Sar satellite absolute radiation calibration method, device, equipment and storage medium

Abstract

The present application relates to the technical field of satellite calibration, and in particular to a SAR satellite absolute radiation calibration method, device, equipment and storage medium, wherein the SAR satellite absolute radiation calibration method comprises: determining a natural distribution target as a calibration field; imaging the calibration field respectively to obtain a calibrated SAR image and a medium-high orbit SAR image; establishing a corresponding relationship between noise power and an incident angle, and removing noise power from the calibrated SAR image and the medium-high orbit SAR image based on the corresponding relationship; performing geometric consistency processing; calculating the backscattering coefficient of the calibration field; and calculating the calibration constant of the medium-high orbit SAR satellite to be calibrated. The above method solves the technical bottleneck of difficulty in arranging a large-size corner reflector outside the field and inability to cover the full field of view of large-width imaging by introducing a calibrated SAR satellite as a radiation reference and combining it with the natural distribution target.

Description

Technical Field

[0001] This invention relates to the field of satellite calibration technology, and more specifically, to a method, apparatus, device, and storage medium for absolute radiometric calibration of SAR satellites. Background Technology

[0002] Absolute radiometric calibration of Synthetic Aperture Radar (SAR) satellites is a crucial technical step in establishing the precise physical relationship between the digital quantization values ​​of SAR images and the true backscattering coefficients of ground objects. Its core lies in solving for the radiometric calibration constant, ensuring that the retrieved backscattering coefficients have absolute physical meaning and quantitative accuracy. This is a prerequisite for the quantitative application of SAR data in fields such as disaster monitoring, land use, and ocean observation. For traditional low-orbit SAR satellites, absolute radiometric calibration typically employs the point target method. This involves deploying corner reflectors with known precise radar cross-sections on the ground as reference targets, and retrieving the calibration constants by measuring their responses in SAR images.

[0003] However, for medium- and high-orbit SAR satellites (such as geostationary orbit SAR) with orbital altitudes reaching tens of thousands of kilometers and single-shot imaging swaths often exceeding several thousand kilometers, traditional point target calibration methods face fundamental defects: due to the high orbital altitude, the signal-to-clutter ratio of ground point targets in the image is extremely low. If small- and medium-sized corner reflectors commonly used in low-orbit satellites are used, it is difficult to accurately extract them from background noise and sidelobe interference. If large-sized corner reflectors are used to ensure the signal-to-clutter ratio, the field deployment operation is extremely inconvenient and the accuracy cannot be guaranteed. At the same time, in order to ensure the calibration accuracy of different wave positions in the wide-swath image, a large number of such large-sized corner reflectors need to be deployed for all wave positions. The cost, operational difficulty, and maintenance workload make it almost impossible to implement in engineering, making it difficult for medium- and high-orbit wide-swath SAR satellites to achieve reliable absolute radiometric calibration. Summary of the Invention

[0004] The purpose of this application is to provide a SAR satellite absolute radiometric calibration method, apparatus, device, and storage medium to solve the above-mentioned technical problems.

[0005] In a first aspect, embodiments of this application provide a method for absolute radiometric calibration of a SAR satellite. The method includes: determining a naturally distributed target as a calibration field; imaging the calibration field using a calibrated SAR satellite and a medium-to-high orbit SAR satellite to be calibrated, respectively, to obtain a calibrated SAR image and a medium-to-high orbit SAR image; establishing a correspondence between noise power and incident angle based on the characteristic that noise power changes with the incident angle in the range direction during wide-swath imaging of the medium-to-high orbit SAR satellite, and removing noise power from the calibrated SAR image and the medium-to-high orbit SAR image based on the correspondence; performing geometric consistency processing on the calibrated SAR image and the medium-to-high orbit SAR image after noise power removal; calculating the backscattering coefficient of the calibration field based on the calibrated SAR image after noise power removal and geometric consistency processing; and calculating the calibration constant of the medium-to-high orbit SAR satellite to be calibrated using the backscattering coefficient of the calibration field and the medium-to-high orbit SAR image after noise power removal and geometric consistency processing.

[0006] Furthermore, before using the natural distribution target as the calibration field, the method further includes: performing a homogeneity test on the candidate region; wherein the homogeneity test is used to evaluate the spatial uniformity of the backscattering characteristics of the candidate region to verify the spatial uniformity of the candidate region as a distribution target; and / or performing a temporal stability test on the candidate region; wherein the temporal stability test is used to evaluate the stability of the backscattering coefficient of the candidate region over time to verify the temporal stability of the candidate region as a distribution target.

[0007] Further, the uniformity test includes: selecting image blocks based on the SAR power image of the candidate region, calculating the mean and variance of the pixel power within the image block, and determining whether the candidate region meets the spatial uniformity requirement as a distribution target based on the mean and the variance; wherein, the mean is used to characterize the average backscattering coefficient of the candidate region; and the variance is used to characterize the degree to which the power value of each pixel within the image block deviates from the mean.

[0008] The temporal stability test includes: acquiring multiple SAR images of the candidate region, calculating the root mean square error of the temporal backscattering coefficient, and determining whether the candidate region meets the temporal stability requirements for being a distributed target based on the root mean square error; wherein, the multiple SAR images meet the conditions of the same satellite, the same ascent and descent orbit, the same incident angle, and different acquisition times; the root mean square error is used to characterize the stability of the backscattering coefficient of the candidate region over time.

[0009] Further, establishing the correspondence between noise power and incident angle includes: dividing the medium-high orbit SAR image into multiple sub-bands along the range direction, with the incident angle variation of each sub-band not exceeding a preset angle threshold, and extracting the average incident angle of each sub-band; obtaining the noise power of each sub-band, and establishing a set of sample pairs of the average incident angle and the noise power; performing curve fitting based on the set of sample pairs to establish a continuous fitting relationship between noise power and incident angle, and obtaining the correspondence between noise power and incident angle.

[0010] Furthermore, obtaining the noise power of each sub-band includes: setting the medium-high orbit SAR satellite to a receive-only mode; wherein, the receive-only mode is a working mode in which the transmitter is turned off and only echo signals are received, so as to obtain a pure noise image without ground object echo signals; and calculating the noise power of each sub-band based on the pure noise image.

[0011] Furthermore, the geometric consistency processing of the calibrated SAR image and the medium-high orbit SAR image after noise power removal includes: performing resolution adaptation processing on the calibrated SAR image to make the resolution of the calibrated SAR image consistent with the resolution of the medium-high orbit SAR image, so as to eliminate the contrast error introduced by the different pixel representation areas; and performing observation geometry normalization processing on the calibrated SAR image and the medium-high orbit SAR image after resolution adaptation processing, so as to eliminate or reduce the backscattering coefficient deviation caused by the difference in incident angle.

[0012] Furthermore, the resolution adaptation processing includes: performing multi-look averaging processing on the single-look complex image of the calibrated SAR image in the azimuth and range directions, so that the equivalent resolution of the calibrated SAR image in the slant range domain is consistent with the resolution of the medium-high orbit SAR image; the observation geometry normalization processing includes: based on the first incident angle of the calibrated SAR satellite and the second incident angle of the medium-high orbit SAR satellite to be calibrated, normalizing the backscattering coefficients observed by the calibrated SAR image to the observation geometry conditions of the medium-high orbit SAR satellite to be calibrated, so as to eliminate the backscattering coefficient deviation caused by the difference in incident angle.

[0013] Secondly, embodiments of this application provide a SAR satellite absolute radiometric calibration device, comprising:

[0014] The calibration field determination module is used to determine the naturally distributed target as the calibration field;

[0015] The imaging control module is used to image the calibration field using calibrated SAR satellites and medium-high orbit SAR satellites to be calibrated, respectively, to acquire calibrated SAR images and medium-high orbit SAR images.

[0016] The noise power removal module is used to establish a correspondence between noise power and incident angle based on the characteristics of noise power variation with distance-to-incident angle during wide-swath imaging of the medium-high orbit SAR satellite, and to remove noise power from the calibrated SAR image and the medium-high orbit SAR image based on the correspondence.

[0017] A geometric consistency processing module is used to perform geometric consistency processing on the calibrated SAR image and the medium-high orbit SAR image after noise power removal;

[0018] The backscattering coefficient calculation module is used to calculate the backscattering coefficient of the calibration field based on the calibrated SAR image after noise power removal and geometric consistency processing.

