based on alien satellite crossing is provided, which may also include a process of calculating the calibration coefficient of the SAR historical image, and the specific process includes: inputting the recalibrated cross data pair into a Gaussian mixture model, extracting the high-frequency area in each two-dimensional data pair, and calculating the interquartile range of the two-dimensional data pair; based on the high-frequency area, using the Freedman-Diaconis rule to calculate the bin width, and using the joint distribution of the two-dimensional histogram statistical variables to obtain the high-frequency area data; if the recalibrated cross data pair is a data pair of different types of areas, using the high-frequency area data to fit the Gaussian mixture model of each two-dimensional Gaussian distribution, and using the K-means method to initialize the initial parameters of the Gaussian mixture model; fitting the Gaussian mixture model to the recalibrated cross data pair, and calculating the calibration coefficient of the SAR historical image.

[0103] In this embodiment, if Figure 4 As shown, the process of calculating the calibration coefficient of the SAR historical image may specifically include:

[0104] The obtained two-dimensional recalibrated cross data pairs are used as input to extract the high-frequency area in each two-dimensional cross data pair; extracting the high-frequency area means extracting the corresponding (DN uncal , σ cal ) value, retain the scattering characteristics of the main uniform and stable objects as much as possible, and exclude the influence of other types of objects and noise;

[0105] Calculate the interquartile range (IQR) of a two-dimensional data pair. The calculation formula is: σ cal,IQR =IQR(σ cal );DN uncal,IQR =IQR(DN uncal );IQR=Q3-Q1;wherein, Q3 is the third quartile and Q1 is the first quartile;

[0106] Use the Freedman-Diaconis rule to calculate the bin width: The calculation formula can be expressed as: σ cal,binwidth =2·σ cal,IQR ·n -1 / 3 DN uncal,binwidth =2·DN uncal,IQR ·n -1 / 3 ; where n is the number of data points; the number of bins is calculated based on the bin width:

[0107] Use a two-dimensional histogram to count the joint distribution of two variables, find the highest frequency area and its index in the group, if the amount of data in the highest frequency area obtained by statistics is small, its boundary can be calculated according to the index of the high frequency area, and through the expansion factor, the high frequency area can be expanded to the corresponding area located in the middle 50% of the data volume;

[0108] When multiple data pairs of different types of regions are input, these high-frequency region data need to be fitted with a Gaussian mixture model (GMM) of multiple two-dimensional Gaussian distributions: Among them, π i is the weight of the ith Gaussian distribution, μ i is the mean, Σ i is the covariance matrix;

[0109] In one embodiment, the process of calculating the calibration coefficient of the SAR historical image may also include: fitting the recalibrated cross data pair with a Gaussian mixture model to extract the corresponding expectation and Gaussian weight of the Gaussian distribution in the high-frequency area; multiplying the Gaussian weight and the expectation ratio corresponding to the Gaussian weight to obtain the calibration coefficient of the SAR historical image.

[0110] Specifically, the computer device can use the K-means method to initialize the initial parameters of the Gaussian mixture model, and then fit the Gaussian mixture model to the cross data pair to extract multiple corresponding expectations and weights of the high-frequency region Gaussian distribution used; the K-means initialization parameter process can be expressed as: mean μ K : The initialization mean of GMM is the Kmeans cluster center; μ k = K-means cluster center (k = 1, 2, ..., K); covariance matrix Σ k : For each cluster, calculate the covariance matrix of the data points and use it as the initial covariance matrix for the cluster: Weight π k : The initial weight of GMM is the ratio of the number of data points in each cluster to the total number of data points Among them, N k is the number of data points in the kth cluster, and N is the total number of data points.