[0019] The calibration constant calculation module is used to calculate the calibration constant of the medium-high orbit SAR satellite to be calibrated using the backscattering coefficient of the calibration field and the medium-high orbit SAR image after noise power removal and geometric consistency processing.

[0020] Thirdly, embodiments of this application provide an electronic device, the device including: a memory and a processor, wherein the memory and the processor communicate with each other through an internal connection path, the memory is used to store instructions, the processor is used to execute the instructions stored in the memory, and when the processor executes the instructions stored in the memory, the processor causes the processor to perform the method in any of the above-described embodiments.

[0021] Fourthly, embodiments of this application provide a computer-readable storage medium that stores a computer program, wherein when the computer program is run on a computer, the methods in any of the above-described embodiments are executed.

[0022] The advantages or beneficial effects of the above technical solutions include at least the following:

[0023] First, by introducing calibrated SAR satellites as a radiation reference and combining them with naturally distributed targets, a radiation value transfer link was established from the calibrated satellites through distributed targets to the medium- and high-orbit satellites to be calibrated. The large-area statistical averaging characteristics of distributed targets were used to overcome the defect of insufficient signal-to-clutter ratio of artificial point targets when medium- and high-orbit satellites are used for high-orbit imaging. Absolute radiometric calibration without relying on ground artificial corner reflectors was achieved, and the technical bottleneck of large-size corner reflectors being difficult to deploy in the field and unable to cover the entire field of view of wide-swath imaging was solved.

[0024] Secondly, considering the characteristic that the large single imaging swath of medium and high orbit SAR satellites leads to drastic changes in the incident angle in the range direction, resulting in uneven spatial distribution of noise power, a continuous fitting relationship between noise power and incident angle is established. Based on this, precise noise power removal is performed on calibrated satellite images and satellite images to be calibrated. This effectively corrects the near-far signal-to-noise ratio difference of wide-swath images, eliminates the influence of noise power variation with range direction on calibration accuracy, and ensures the accuracy of solving the radiometric calibration constant.

[0025] Third, by performing multi-view processing on the calibrated SAR satellite images to achieve resolution adaptation with the medium- and high-orbit SAR images to be calibrated in the slant range domain, and by performing observation geometry normalization processing based on the difference in incident angle, the backscattering coefficients observed by the calibrated satellites are converted to the observation geometry conditions of the satellites to be calibrated. This effectively eliminates the backscattering coefficient deviation caused by the difference in the area represented by different satellite pixels and the difference in observation angle, ensuring the consistency and reliability of the transmission of radiometric values ​​among multi-satellite data.

[0026] Fourth, by selecting flat, uniform, and time-stable natural feature areas as the distribution target calibration field, the reliance on a large number of large-size artificial corner reflectors in traditional methods is completely avoided. There is no need for the transportation, deployment, measurement, and maintenance of ground equipment, which greatly reduces the implementation cost and operational complexity of absolute radiometric calibration of medium- and high-orbit wide-swath SAR satellites, making routine, large-scale on-orbit calibration of medium- and high-orbit SAR satellites engineering feasible.

[0027] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the embodiments of this application. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 A schematic flowchart illustrating the SAR satellite absolute radiometric calibration method provided in this application embodiment;

[0030] Figure 2 A flowchart illustrating a SAR satellite absolute radiometric calibration method in a specific application scenario provided in this application embodiment;

[0031] Figure 3 This is a schematic diagram of the structure of the SAR satellite absolute radiometric calibration device provided in the embodiments of this application;

[0032] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0033] The embodiments of the technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only examples, not intended to limit the scope of protection of the present invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and the foregoing description of the accompanying drawings, are intended to cover non-exclusive inclusion. In the description of the embodiments of this application, technical terms such as "first," "second," etc., are only used to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly indicating the number, specific order, or primary or secondary relationship of the indicated technical features. In the description of the embodiments of this application, "a plurality of" means two or more, unless otherwise explicitly specified. The reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0034] Absolute radiometric calibration of SAR satellites involves establishing the image digital number (DN) and the true backscattering coefficient (σ) of ground objects. 0 The key to determining the precise physical relationship between σ and σ lies in solving for the scaling constant, which allows the inverted σ to be obtained. 0 The absolute physical meaning and quantitative accuracy of SAR data are prerequisites for its quantitative application in fields such as disaster monitoring, land use, and ocean observation.

[0035] For traditional low-Earth orbit SAR satellites, absolute radiometric calibration typically employs the "point target method," which involves deploying corner reflectors with known and precise radar cross-sections (RCS) on the ground as reference targets and retrieving calibration constants by measuring their responses in SAR images. However, for the rapidly developing medium- and high-Earth orbit SAR satellites (such as geostationary orbit SAR), this method faces significant challenges:

[0036] 1. Difficulty in target identification: Medium and high orbit satellites can reach altitudes of tens of thousands of kilometers. Ground point targets, if using small to medium-sized corner reflectors (right-angled sides less than 3m) commonly used in low orbit, have extremely low signal-to-clutter ratios in the images, making them difficult to accurately extract from background noise and sidelobe interference. To achieve high signal-to-clutter ratios for point targets, large-sized corner reflectors are required. For example, the 20m resolution images from the Land Survey-4 01 satellite require at least a trihedral corner reflector with right-angled sides of 5m or more, which is extremely inconvenient to operate in the field and cannot guarantee accuracy.

[0037] 2. High engineering implementation cost: Medium and high orbit SAR satellites (such as Land Survey 4 01 satellite) adopt a multi-wavelength scanning mode, and the swath width of a single imaging often exceeds 3,000 kilometers. In order to ensure the absolute radiometric calibration accuracy of different wavelengths of the entire image, a large number of large-size corner reflectors are deployed as point targets for radiometric calibration in all wavelength images. The cost, operation and maintenance difficulty make this method almost impossible to implement in engineering.

[0038] Therefore, finding a calibration constant calculation method that does not rely on artificial point targets and can adapt to the orbit and imaging characteristics of medium and high orbit SAR satellites has become a technical bottleneck that urgently needs to be overcome in this field.

[0039] In view of this, this application provides a method for absolute radiometric calibration of SAR satellites. This method introduces a calibrated SAR satellite as a radiometric reference and combines it with naturally distributed targets to establish a radiometric value transfer link from the calibrated satellite through the distributed targets to the medium- and high-orbit satellite to be calibrated. It overcomes the defect of insufficient signal-to-clutter ratio of artificial point targets when medium- and high-orbit satellites are used for high-orbit imaging by utilizing the large-area statistical averaging characteristics of distributed targets. It realizes absolute radiometric calibration without relying on ground artificial corner reflectors and solves the technical bottleneck of the difficulty in deploying large-size corner reflectors in the field and the inability to cover the entire field of view of wide-width imaging.

[0040] like Figure 1 As shown in the embodiment of this application, a method for calibrating the absolute radiometric value of a SAR satellite is provided, including:

[0041] Step S110: Determine the natural distribution target as the calibration field.

[0042] The aforementioned naturally distributed targets refer to natural landform areas with a certain spatial extension, whose radar backscattering characteristics exhibit statistical uniformity in space and remain relatively stable over time. These typically include flat terrain with uniform surface media, such as deserts, grasslands, or bare soil. The statistically averaged characteristics of their planar scattering units provide a stable radar echo power reference. The calibration field refers to a specific geographical area, pre-selected and verified, used for SAR satellite absolute radiometric calibration. This area must meet SAR imaging geometry requirements and possess quantifiable backscattering coefficient reference values, serving as a physical benchmark connecting the digital quantized values ​​of SAR images with the actual radar scattering characteristics of ground objects.

[0043] It is understandable that using naturally distributed targets as the calibration field aims to solve the technical problem that traditional manual point target calibration methods fail due to the high orbital altitude and wide imaging swath of medium- and high-orbit SAR satellites. On the one hand, the orbital altitude of medium- and high-orbit satellites can reach tens of thousands of kilometers, resulting in extremely low signal-to-clutter ratios for conventionally sized corner reflectors in the images, making accurate extraction impossible. If large-sized corner reflectors are used, the accuracy of field deployment is difficult to guarantee, and operation is extremely inconvenient. On the other hand, wide-swath imaging requires covering a field of view of thousands of kilometers, making it almost impossible to deploy a large number of artificial point targets for all wave positions in engineering. Naturally distributed targets do not require ground equipment deployment and can directly utilize their existing large-area uniform scattering characteristics as a radiation reference, thereby achieving accurate solutions for calibration constants without relying on artificial facilities.

[0044] Optionally, before step S110, the above-mentioned SAR satellite absolute radiometric calibration method may further include: performing a uniformity test on the candidate region; wherein the uniformity test is used to evaluate the spatial uniformity of the backscattering characteristics of the candidate region to verify the spatial uniformity of the candidate region as a distribution target; and / or performing a temporal stability test on the candidate region; wherein the temporal stability test is used to evaluate the stability of the backscattering coefficient of the candidate region over time to verify the temporal stability of the candidate region as a distribution target.

[0045] It is understandable that the above scheme performs uniformity and temporal stability checks on candidate regions because: the spatial uniformity and temporal stability of the distributed target, as the physical benchmark for SAR absolute radiometric calibration, directly determine the accuracy and reliability of the radiometric value transmission; if there is spatial non-uniformity in the candidate region, it will cause the pixel power statistical distribution within the image patch to be discrete, making it impossible to obtain a representative average backscattering coefficient; and if there is temporal instability, the backscattering coefficient reference value established based on historical observation data will not accurately reflect the true radiometric characteristics at the current imaging moment, thereby introducing calibration error and affecting the solution accuracy of the calibration constant of medium and high orbit SAR satellites.

[0046] The above uniformity test is an evaluation process of the spatial consistency of radar backscattering characteristics in the candidate calibration area. It aims to verify whether the area has statistically significant spatial uniformity within the SAR image coverage area, that is, whether the dispersion of the echo power of each scattering unit in the area meets the technical requirements for a distributed target.

[0047] The aforementioned temporal stability test is an evaluation process of the temporal consistency of the backscattering coefficients of the candidate calibration region. It aims to verify whether the backscattering characteristics of the region remain constant under different acquisition phases, that is, to exclude radar echo fluctuations caused by changes in the surface condition over time, and to ensure that it has the ability to serve as a long-term stable radiation benchmark.

[0048] Optionally, the above uniformity test includes: selecting image blocks based on the SAR power image of the candidate region, calculating the mean and variance of the pixel power within the image block, and determining whether the candidate region meets the spatial uniformity requirements as a distribution target based on the mean and variance; wherein, the mean is used to characterize the average backscattering coefficient of the candidate region; and the variance is used to characterize the degree to which the power value of each pixel within the image block deviates from the mean.

[0049] In the above scheme, selecting representative image patches on the SAR power image is the fundamental step in implementing uniformity testing. This technique discretizes a large-scale, continuously distributed target into statistically analyzable sample units by defining a two-dimensional pixel array of a specific size within the candidate calibration region. This allows for a quantitative assessment of the spatial consistency of radar scattering characteristics while maintaining the physical continuity of ground features. The selection of image patches must cover the main land cover types in the candidate area and avoid boundary effects and strong scattering anomalies to ensure that the extracted statistical features accurately reflect the inherent scattering properties of the area rather than local interference.

[0050] The core technical approach for evaluating spatial uniformity is to statistically calculate the mean and variance of the power values ​​of all pixels within a selected image patch. The mean pixel power, as a measure of the central tendency of the image patch's energy response, physically corresponds to the average radar backscattering coefficient per unit area of ​​land cover within that region, reflecting the basic scattering level of the calibration site under specific radar wavelengths and incident angles. The variance of the pixel power characterizes the degree of dispersion of each scattering unit within the image patch relative to this mean. The magnitude of the variance directly quantifies the range of echo power fluctuations caused by factors such as surface micro-topographic undulations, spatial variations in dielectric constant, and radar speckle noise.

[0051] An example implementation of determining the spatial uniformity of candidate regions based on the aforementioned statistical measures is as follows: When the variance value is within a preset reasonable range, it indicates that the fluctuation of pixel power within the image patch is within an acceptable range of statistical fluctuations, and the backscattering characteristics of ground features exhibit a high degree of consistency, satisfying the spatial uniformity condition required for distribution targets. Conversely, if the variance exceeds the threshold, it means that the region has significant ground feature heterogeneity or scattering inhomogeneity, and is not suitable as a reference region for radiometric calibration. Through this quantitative evaluation based on pixel-level power statistics, uniform natural ground features with stable scattering characteristics can be objectively screened, laying a reliable spatial benchmark for subsequent absolute radiometric calibration using the statistical averaging principle.

[0052] Optionally, the aforementioned temporal stability test includes: acquiring multiple SAR images of the candidate region, calculating the root mean square error of the temporal backscattering coefficient, and determining whether the candidate region meets the temporal stability requirements for being a distributed target based on the root mean square error; wherein, the multiple SAR images meet the conditions of the same satellite, the same ascent and descent orbits, the same incident angle, and different acquisition times; the root mean square error is used to characterize the stability of the backscattering coefficient of the candidate region over time.

[0053] The aforementioned temporal stability test is a crucial step in evaluating the temporal consistency of backscattering coefficients in candidate calibration fields. Its technical necessity stems from the fundamental requirement of stable radiometric references in SAR absolute radiometric calibration. Since calibration operations often rely on historical observation data or satellite data spanning different time phases for radiometric value transfer, the surface physical properties of the calibration field must remain constant over the observation time span to eliminate backscattering coefficient drift caused by factors such as changes in soil moisture content, vegetation phenological cycles, changes in surface roughness, or human activities. This ensures that the radiometric references derived from historical observations still possess physical validity and numerical comparability at the current imaging time.

[0054] To accurately isolate the independent influence of time factors on the backscattering coefficient, the acquisition of multi-period SAR images requires strict control over the high consistency of observation geometry conditions, namely, using the same satellite platform, the same ascent and descent orbit directions, and the same radar beam incident angle. This strict control strategy aims to eliminate differences in scattering mechanisms caused by variations in satellite system response, changes in observation direction (leading to changes in the interaction between the radar beam and surface micro-topography), or changes in incident angle (causing differences in radar echo intensity with angle response). This ensures that fluctuations in the backscattering coefficient in the time series purely reflect the temporal variation of the true physical state of the surface, rather than spurious changes caused by variations in observation geometry conditions.

[0055] The root mean square error (RMSE), as a core statistical indicator for quantifying temporal stability, characterizes the dispersion level and fluctuation range of the calibration field's backscattering characteristics over time by calculating the deviation of the average value of each period in the backscattering coefficient sequence from the overall mean. Mathematically, this indicator comprehensively considers the cumulative effect of the deviation between observed and expected values ​​in each period, sensitively capturing subtle changes in surface conditions. Its magnitude directly quantifies the ability of a candidate region to maintain constant radar scattering characteristics within the observation period, providing a quantifiable numerical basis for objectively assessing the temporal stability of the calibration field.

[0056] Based on the calculation results of the root mean square error (RMSE), a preset stability threshold is used to determine whether the candidate region meets the temporal stability requirements for distribution targets. Specifically, when the RMSE is at a low level, it indicates that the backscattering coefficient of the region fluctuates slightly within the observation period, qualifying it as a long-term stable radiation benchmark and supporting cross-temporal radiometric calibration operations. Conversely, if the RMSE exceeds the acceptable range, it indicates that the surface condition has significant time-varying characteristics, and its scattering characteristics lack time repeatability, making it unsuitable as a reference benchmark for absolute radiometric calibration. This effectively avoids calibration errors introduced by the temporal instability of the calibration field, ensuring the temporal consistency of radiation value transmission and the reliability of calibration results.

[0057] Step S120: Use the calibrated SAR satellite and the medium-high orbit SAR satellite to be calibrated to image the calibration field, and obtain the calibrated SAR image and the medium-high orbit SAR image.

[0058] The reason why step S120 above uses a calibrated SAR satellite as a radiation reference source is that the calibration constants of the calibrated satellite have been accurately determined through prior calibration operations, and there is a known and traceable physical relationship between the pixel power of the acquired SAR images and the backscattering coefficient of ground objects, thus serving as a starting point for the transfer of absolute radiation values. This satellite is typically a low-orbit SAR satellite with a mature on-orbit calibration history and stable radiation performance. Its observation data can provide reference values ​​for the absolute backscattering coefficient of the calibration field area under specific radar bands and observation geometry conditions, providing a physical anchor point for establishing the radiation reference of the satellite to be calibrated subsequently. The medium- and high-orbit SAR satellites to be calibrated have the technical characteristics of high orbital altitude and large single-shot imaging swath width. Their orbital altitude can reach tens of thousands of kilometers, and their imaging swath width often exceeds several thousand kilometers. This high-orbit, wide-swath imaging mode makes traditional calibration methods relying on manual point targets difficult to implement. By imaging the same target area using the satellite, raw data containing radar echo information of the calibration field is obtained, forming a SAR image to be processed. The conversion relationship between the pixel power of the image and the true backscattering coefficient of the ground objects needs to be established by solving the calibration constant.