[0111] Next, the computer device can fit the two-dimensional Gaussian mixture model to each two-dimensional data pair input, and extract the mean and corresponding weight corresponding to each Gaussian distribution. Use the corresponding Gaussian weight and its corresponding two-dimensional Gaussian distribution backscatter coefficient expectation μ i (σ cal ) and the digital quantization value expected DN uncal The historical SAR image calibration coefficient calculated using the two-dimensional Gaussian mixture model is multiplied by the ratio of

[0112]

[0113] Specifically, in this embodiment, the computer device can use the obtained multiple pairs of two-dimensional arrays as data input to recalibrate the historical data of the spaceborne SAR to be calibrated; the two pairs of recalibrated image pairs corresponding to the two groups of image data are used as the input of the two-dimensional Gaussian mixture model, and the Gaussian mixture model is fitted to all the input data using the model. The fitting result is shown in the figure Figure 5 (c); then extract the high-frequency data area according to the four-bin criterion, and then fit the extracted high-frequency data area (accounting for about 50% of the total data volume) with a two-dimensional Gaussian mixture model, as shown in Figure 5 (d) and solve the weight value and mean of the corresponding data pair, combining the weight and backscattering coefficient mean μ obtained by combining the two pairs of fitting data used in each group i (σ cal ) and the digital quantization value mean μ i (DN uncal ) are multiplied by the ratio of the two values ​​to obtain the radiation recalibration coefficient of the alien SAR historical data based on the Gaussian mixture model. The detailed solution results of each historical data pair based on the Gaussian mixture model are shown in the following table:

[0114]

[0115]

[0116] Step 210, verify the calibration coefficients of the SAR historical images and output the historical data recalibration results.

[0117] The computer equipment can use the historical SAR image calibration coefficient k GMM高频区域 and the digital quantization value μ of the historical satellite-borne SAR image data to be calibrated i (DN uncal ) is multiplied to obtain the result value of the backscatter coefficient based on the recalibration; then the reference backscatter coefficient value is calculated by the existing or official nominal calibration coefficient of the satellite to be recalibrated; finally, the root mean square error (RMSE) is solved pixel by pixel. If the RMSE obtained meets the design accuracy of the satellite's on-orbit radiation calibration, the historical data recalibration result is output: Among them, σ i is the backscatter coefficient value calculated based on the existing or official nominal calibration coefficients of the satellite to be calibrated, The backscatter coefficient value is recalibrated based on the historical data calculated by the present invention.

[0118] If it does not meet the requirements, the computer equipment can compare the average backscatter coefficient of the same type of ground object of the calibrated satellite in the similar time; if the error between the calculated average backscatter coefficient and the average of the same type of ground objects is less than the design accuracy of the satellite's in-orbit radiation calibration, the historical data recalibration result is output; otherwise, the recalibration result is rejected, and the data pair selection, preprocessing and recalibration process are repeated.

[0119] Specifically, in this embodiment, the computer device can use the obtained calibration coefficients to recalibrate the historical image data to be calibrated, and obtain recalibrated backscatter coefficient image data based on the historical image data; use the recalibrated data and; compare the standard backscatter coefficient image data directly obtained by the provided calibration coefficients, and calculate the root mean square error between the theoretical backscatter coefficient and the recalibrated backscatter coefficient result; then use the classical least squares method to solve and compare with the results in this embodiment.

[0120] The results show that in this embodiment, by establishing a recalibration process for the historical data of spaceborne SAR with different payloads in the same band, the differences in resolution, incident angle, etc. between satellite data with different payloads are solved, and a clustering method based on a mixed Gaussian model is used in the final solution of the calibration coefficient, and a recalibration coefficient is obtained by weighted fitting of multiple recalibration data pairs; among them, the recalibration accuracy meets the requirements of the radiation calibration accuracy of GF-3A (<1.5dB), and the effect is significantly better than the current mainstream least squares solution method, as shown in the following table:

[0121]

[0122] In one embodiment, Figure 6 As shown, a SAR historical data recalibration method based on heterogeneous satellite crossover is provided, which mainly includes three parts: historical data acquisition, historical data recalibration data pair preprocessing, and historical data crossover recalibration. Among them,