[0059] The essence of dual-satellite collaborative observation lies in constructing a radiometric value transfer link from the calibrated satellite through naturally distributed targets to the satellite to be calibrated. Specifically, since both satellites observe the same ground target, the backscattering coefficients of this calibration field are physically unique and consistent. Although the observation geometry, spatial resolution, and radar system parameters of the two satellites differ, the inherent scattering characteristics of the ground target provide a common radiometric reference for both. Through this dual-satellite cross-observation mode, the absolute radiometric reference of the calibrated satellite can be transferred to the satellite to be calibrated, achieving calibration constant inversion without the need for manual point target intervention.

[0060] In addition, dual-satellite observations must be completed within a window of no more than one week to ensure that the physical state of the calibration site (such as soil moisture and vegetation cover) has not changed significantly. At the same time, imaging is required to be carried out under similar meteorological conditions to avoid changes in surface roughness or dielectric properties caused by weather factors such as precipitation and wind. Furthermore, the same ascending and descending orbit directions and left and right side-looking settings are used to ensure that the interaction between the radar beam and the surface micro-topography is consistent, thereby minimizing the changes in backscattering coefficients caused by differences in observation conditions and ensuring the comparability of observation data from the two satellites and the accuracy of radiometric value transmission.

[0061] Step S130: Based on the characteristic that noise power changes with the incident angle in the range direction during wide-swath imaging of medium- and high-orbit SAR satellites, establish the correspondence between noise power and incident angle, and remove noise power from calibrated SAR images and medium- and high-orbit SAR images respectively based on the correspondence.

[0062] In the wide-swath imaging mode of medium- and high-orbit SAR satellites, the coverage area of ​​a single imaging session can reach thousands of kilometers. This results in a significant spatial gradient change in the incident angle of the radar beam in the range direction, with the difference in incident angle between the near and far ends of the entire orbit image reaching tens of degrees. This large-span change in incident angle directly affects the spatial distribution characteristics of the system noise power, causing the noise power to no longer exhibit a constant distribution as assumed by traditional narrow-swath imaging. Instead, it shows a systematic variation trend with range position. The physical root cause lies in the coupling relationship between factors such as radar receiver noise figure, antenna gain pattern, and range transmission loss and the incident angle.

[0063] The reason for establishing the correspondence between noise power and incident angle is that it is necessary to accurately extract the noise power that varies with spatial location from the original echo signal in order to recover the true radar scattering contribution of ground targets. If this noise power variation dependent on the incident angle is not considered and the global average noise power is directly used for subtraction, it will lead to a systematic deviation in the signal-to-noise ratio correction between the near and far ends of the wide-swath image. If the noise power is underestimated when the near-end incident angle is small, the effective signal will be over-subtracted, while if the noise power is overestimated when the far-end incident angle is large, too much noise component will remain. Both will destroy the physical quantitative relationship between image pixel power and backscattering coefficient.

[0064] The above scheme performs noise power removal based on the incident angle characteristics on both calibrated SAR images and medium-to-high orbit SAR images to ensure that the observation data from the two satellites have a consistent physical basis at the radiometric processing level. Since the subsequent calculation of calibration constants depends on the cross-comparison of the observation data from both satellites, any residual noise component from either satellite will directly contribute to the calculation error of the final calibration constants. Therefore, it is necessary to perform precise noise power removal on both images so that the power of the processed image strictly corresponds to the backscattered echo of ground objects, eliminating the additional power introduced by system thermal noise, and providing a clean signal basis for the subsequent calibration constant inversion based on the physical model.

[0065] Optionally, the above-mentioned establishment of the correspondence between noise power and incident angle includes: dividing the medium- and high-orbit SAR image into multiple sub-bands along the range direction, with the incident angle variation of each sub-band not exceeding a preset angle threshold, and extracting the average incident angle of each sub-band; obtaining the noise power of each sub-band, and establishing a set of sample pairs of average incident angle and noise power; performing curve fitting based on the set of sample pairs to establish a continuous fitting relationship between noise power and incident angle, and obtaining the correspondence between noise power and incident angle.

[0066] The aforementioned division of SAR images into multiple sub-bands along the range direction is a spatial discretization strategy to address the continuously varying incident angle characteristics of wide-swath imaging. By dividing the continuously varying range into several intervals with limited incident angle spans, the incident angle can be approximated as constant within each sub-band, thereby obtaining a representative average incident angle parameter within that local area. The incident angle variation corresponding to each sub-band does not exceed a preset angle threshold. This technical constraint aims to balance the contradiction between incident angle resolution and statistical sample size: a threshold that is too small can improve the incident angle positioning accuracy, but it will lead to insufficient pixel samples within the sub-band and an increase in the variance of noise power statistical estimation; a threshold that is too large can increase the sample size and improve statistical stability, but it will blur the details of incident angle variation and lose the sensitivity of noise power to changes with angle.

[0067] The above scheme extracts the average incident angle of each sub-band, which essentially determines the radar observation geometric center corresponding to that sub-band. This parameter serves as a physical representation of the sub-band's spatial location, establishing a mapping relationship between image coordinates and radar beam illumination angle. The set of sample pairs constructed based on the average incident angle of each sub-band and its corresponding noise power constitutes a discrete observation dataset describing the variation of noise power with the incident angle. Mathematically, this set is presented as a scatter distribution on a two-dimensional plane, with each data point representing a measured value of the system noise level at a specific observation angle. Curve fitting is performed based on the sample pair set, aiming to approximate the discrete observation data with a continuous mathematical function model, establishing a continuous fitting relationship between noise power and incident angle. This mapping process from discrete samples to a continuous function not only smooths the measurement noise caused by random fluctuations but, more importantly, enables the interpolation and prediction capability of noise power at any incident angle position. By obtaining the continuous correspondence between noise power and incident angle, the discrete sub-band level noise estimation can be extended to a continuous noise power model along the range direction, thereby providing an accurate noise power estimate matching its corresponding incident angle for each pixel position in the image, supporting subsequent pixel-by-pixel noise power removal operations.

[0068] Optionally, obtaining the noise power of each sub-band includes: setting the medium- and high-orbit SAR satellite to a receive-only mode; wherein, the receive-only mode is a working mode in which the transmitter is turned off and only echo signals are received, so as to obtain a pure noise image without ground object echo signals; and calculating the noise power of each sub-band based on the pure noise image.

[0069] The above describes setting medium- and high-orbit SAR satellites to receive-only mode, i.e., turning off the radar transmitter while keeping the receiver in normal working condition. In this mode, the radar system does not actively emit electromagnetic energy; the receiving channel only collects internal thermal noise and background radiation noise from the external environment, forming a pure noise image without any backscattered echo components from ground objects. The technical essence of this working mode lies in artificially constructing a signal-input-free receiving state, allowing the digital sampled values ​​output by the receiver to purely reflect the system's background noise level under echo-free conditions, thus decoupling noise power measurement from the complex signal-plus-noise mixture environment. The purpose of acquiring a pure noise image without ground object echo signals is to provide a clean benchmark for noise power estimation, uncontaminated by ground object scattering characteristics. In conventional imaging modes, the signal received by the radar is a superposition of ground object backscattered echoes and system noise, which are difficult to directly separate in the power domain. However, the pure noise image completely eliminates the presence of ground object echoes through physical means, making the power value of each pixel in the image directly correspond to the system noise power at that receiving time and location in the range direction, laying the data foundation for subsequent accurate quantification of noise levels.

[0070] The above calculation of noise power for each sub-band based on a purely noisy image essentially utilizes a statistical averaging method to extract representative noise power values ​​corresponding to a specific incident angle within a defined sub-band spatial range. Since the pixel power of a purely noisy image contains only random noise components, by performing an arithmetic average of all pixel powers within the sub-band, the random fluctuations of the noise itself can be effectively suppressed, obtaining a stable mean noise power value under that incident angle condition. This establishes an accurate mapping sample between the incident angle and noise power, providing reliable observational data support for constructing a continuous fitting model of noise power variation with the incident angle.

[0071] Step S140: Perform geometric consistency processing on the calibrated SAR image and the medium-to-high orbit SAR image after noise power removal.

[0072] Optionally, step S140 may include: performing resolution adaptation processing on the calibrated SAR image to make the resolution of the calibrated SAR image consistent with the resolution of the medium- and high-orbit SAR image, so as to eliminate the contrast error introduced by the different pixel representation areas; and performing observation geometry normalization processing on the calibrated SAR image and the medium- and high-orbit SAR image after resolution adaptation processing, so as to eliminate or reduce the backscattering coefficient deviation caused by the difference in incident angle.

[0073] It is understandable that SAR satellites at different orbital altitudes typically exhibit significant differences in azimuth and range spatial resolution due to variations in imaging geometry and system design. Low-Earth orbit (LEO) calibrated satellites often possess higher spatial resolution, with each pixel in their complex image corresponding to a smaller ground area. In contrast, medium- and high-Earth orbit (MEO) satellites, due to their increased orbital altitude, are usually designed with coarser resolution to maintain synthetic aperture processing gain and signal-to-noise ratio, resulting in a larger physical area covered by each pixel. This difference in the area represented by a pixel directly leads to differences in radar echo power. Since received power is proportional to the illuminated area, directly comparing the pixel power of images at different resolutions introduces a systematic area-weighted error unrelated to the actual scattering characteristics of the ground, undermining the comparability of radiation values. The observation geometry configurations of calibrated satellites and MEO satellites awaiting calibration for the same distributed target typically differ, primarily in the radar beam incident angle. The radar backscattering coefficient of natural ground features exhibits significant angle dependence, and the interaction between surface roughness, dielectric properties, and the electromagnetic wavefront displays nonlinear response characteristics as the incident angle changes. When two satellites observe the same ground feature at different incident angles, even if the physical properties of the feature remain constant, the intensity of its radar echo will differ due to the difference in the angular response function. Ignoring this geometric difference in observation and directly using the backscattering coefficient measured by the calibrated satellite as the reference value for the satellite to be calibrated will result in a radiative transmission deviation caused by the angular effect, leading to a shift in the calibration reference during geometric transformation. By performing resolution adaptation processing on the calibrated SAR image, adjusting its spatial resolution to match that of the medium-to-high orbit SAR image to be calibrated, the pixels of the two images achieve a strict match in terms of ground coverage area, thereby eliminating the power comparison error introduced by the difference in the area of ​​the resolution unit.

[0074] Furthermore, the observation geometry normalization process establishes a correction model for the backscattering coefficient as a function of the incident angle. This transforms the radiometric values ​​observed by the calibrated satellite under its actual incident angle conditions to those under the incident angle geometry conditions of the satellite to be calibrated, thus unifying the radiometric reference between different observation angles. These two geometric consistency processes together construct a radiometric value transfer channel across different satellite platforms and observation conditions. This ensures that in subsequent calibration constant calculations, the observation data from the two satellites can be compared within a consistent physical space and geometric reference frame, laying a data consistency foundation for achieving high-precision absolute radiometric calibration based on distributed targets.

[0075] Optionally, the resolution adaptation process described above includes: performing multi-view averaging on the single-view complex image of the calibrated SAR image in the azimuth and range directions, so that the equivalent resolution of the calibrated SAR image in the slant range domain is consistent with the resolution of the medium- and high-orbit SAR image.

[0076] Single-look complex images, as advanced outputs of SAR raw imaging processing, retain the phase information of radar echo signals, maintaining the highest spatial resolution achievable by the system in both the azimuth and range directions. Each pixel in this type of image contains a complex value, corresponding to the in-phase and quadrature components of the radar echo, and its intensity is calculated by squaring the complex modulus. Since the range resolution of a SAR system is determined by the transmitted signal bandwidth, and the azimuth resolution by the synthetic aperture length, low-Earth orbit SAR satellites typically have fine resolution in both dimensions, resulting in a small ground area corresponding to a single pixel. Multi-look averaging is an incoherent processing technique that sacrifices spatial resolution for improved radiometric quality. It is implemented by combining and averaging multiple adjacent single-look pixels in both the azimuth and range directions. In the azimuth direction, multi-look processing reduces speckle noise caused by radar coherent imaging mechanisms by merging signals from different Doppler frequency ranges; in the range direction, it further smooths random scattering fluctuations by merging signals within adjacent slant range gates. This processing method merges the energy responses of multiple high-resolution pixels into a single lower-resolution pixel, making the processed image statistically closer to the macroscopic scattering characteristics of natural ground objects. The slant range domain, as the inherent coordinate system of the raw SAR data, directly corresponds to the sampling interval of the radar signal processing, while the ground range domain resolution is modulated by the incident angle. By adapting the resolution of the calibrated SAR image in the slant range domain to match that of the high-orbit SAR image to be calibrated, it ensures that the pixels in both images have the same slant range sampling interval in the range direction, thus achieving physical consistency in the area of ​​the ground resolution unit represented by the pixel. When resolution matching is achieved in both the azimuth and range directions through multi-look processing, the physical ground area represented by each pixel in the two images tends to be consistent, allowing the calculation of the backscattering coefficient based on pixel power to be established on the same area benchmark, effectively eliminating the systematic deviation in power values ​​caused by resolution differences.

[0077] The aforementioned resolution adaptation technique essentially reconstructs the spatial sampling consistency of two images through spatial filtering, enabling radiometric comparisons of observation data from different satellite platforms with different original resolutions at a unified ground resolution unit scale. The pixel power of the adapted calibrated SAR image is comparable to that of medium- and high-orbit SAR images because they correspond to radar echo energy integrals within the same size ground resolution unit, thus supporting subsequent absolute radiometric calibration calculations based on distributed targets.

[0078] Optionally, the above observation geometry normalization process includes: based on the first incident angle of the calibrated SAR satellite and the second incident angle of the medium-high orbit SAR satellite to be calibrated, normalizing the backscattering coefficient of the calibrated SAR image observation to the observation geometry conditions of the medium-high orbit SAR satellite to be calibrated, so as to eliminate the backscattering coefficient deviation caused by the difference in incident angle.

[0079] Due to differences in orbital altitude and beam pointing design, low-Earth orbit SAR satellites and medium-to-high orbit SAR satellites to be calibrated often produce different radar beam incidence angles when observing the same distributed target. The numerical difference between the first incidence angle of the calibrated satellite and the second incidence angle of the satellite to be calibrated directly alters the physical geometry of the interaction between electromagnetic waves and surface micro-topography and dielectric properties. The radar backscattering coefficient of natural features is not a constant physical quantity, but exhibits a specific angular response curve as the incidence angle changes. Factors such as surface roughness, local slope distribution, and penetration depth all exhibit differentiated scattering mechanisms under different incidence conditions, resulting in significantly different radar echo intensities for the same feature under different incidence angles. Normalizing the backscattering coefficient obtained from the observations of the calibrated SAR images to the observation geometry of the medium-to-high orbit SAR satellite to be calibrated essentially establishes an angular domain radiative value conversion mechanism. This technique is based on the quantitative difference in the incident angles of the two satellites. The backscattering coefficient value measured by the calibrated satellite at its specific incident angle is mapped, through an angle response model, to the geometric conditions corresponding to the second incident angle of the satellite to be calibrated. This allows for the calculation of the equivalent backscattering coefficient of the ground feature at the observation angle of the satellite to be calibrated. This normalization transformation enables comparison of radiometric measurements from different observation geometries on a unified incident angle reference, eliminating differences in scattering characteristics caused by different observation angles.

[0080] The above scheme effectively eliminates the backscattering coefficient deviation caused by differences in incident angle by implementing observation geometry normalization, ensuring that the absolute radiation reference provided by the calibrated satellite can be accurately transferred to the observation frame of the satellite to be calibrated. The normalized backscattering coefficient value reflects the scattering characteristics of ground objects under the actual incident conditions of the satellite to be calibrated, making the distributed target radiation reference calculated based on the data of the calibrated satellite consistent with the observation data of the satellite to be calibrated geometrically. This supports the accurate solution of subsequent calibration constants and avoids systematic radiative transfer errors introduced by inconsistencies in observation geometry.