[0123] Historical data acquisition: used to select imaging areas and cross-image data pairs that meet the requirements of two satellites with the same band but different payloads;

[0124] Preprocessing of historical data recalibration data pairs: It provides a process for obtaining historical data radiation recalibration data area production for the target area selected in S1, performs necessary preprocessing for the subsequent production of recalibration cross data pairs, and realizes the registration and cropping of historical image data of two spaceborne SAR satellites with the same band but different payloads;

[0125] Historical data cross-recalibration part: The output pre-processed and aligned historical image data are used to perform resolution resampling according to the imaging resolution of the two satellites, and the incident angle is corrected for the data pairs with different incident angles. One or more sets of recalibrated cross-data pairs that meet the requirements are reconstructed as input for solving the recalibration coefficients based on the Gaussian mixture model method. Finally, the two-dimensional Gaussian mixture model is used to fit the calibration coefficients for the radiation recalibration of the historical satellite-borne SAR data to be calibrated, and the coefficients are compared with the standard values ​​to verify the accuracy of the recalibration results.

[0126] It should be understood that, although the steps in the above-mentioned flowcharts are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the above-mentioned flowcharts may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0127] In one embodiment, Figure 7 As shown, a SAR historical data recalibration system based on alien satellite crossover is provided, comprising: a data acquisition module 710, a data preprocessing module 720, a data pair construction module 730, a calibration coefficient calculation module 740 and a result verification module 750, wherein:

[0128] The data acquisition module 710 is used to collect the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract the data using the same polarization mode from the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data as the spaceborne SAR recalibration data pair;

[0129] The data preprocessing module 720 is used to preprocess the spaceborne SAR recalibration data pair, perform terrain correction on the preprocessed spaceborne SAR recalibration data pair, perform longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, and perform image registration using a SAR image registration method to generate a recalibration historical data pair;

[0130] A data pair construction module 730 is used to perform resolution resampling processing on the recalibration historical data pairs, and to perform incident angle correction on the recalibration historical data pairs with incident angle differences, so as to construct recalibration cross data pairs;

[0131] The calibration coefficient calculation module 740 is used to input the recalibration cross data pair into the Gaussian mixture model, fit the recalibration cross data pair through the Gaussian mixture model, and calculate the calibration coefficient of the SAR historical image;

[0132] The result verification module 750 is used to verify the calibration coefficients of the SAR historical images and output the historical data recalibration results.

[0133] In one embodiment, the data acquisition module 710 is also used to collect calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data with the same coverage area, the same imaging ascending and descending orbits, and imaging time intervals within a threshold range under different satellites in the same band.

[0134] In one embodiment, the data preprocessing module 720 is also used to perform the same preprocessing operation on the data in the spaceborne SAR recalibration data pair; wherein the preprocessing operation includes thermal noise removal, multi-look processing, speckle filtering, and radiation correction.

[0135] In one embodiment, the data preprocessing module 720 is also used to perform terrain correction processing on the preprocessed spaceborne SAR recalibration data using the geocoding method under the same coordinate system, and to center the preprocessed spaceborne SAR recalibration data so that the image data of different payloads have the same geographic information at corresponding pixel points.

[0136] In one embodiment, the data pair construction module 730 is also used to use the lower resolution in the recalibrated historical data pair as a resampling benchmark; based on the resampling benchmark, a weighted calculation is performed on each pixel point of the image in the recalibrated historical data pair through a downsampling function to obtain a new pixel value.

[0137] In one embodiment, the data pair construction module 730 is also used to calculate the incident angle corresponding to the recalibrated historical data pair. If the difference between the incident angle and the reference incident angle is within a threshold range, there is no incident angle difference; if the difference between the incident angle and the reference incident angle exceeds the threshold range, it is determined that there is an incident angle difference, and the incident angle correction model is used for correction.