[0081] Step S150: Calculate the backscattering coefficients of the calibration field based on the calibrated SAR image after noise power removal and geometric consistency processing.

[0082] After noise power removal and geometric consistency processing, the pixel power values ​​of the calibrated SAR image accurately correspond to the true radar echo contribution of ground objects, eliminating error components introduced by system thermal noise and observation geometric differences. Since the satellite's radiometric calibration constant has been accurately determined through prior calibration, a definite physical quantitative relationship exists between image pixel power and ground object backscattering coefficients. This relationship is rigorously described by the absolute calibration model derived from the radar equations, involving key physical quantities such as radar system parameters, transfer functions, and ground resolution unit areas. By selecting an image patch covering the calibration field on the calibrated SAR image, the average power value of all pixels within that region is calculated. This statistic represents the average radar echo intensity of the calibration field under the observation conditions of the calibrated satellite. This average power value has been stripped of system noise components through previous noise power removal processing and corrected for resolution differences and incident angle effects through geometric consistency processing, thus accurately reflecting the true scattering energy level of the calibration field. Based on the mathematical expression of the absolute calibration model, the backscattering coefficient of the calibration field can be obtained by dividing the average power value of the image patch by the product of the calibration constant of the calibrated satellite and the area of ​​the corresponding ground resolution unit. This coefficient, expressed in decibels or as a linear value, characterizes the radar cross-section of a unit area of ​​ground features in the calibration field under specific radar bands and incident angles, and is a radiation quantity with absolute physical significance. This quantity represents both the inherent scattering properties of the calibration field under the observation geometry of the calibrated satellite and, through prior geometric normalization processing, the ability to map to the observation geometry of the satellite to be calibrated, thus constituting a key physical quantity connecting the radiation references of the two satellites.

[0083] The obtained backscattering coefficients of the calibration field can serve as an absolute radiometric reference. Their values ​​are independent of specific satellite platforms or imaging conditions, and are determined solely by the electromagnetic scattering characteristics of the ground objects themselves. The backscattering coefficients will be used as the true input values ​​for subsequently calculating the calibration constants of the high-orbit SAR satellites to be calibrated. This allows the satellites to calibrate to deduce their own radiometric calibration constants by comparing their observed power with this absolute reference, thereby establishing a conversion relationship between their image digital quantization values ​​and the true backscattering coefficients of the ground objects.

[0084] Step S160: Calculate the calibration constants of the medium-high orbit SAR satellite to be calibrated using the backscattering coefficients of the calibration field and the medium-high orbit SAR images after noise power removal and geometric consistency processing.

[0085] After noise power removal and geometric consistency processing, the pixel power values ​​of medium- and high-orbit SAR images accurately represent the backscattered echo energy of ground objects received by the radar, eliminating interference from system noise and observation geometric differences. A linear relationship exists between the image pixel power and the true backscattering coefficient of ground objects, described by radar equations. This relationship includes a scaling factor characterizing the overall transmission characteristics of the radar system, i.e., the radiometric calibration constant to be solved. This calibration constant integrates parameters such as radar transmit power, antenna gain, wavelength, system loss, and imaging transfer function. For a specific SAR system, it is a constant with definite physical meaning, and its value directly determines the scaling ratio for converting the image digital quantization values ​​to backscattering coefficients.

[0086] The backscattering coefficient of the calibration field, as an absolute radiance value accurately measured by calibrated satellites, constitutes the key true input for calculating the calibration constants of the satellite to be calibrated. This coefficient represents the inherent electromagnetic scattering characteristics of a specific natural feature under radar wave illumination and does not change with the observation satellite platform. By comparing the average pixel power of the corresponding calibration field region on a medium-to-high orbit SAR image with this absolute benchmark, and using the mathematical relationship of the absolute calibration model, the pixel power is divided by the product of the calibration field backscattering coefficient and the area of ​​the corresponding ground resolution unit, thus retrieving the calibration constants of the medium-to-high orbit SAR system.

[0087] The obtained calibration constant establishes a precise physical conversion relationship between pixel power in medium- and high-orbit SAR satellite images and backscattering coefficients of ground objects. This allows any SAR image acquired by the satellite to have its original digital quantization values ​​converted into backscattering coefficients with absolute physical meaning using this constant. Once determined, this constant becomes the fundamental parameter for the satellite's radiometric calibration and can be used for the quantitative processing of all subsequent data from similar imaging modes, realizing the conversion from relative radiometric measurements to absolute radiometric measurements.

[0088] The above scheme, by completing the calculation of calibration constants, enables the medium- and high-orbit SAR satellites to be calibrated to provide quantitative absolute radiometric data products. The image pixel values ​​obtained by the satellites can be accurately inverted through the calibration equation to retrieve the backscattering coefficients of ground objects, which meets the technical requirements for the accuracy of radiometric values ​​in quantitative remote sensing applications such as disaster monitoring, ground object classification, and parameter inversion. Thus, high-precision absolute radiometric calibration of medium- and high-orbit wide-swath SAR satellites is achieved without relying on ground-based artificial point targets.

[0089] like Figure 2 As shown, to facilitate understanding of the working principle of the above-described SAR satellite absolute radiometric calibration method, this application embodiment also provides a specific application example of this method in a certain application scenario. In this application scenario, the above-described SAR satellite absolute radiometric calibration method mainly includes:

[0090] Step 1: Establishing the absolute calibration model;

[0091] Based on the characteristics of medium- and high-orbit SAR imaging, an absolute calibration model is established that takes into account the effects of noise, ground object resolution, and other factors.

[0092] Based on the radar equations, the relationship between the backscattering coefficient of distributed targets and the pixel power of radar images can be obtained. :

[0093]

[0094] in, For radar transmission power, For antenna gain, Antenna direction angle The distance at which the radar reaches its target. For system losses, For the signal wavelength, This refers to the radar cross-section per unit area. This represents the area of ​​the radar ground resolution unit. Let be the transfer function of the radar from the antenna output to the image processor output. This represents noise power.

[0095] Ground resolution unit in the formula for:

[0096]

[0097] in, This refers to the azimuth spatial resolution. For ground distance resolution, For slant range resolution, The angle of incidence is denoted as .

[0098] Regarding the above formula, let:

[0099]

[0100] in, The value includes the transfer functions of various parts such as radar reception, recording, and imaging processing. For the same SAR system, the transmit power... ,wavelength System parameters can be considered constants, while system losses... Antenna gain Parameters such as pixel slant range R can also be considered constants after system calibration, antenna pattern correction, and incident angle correction. Therefore, The calibration constants obtained for radiation calibration.

[0101] The general form of the scaling equation can be simplified as follows:

[0102]

[0103]

[0104] Step 2: Target selection and imaging;

[0105] S2.1: Target selection for distribution;

[0106] Based on the orbital characteristics of medium- and high-orbit SAR satellites, a natural area with flat terrain, uniform surface, and stable temporal sequence within the coverage area of ​​medium- and high-orbit SAR satellites was selected as the calibration site. Several uniform image patches were then selected within this area as distributed targets to verify the flatness, uniformity, and stability of the distributed targets.

[0107] Ground coverage area of ​​medium and high orbit SAR satellites:

[0108]

[0109] in, The altitude of the satellite orbit. For the track inclination angle, The azimuth beam angle, For pitch beam angle, For time parameters, Right ascension of the ascending node, The perigee argument, It is the angle closest to the point.

[0110] Flatness check: Based on topographic maps and the Global Digital Elevation Model (DEM, resolution) Slope analysis: Regional surface slope °(Preferred) °).

[0111] Uniformity test: Calculate the mean and variance of the image patch. The mean reflects the average backscattering coefficient of the target, and the variance represents the degree to which all points in the image patch region deviate from the mean.

[0112] If the image patch size is Its mean and variance They are respectively:

[0113]

[0114]

[0115] in, For SAR power images in The value of the point.

[0116] Stability verification: Multiple sets of images of the site to be verified were selected from the same satellite, with the same ascent and descent orbits, the same incident angle, but different acquisition times. The root mean square error of the temporal backscattering coefficients was calculated. :

[0117]

[0118] Where L is the number of satellite images selected. Let be the average value of the backscattering coefficient within the target region at time i. Let i be the average value of i over a series of time periods.