[0138] In one embodiment, the data pair construction module 730 is also used to extract the first pixel value corresponding to the historical data to be calibrated in the recalibrated historical data pair, and use the first pixel value as the independent variable of the recalibrated cross data pair; extract the second pixel value corresponding to the calibrated standard historical data after downsampling in the recalibrated historical data pair, and use the second pixel value as the dependent variable of the recalibrated cross data pair; combine the independent variable and the dependent variable to form a two-dimensional data pair, and use the two-dimensional data pair as the data corresponding to the recalibrated cross data pair to construct the recalibrated cross data pair.

[0139] In one embodiment, the calibration coefficient calculation module 740 is also used to input the recalibrated cross data pairs into the Gaussian mixture model, extract the high-frequency area in each two-dimensional data pair, and calculate the interquartile range of the two-dimensional data pair; based on the high-frequency area, the Freedman-Diaconis rule is used to calculate the bin width, and the joint distribution of the two-dimensional histogram statistical variables is used to obtain the high-frequency area data; if the recalibrated cross data pairs are data pairs of different types of areas, the high-frequency area data is used to fit the Gaussian mixture model of each two-dimensional Gaussian distribution, and the K-means method is used to initialize the initial parameters of the Gaussian mixture model; the Gaussian mixture model is fitted to the recalibrated cross data pairs to calculate the calibration coefficient of the SAR historical image.

[0140] In one embodiment, the calibration coefficient calculation module 740 is also used to fit the Gaussian mixture model to the recalibrated cross data pairs, extract the corresponding expectation and Gaussian weight of the Gaussian distribution in the high-frequency area; multiply the Gaussian weight and the expected ratio corresponding to the Gaussian weight to obtain the calibration coefficient of the SAR historical image.

[0141] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a SAR historical data recalibration method based on alien satellite crossing is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse, etc.

[0142] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0143] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of a SAR historical data recalibration method based on alien satellite crossing when executing the computer program.

[0144] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of a SAR historical data recalibration method based on alien satellite crossing are implemented.

[0145] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0146] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0147] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A SAR historical data recalibration method based on heterogeneous satellite crossover, characterized in that: The method comprises: Collect calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract data using the same polarization mode from the calibrated spaceborne SAR historical image data and to-be-calibrated spaceborne SAR historical image data as spaceborne SAR recalibration data pairs; Preprocessing the spaceborne SAR recalibration data pair, and performing terrain correction on the preprocessed spaceborne SAR recalibration data pair, performing longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, and performing image registration using a SAR image registration method to generate a recalibration historical data pair; Performing resolution resampling processing on the recalibrated historical data pairs, and performing incident angle correction on the recalibrated historical data pairs with incident angle differences, to construct recalibrated cross data pairs; Inputting the recalibrated cross data pair into a Gaussian mixture model, fitting the recalibrated cross data pair through the Gaussian mixture model, and calculating the calibration coefficient of the SAR historical image; The calibration coefficients of the SAR historical images are verified, and the historical data recalibration results are output.

2. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: Collect calibrated spaceborne SAR historical image data and uncalibrated spaceborne SAR historical image data from different satellites in the same band, including: The calibrated spaceborne SAR historical image data and the spaceborne SAR historical image data to be calibrated are collected from different satellites in the same band with the same coverage area, the same imaging ascending and descending orbits, and imaging time intervals within the threshold range.

3. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: Preprocessing the spaceborne SAR recalibration data pair includes: Performing the same preprocessing operation on the data in the spaceborne SAR recalibration data pair; The preprocessing operations include thermal noise removal, multi-view processing, speckle filtering, and radiation correction.

4. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: The pre-processed spaceborne SAR recalibration data is subjected to terrain correction, including: The pre-processed spaceborne SAR recalibration data are subjected to terrain correction processing using a geocoding method under the same coordinate system, and the pre-processed spaceborne SAR recalibration data are centered, so that image data of different payloads have the same geographic information at corresponding pixel points.

5. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: Performing resolution resampling processing on the recalibrated historical data pair includes: Using a lower resolution of the recalibrated historical data pair as a resampling benchmark; Based on the resampling benchmark, a weighted calculation is performed on each pixel point of the image in the recalibrated historical data pair through a downsampling function to obtain a new pixel value.

6. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: Performing incident angle correction on the recalibrated historical data pair with incident angle differences, comprising: Calculating an incident angle corresponding to the recalibration historical data pair, if the difference between the incident angle and the reference incident angle is within a threshold range, then there is no incident angle difference; If the difference between the incident angle and the reference incident angle exceeds the threshold range, it is determined that there is an incident angle difference, and an incident angle correction model is used for correction.

7. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 1 is characterized in that: Construct recalibration cross-data pairs, including: Extracting a first pixel value corresponding to the historical data to be calibrated in the recalibrated historical data pair, and using the first pixel value as an independent variable of the recalibrated cross data pair; Extracting a second pixel value corresponding to the calibrated standard historical data after downsampling in the recalibrated historical data pair, and using the second pixel value as a dependent variable of the recalibrated cross data pair; The independent variables and the dependent variables are combined to form a two-dimensional data pair, and the two-dimensional data pair is used as data corresponding to the recalibrated cross data pair to construct the recalibrated cross data pair.

8. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 7 is characterized in that: Inputting the recalibrated cross data pair into a Gaussian mixture model, fitting the recalibrated cross data pair through the Gaussian mixture model, and calculating the calibration coefficient of the SAR historical image, including: Inputting the recalibrated cross data pairs into a Gaussian mixture model, extracting the high-frequency region in each of the two-dimensional data pairs, and calculating the interquartile range of the two-dimensional data pairs; Based on the high-frequency area, the bin width is calculated using the Freedman-Diaconis rule, and the joint distribution of the two-dimensional histogram statistical variables is used to obtain the high-frequency area data; If the recalibrated cross data pairs are data pairs of different types of regions, the high-frequency region data are used to fit Gaussian mixture models of each two-dimensional Gaussian distribution, and the initial parameters of the Gaussian mixture model are initialized using the K-means method; A Gaussian mixture model is fitted on the recalibrated cross data pair to calculate the calibration coefficient of the SAR historical image.

9. The SAR historical data recalibration method based on heterogeneous satellite crossover according to claim 8 is characterized in that: The recalibrated cross data pairs are fitted with a Gaussian mixture model to calculate the calibration coefficients of the SAR historical images, including: Fitting the recalibrated cross data pair with a Gaussian mixture model to extract the corresponding expectation and Gaussian weight of the Gaussian distribution in the high-frequency region; The Gaussian weight and the expected ratio corresponding to the Gaussian weight are multiplied to obtain the calibration coefficient of the SAR historical image.

10. A SAR historical data recalibration system based on cross-satellite, characterized in that: The system comprises: The data acquisition module is used to collect the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data of different satellites in the same band, and extract the data using the same polarization mode from the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data as the spaceborne SAR recalibration data pair; A data preprocessing module is used to preprocess the spaceborne SAR recalibration data pair, perform terrain correction on the preprocessed spaceborne SAR recalibration data pair, perform longitude and latitude clipping on the calibrated spaceborne SAR historical image data and the to-be-calibrated spaceborne SAR historical image data after terrain correction, and perform image registration using a SAR image registration method to generate a recalibration historical data pair; A data pair construction module, used for performing resolution resampling processing on the recalibration historical data pairs, and performing incident angle correction on the recalibration historical data pairs with incident angle differences, to construct recalibration cross data pairs; A calibration coefficient calculation module, used for inputting the recalibration cross data pair into a Gaussian mixture model, fitting the recalibration cross data pair through the Gaussian mixture model, and calculating the calibration coefficient of the SAR historical image; The result verification module is used to verify the calibration coefficients of the SAR historical images and output the historical data recalibration results.

Images