[0119] After uniformity and stability tests, sites that meet the design requirements for satellite radiation indicators were selected as calibration sites and distribution targets for absolute radiation calibration of medium and high orbit SAR satellites.

[0120] S2.2: Imaging of distributed targets;

[0121] Image the same target area using both calibrated SAR satellites and medium-high orbit SAR satellites to be calibrated. The imaging requirements include both calibrated SAR satellites and medium-high orbit SAR satellites. The same target area should be imaged at least three times, with a time span not exceeding one week. The target area should be under similar meteorological conditions, and the same ascending / descending orbit direction and the same left and right side-view geometric settings should be used during imaging.

[0122] Step 3: Noise power elimination;

[0123] Medium- and high-orbit SAR satellites, such as LandScanner-4 01, produce a single-image swath of 3000 km, stitched together from 7 sub-images along the range direction. The incident angle varies by nearly 30° between the near and far ends of the full-orbit image, while a single sub-image has a swath width of 500 km and an incident angle variation of approximately 5°. To address the characteristic of large variations in noise power with the incident angle along the range direction under wide-swath imaging by medium- and high-orbit SAR satellites, a fitting relationship between noise power and incident angle is established to accurately eliminate noise power from the selected targets in step two.

[0124] S3.1: Noise data imaging;

[0125] The satellite is set to receive-only mode (transmitter off) to obtain image data. The image data has no ground object echo signal, and the power is calculated as noise power.

[0126] S3.2: Range-direction partitioning and angle of incidence calculation;

[0127] The SAR image is divided into sub-bands along the range direction based on the original sub-scene, with the width of each sub-band corresponding to the change in the incident angle. Let the i-th subband be ? Extract the average incident angle of each sub-band ;

[0128] S3.3: Subband Noise Power Calculation

[0129] Subbands of pure noise Calculate the mean noise power within the sub-band. :

[0130]

[0131] in, For children The number of pixels within, for Inner The noise power value of each pixel.

[0132] Establish a set of sample pairs of sub-band average incident angle and mean noise power:

[0133]

[0134] S3.4: Noise Power-Incident Angle Curve Fitting

[0135] To address the characteristics of wide-swath SAR satellites in medium and high orbits, such as large incident angle ranges and the correlation between noise power and incident angle, a nonlinear fitting function is used to fit the sample set. A global fit is performed, using a quadratic polynomial function as the fitting function:

[0136]

[0137] in, For any incident angle, Here, a, b, and c represent the noise power corresponding to the incident angle; a, b, and c are the fitting coefficients, calculated using the least squares method on the sample set. The solution is obtained.

[0138] S3.5: Noise power deduction;

[0139] Based on the location of the distributed target in the image x is the distance coordinate, which is substituted into the fitted continuous model. This yields the precise noise power value at the corresponding location. For the original echo power Noise power removal is performed to obtain the effective target echo power. :

[0140]

[0141] Subtracting the obtained noise power, the relationship between image power and backscattering coefficient is as follows:

[0142]

[0143]

[0144] Step 4: Image geometric consistency processing;

[0145] Based on the processing in step three, observation geometric correction and resolution adaptation processing are performed on the calibrated SAR images and medium-to-high orbit SAR images, mainly including:

[0146] S4.1: Resolution Adaptation Processing

[0147] High-resolution single-look complex (SLC) images from calibrated SAR satellites undergo multi-look processing at the slant range to ensure their equivalent resolution in the slant range domain matches that of the medium-to-high orbit SAR image to be calibrated. This eliminates contrast errors introduced by differences in pixel representation area. The specific steps are as follows:

[0148] (1) Determine the target resolution: Determine the nominal resolution of the medium-to-high orbit SAR image to be calibrated, and assume its azimuth resolution is . The slant range resolution is Simultaneously, the raw resolution of the calibrated reference SAR satellite SLC imagery is obtained. and .

[0149] (2) Calculate the required number of views: Multiview processing is achieved by incoherently averaging multiple single-view pixels (i.e., images from different parts of the Doppler spectrum). The required number of views is determined by the ratio of the target resolution to the original resolution.

[0150] Orientation and view number :

[0151]

[0152] Range of view :

[0153]

[0154] Total equivalent number of views :

[0155]

[0156] in, and Take the integer value that is closest to the target resolution.

[0157] (3) Perform multi-view averaging: Directly perform spatial averaging on the intensity map of the calibrated SLC image, dividing the image into size segments in the azimuth and range directions. The multi-view averaging formula for intensity images is as follows: (The formula is missing from the provided text.)

[0158]

[0159] in, It is the pixel index of the multi-view image. It is the pixel index within the corresponding window in the original SLC image.

[0160] After resolution adaptation, the calibrated SAR image and the uncalibrated high-orbit SAR image can be considered to have approximately the same ground resolution unit in the slant range domain. Therefore, they are converted to the ground range domain, i.e.:

[0161]

[0162]

[0163]

[0164] S4.2: Observation geometry normalization processing;

[0165] Because the selected target distribution area is located in a flat region, the difference in satellite incident angle between the calibrated satellite images and the high-orbit SAR satellite images to be calibrated is mainly affected by the difference in satellite incident angle. Step S4.2 aims to eliminate the influence of differences in observation geometry (incident angle, azimuth angle) between the reference satellite and the high-orbit SAR satellite to be calibrated on the comparison of the backscattering coefficients of ground objects. The main steps include:

[0166] (1) Determine the incident angles of the calibrated satellite A and the high-orbit SAR satellite B to be calibrated. , ;

[0167] (2) Normalize the backscattering coefficient values ​​observed by the calibrated reference satellite using a cosine model. Under the condition of the incident angle of the satellite to be calibrated :

[0168]

[0169] Among them, the index The backscattering coefficient can be determined experimentally based on the surface type of the calibration site, assuming that it is related to... It is directly proportional, so here we take m=1.

[0170] Step 5: Calculate the backscattering coefficient of the distributed target;

[0171] After processing in steps three and four, for SAR satellite images with known calibration constants, the backscattering coefficients of distributed targets are calculated based on the absolute calibration model formula and the relationship between image pixel power and backscattering coefficient. The main steps include:

[0172] (1) Determine the calibration constants of the calibrated SAR satellite ;

[0173] (2) For the calibrated SAR images after noise power removal and geometric consistency processing in steps three and four, select the target distribution of the image blocks determined in step two, and calculate the average energy value of all pixels in each image block. :

[0174]

[0175] in, Let be the average energy of the i-th pixel, and n be the number of pixels in the image block.

[0176] Calculate the backscattering coefficient of the distributed target :

[0177]

[0178] The backscattering coefficient measurement value converted to the perspective of a medium-to-high orbit satellite should be: :

[0179]

[0180] Step 6: Calculation of calibration constants for medium and high orbit SAR satellites;

[0181] After processing in steps three and four, based on the absolute calibration model formula and using the distributed target backscattering coefficient obtained in step four, the calibration constant of medium- and high-orbit SAR satellites is calculated, establishing the relationship between pixel power and backscattering coefficient of medium- and high-orbit SAR images. The main steps include:

[0182] (1) For the medium- and high-orbit SAR images with noise power removed in steps three and four, select the same image patch distribution targets as in step five, and register them with the image patches in the calibrated SAR images. Calculate the average energy value of all pixels within the distribution targets of each image patch. :

[0183]

[0184] in, is the average energy of the i-th pixel of a medium-to-high orbit SAR satellite.

[0185] (2) Based on the results obtained in step four Calculate the calibration constants for the medium-to-high orbit SAR satellite to be calibrated. :

[0186]

[0187] consider :

[0188]

[0189] The final calibration equation for medium- and high-orbit SAR satellites is:

[0190]

[0191] The calibration constants for medium- and high-orbit SAR satellites are obtained from the above formula, and the absolute radiometric calibration of medium- and high-orbit SAR satellites is completed.

[0192] like Figure 3 As shown, based on the same inventive concept, this application also provides a SAR satellite absolute radiometric calibration device 200, comprising:

[0193] The calibration field determination module 210 is used to determine the naturally distributed target as the calibration field;

[0194] The imaging control module 220 is used to image the calibration field using calibrated SAR satellites and medium-high orbit SAR satellites to be calibrated, respectively, and to acquire calibrated SAR images and medium-high orbit SAR images.

[0195] The noise power removal module 230 is used to establish a correspondence between noise power and incident angle based on the characteristics of noise power changing with the incident angle from the range direction during wide-swath imaging of medium- and high-orbit SAR satellites, and to remove noise power from calibrated SAR images and medium- and high-orbit SAR images based on the correspondence.

[0196] The geometric consistency processing module 240 is used to perform geometric consistency processing on the calibrated SAR image and the medium-high orbit SAR image after noise power removal.

[0197] The backscattering coefficient calculation module 250 is used to calculate the backscattering coefficient of the calibration field based on the calibrated SAR image after noise power removal and geometric consistency processing.

[0198] The calibration constant calculation module 260 is used to calculate the calibration constant of the medium- and high-orbit SAR satellite to be calibrated using the backscattering coefficient of the calibration field and the medium- and high-orbit SAR image after noise power removal and geometric consistency processing.

[0199] It is understood that the SAR satellite absolute radiometric calibration device 200 provided in this application embodiment can realize any one of the functions of the above-mentioned SAR satellite absolute radiometric calibration method. For the method embodiment section, please refer to the method embodiment section for the way each function is realized and the working principle. The device embodiment section will not repeat it here.

[0200] Figure 4 A structural block diagram of an electronic device according to an embodiment of the present invention is shown. Figure 4 As shown, the electronic device includes a memory 310 and a processor 320. The memory 310 stores a computer program that can run on the processor 320. When the processor 320 executes the computer program, it implements the SAR satellite absolute radiometric calibration method in the above embodiments. The number of memories 310 and processors 320 can be one or more.

[0201] The electronic device / terminal / server also includes:

[0202] The communication interface 330 is used to communicate with external devices and perform data exchange and transmission.

[0203] If the memory 310, processor 320, and communication interface 330 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0204] Optionally, in a specific implementation, if the memory 310, processor 320 and communication interface 330 are integrated on a single chip, the memory 310, processor 320 and communication interface 330 can communicate with each other through an internal interface.

[0205] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method provided in this application.

[0206] This application also provides a chip, which includes a processor for calling and executing instructions stored in a memory, causing a communication device on which the chip is installed to perform the method provided in this application.

[0207] This application also provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, output interface, processor, and memory are connected through an internal connection path. The processor is used to execute code in the memory. When the code is executed, the processor is used to execute the method provided in this application.

[0208] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting the Advanced Reduced Instruction Set Computing (RISC) machine (ARM) architecture.

[0209] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0210] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0211] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0212] Furthermore, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0213] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for absolute radiometric calibration of SAR satellites, characterized in that, The method includes: Determine the naturally distributed target as the calibration field; The calibration field is imaged using calibrated SAR satellites and medium-high orbit SAR satellites to be calibrated, respectively, to obtain calibrated SAR images and medium-high orbit SAR images; Based on the characteristic that noise power changes with the incident angle in the range direction during wide-swath imaging of the medium- and high-orbit SAR satellite, a correspondence between noise power and incident angle is established, and noise power is removed from the calibrated SAR image and the medium- and high-orbit SAR image respectively based on the correspondence. Geometric consistency processing is performed on the calibrated SAR image and the medium-high orbit SAR image after noise power removal; Based on the calibrated SAR image after noise power removal and geometric consistency processing, the backscattering coefficient of the calibration field is calculated; The calibration constants of the medium-high orbit SAR satellite to be calibrated are calculated using the backscattering coefficients of the calibration field and the medium-high orbit SAR image after noise power removal and geometric consistency processing.

2. The SAR satellite absolute radiometric calibration method according to claim 1, characterized in that, Before using the naturally distributed target as the calibration field, the method further includes: The candidate region is subjected to a uniformity test; wherein the uniformity test is used to evaluate the spatial uniformity of the backscattering characteristics of the candidate region in order to verify the spatial uniformity of the candidate region as a distribution target. And / or, perform a temporal stability test on the candidate region; wherein the temporal stability test is used to evaluate the stability of the backscattering coefficient of the candidate region over time, so as to verify the temporal stability of the candidate region as a distribution target.

3. The SAR satellite absolute radiometric calibration method according to claim 2, characterized in that, The uniformity test includes: selecting image blocks based on the SAR power image of the candidate region, calculating the mean and variance of the pixel power within the image block, and determining whether the candidate region meets the spatial uniformity requirements as a distribution target based on the mean and variance; wherein, the mean is used to characterize the average backscattering coefficient of the candidate region; and the variance is used to characterize the degree to which the power value of each pixel within the image block deviates from the mean. The temporal stability test includes: acquiring multiple SAR images of the candidate region, calculating the root mean square error of the temporal backscattering coefficient, and determining whether the candidate region meets the temporal stability requirements for being a distributed target based on the root mean square error; wherein, the multiple SAR images meet the conditions of the same satellite, the same ascent and descent orbit, the same incident angle, and different acquisition times; the root mean square error is used to characterize the stability of the backscattering coefficient of the candidate region over time.

4. The SAR satellite absolute radiometric calibration method according to claim 1, characterized in that, Establishing the correspondence between noise power and incident angle includes: The medium-high orbit SAR image is divided into multiple sub-bands along the range direction. The incident angle variation of each sub-band does not exceed a preset angle threshold, and the average incident angle of each sub-band is extracted. Obtain the noise power of each sub-band and establish a sample pair set of the average incident angle and the noise power; Based on the sample set, curve fitting is performed to establish a continuous fitting relationship between noise power and incident angle, and the correspondence between noise power and incident angle is obtained.

5. The SAR satellite absolute radiometric calibration method according to claim 4, characterized in that, The acquisition of noise power for each sub-band includes: The medium-high orbit SAR satellite is set to receive-only mode; wherein, the receive-only mode is a working mode in which the transmitter is turned off and only echo signals are received, so as to obtain pure noise images without ground object echo signals; The noise power of each sub-band is calculated based on the pure noise image.

6. The SAR satellite absolute radiometric calibration method according to claim 1, characterized in that, The geometric consistency processing of the calibrated SAR image and the medium-to-high orbit SAR image after noise power removal includes: The calibrated SAR image is subjected to resolution adaptation processing to make the resolution of the calibrated SAR image consistent with the resolution of the medium-high orbit SAR image, so as to eliminate the contrast error introduced by the different pixel representation area. The calibrated SAR image and the medium-high orbit SAR image, after resolution adaptation processing, are subjected to observation geometry normalization processing to eliminate or reduce the backscattering coefficient deviation caused by the difference in incident angle.

7. The SAR satellite absolute radiometric calibration method according to claim 6, characterized in that, The resolution adaptation process includes: performing multi-view averaging on the single-view complex image of the calibrated SAR image in the azimuth and range directions, so that the equivalent resolution of the calibrated SAR image in the slant range domain is consistent with the resolution of the medium-high orbit SAR image. The observation geometry normalization process includes: based on the first incident angle of the calibrated SAR satellite and the second incident angle of the medium-high orbit SAR satellite to be calibrated, normalizing the backscattering coefficient of the calibrated SAR image observation to the observation geometry conditions of the medium-high orbit SAR satellite to be calibrated, so as to eliminate the backscattering coefficient deviation caused by the difference in incident angle.

8. A SAR satellite absolute radiometric calibration device, characterized in that, include: The calibration field determination module is used to determine the naturally distributed target as the calibration field; The imaging control module is used to image the calibration field using calibrated SAR satellites and medium-high orbit SAR satellites to be calibrated, respectively, to acquire calibrated SAR images and medium-high orbit SAR images. The noise power removal module is used to establish a correspondence between noise power and incident angle based on the characteristics of noise power variation with distance-to-incident angle during wide-swath imaging of the medium-high orbit SAR satellite, and to remove noise power from the calibrated SAR image and the medium-high orbit SAR image based on the correspondence. A geometric consistency processing module is used to perform geometric consistency processing on the calibrated SAR image and the medium-high orbit SAR image after noise power removal; The backscattering coefficient calculation module is used to calculate the backscattering coefficient of the calibration field based on the calibrated SAR image after noise power removal and geometric consistency processing. The calibration constant calculation module is used to calculate the calibration constant of the medium-high orbit SAR satellite to be calibrated using the backscattering coefficient of the calibration field and the medium-high orbit SAR image after noise power removal and geometric consistency processing.

9. An electronic device, characterized in that, include: A processor and a memory, wherein instructions are stored in the memory and loaded and executed by the processor to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of claims 1-7.

Images