A method for cross-calibration of synthetic aperture radar radiation based on weighted least squares
By using uncertainty analysis and weighted least squares model in synthetic aperture radar (SAR) radiation cross calibration, the problem of data affecting calibration accuracy on quality differences is solved, and a higher precision calibration result is achieved.
Patent Information
- Application Number
- CN202411002647.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2025-05-02
- Estimated Expiration
- 2044-07-25
AI Technical Summary
The existing synthetic aperture radar (SAR) radiation cross calibration method ignores the quality differences in data pairs, resulting in inaccurate results of calibration coefficients, making it difficult to eliminate the impact of scattering stability, spatiotemporal consistency and scene uniformity on calibration coefficients.
The uncertainty analysis method is used to calculate the uncertainty of the cross-calibrated data pair, and the weight coefficients of each pair are assigned to each pair of data pairs. The calibration coefficients of the pending satellites are obtained through the weighted least squares model, thereby improving the radiation cross-calibration accuracy.
Through the use of the weighted least squares model, the quality differences of data pairs can be considered more accurately, the accuracy of cross-calibration and the calibration error can be reduced.
Smart Images
Figure CN119044906B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of radar calibration, and in particular to a synthetic aperture radar radiation cross calibration method based on weighted least squares. Background Art
[0002] Synthetic Aperture Radar (SAR) absolute radiometric calibration is a process of establishing an accurate relationship between the end-to-end performance of the SAR system to achieve an accurate correspondence between the digital quantization value of the SAR image and the backscatter coefficient of the target. This process aims to infer the physical characteristics of the target, eliminate the influence of different radiation errors, make SAR images acquired at different times, with different loads or in different regions comparable, and provide accurate radiation information and basic data for applications such as environmental monitoring, resource management and disaster assessment to meet the needs of SAR quantitative remote sensing.
[0003] The methods of absolute radiometric calibration can be divided into the calibration method based on manual calibrators and the radiation cross calibration method. The calibration method based on manual calibrators is to deploy a certain number of calibrators at the calibration site, and calculate the absolute radiometric calibration coefficient based on the digital quantization value of the image and the radar cross-section area (RCS) of the calibrator. The radiation cross calibration method is to calibrate the satellite data to be calibrated by observing the same target or scene, using the satellite data that has been calibrated with high precision as a reference or standard. Compared with the calibration method based on manual calibrators, the radiation cross calibration method does not require manual deployment of calibrators and field measurement data, saving a lot of manpower, equipment and funds. At the same time, the radiation cross calibration method can also calibrate historical data, breaking through the limitations of ground synchronous measurement. For this reason, this method has advantages in calibration frequency and timeliness.
[0004] The radiation cross calibration method includes the steps of extracting stable scenes, correcting the incident angle of radar observation, and fitting calibration coefficients based on ordinary least squares method, among which the scene extraction and the correction of the observed incident angle are both preprocessing processes before obtaining the calibration coefficients. The calibration coefficients of the satellite to be calibrated can be directly obtained by establishing a calibration model based on ordinary least squares method, thereby completing the radiation cross calibration. However, the existing method of establishing a calibration model based on ordinary least squares method still has the following technical defects: the cross data pairs in the preprocessing process are necessary data for obtaining the calibration coefficients, and their quality accuracy will affect the reliability and authenticity of the radiation calibration. The existing radiation cross calibration method ignores the quality differences of each data pair, and it is difficult to eliminate the influence of factors such as scattering stability, spatial matching, and data block uniformity on the calibration coefficients, resulting in inaccurate calibration coefficient results.
[0005] Therefore, how to provide a synthetic aperture radar radiation cross calibration method based on weighted least squares that can improve calibration accuracy is a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention
[0006] In view of the above research status, the present invention provides a synthetic aperture radar radiation cross-calibration method based on weighted least squares. In view of the quality differences of each data pair caused by the influence of radiation stability, spatiotemporal consistency and scene uniformity in the cross-calibration process, an uncertainty analysis method is used to calculate the uncertainty of the corresponding influencing factors of the data pair, and then a weight coefficient is assigned to each pair of cross data pairs. Finally, the calibration coefficient of the satellite to be calibrated is obtained through the weighted least squares model, thereby improving the accuracy of radiation cross-calibration.
[0007] The present invention provides a synthetic aperture radar radiation cross calibration method based on weighted least squares, comprising the following steps:
[0008] Extraction of cross-calibration data pairs: obtain two sets of image data of the uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval, namely, the calibrated SAR image and the SAR image to be calibrated. According to the selected backscatter coefficient description feature extraction method, extract the backscatter coefficient of the calibrated SAR in the uniform area and the amplitude value of the SAR to be calibrated in the uniform area as the cross-calibration data pair;
[0009] Calculation of weight coefficients: acquiring historical image data of the calibrated SAR and the SAR to be calibrated of the same target scene uniform area within a specified time interval, extracting the backscattering coefficient of the calibrated SAR historical image data and the amplitude value of the SAR historical image data to be calibrated according to the historical image data, as a historical cross-calibration data pair; calculating the overall uncertainty of the historical cross-calibration data pair according to the calibration error influencing factors; calculating the weight coefficient of the cross-calibration data pair according to the overall uncertainty;
[0010] Generation of calibration coefficients: constructing a calibration coefficient model based on a weighted least squares model, and substituting the cross-calibration data pair and the weight coefficient into the calibration coefficient model to calculate and obtain the calibration coefficient.
[0011] Preferably, in the step of extracting the cross-calibration data pair, the feature extraction method is described according to the selected backscatter coefficient, and extracting the backscatter coefficient of the calibrated SAR in the cross-target scene area includes the following steps:
[0012] S11: Acquire a calibrated SAR image time series data set under the same orbit, the same incident angle, and VV polarization mode;
[0013] S12: After preprocessing the SAR image time series data group, select the same uniform scene in each SAR image data for slicing processing;
[0014] S13: calculating the backscatter coefficient of each slice according to the selected backscatter coefficient description feature extraction method;
[0015] S14: Calculate the error of the backscatter coefficient of the slice in the SAR image time series data group on the corresponding time series; select the backscatter coefficient with the smallest time series error to construct the cross-calibration data pair.
[0016] Preferably, the backscatter coefficient description feature extraction method includes a median method and an average method.
[0017] Preferably, in the step of extracting the cross-calibration data pair, the same uniform regions in the two groups of image data include: uniform regions selected from a plurality of same scene categories in each group of image data.
[0018] Preferably, the method further comprises the step of correcting the incident angle:
[0019] A data pair scene evaluation threshold is determined by using historical image data of a uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval. The data pair scene evaluation threshold is used to perform incident angle correction on the calibrated SAR to obtain a corrected backscatter coefficient of the calibrated SAR. Combined with the amplitude value of the SAR image data to be calibrated, a corrected cross-calibration data pair is constructed as the cross-calibration data pair.
[0020] Preferably, the step of determining the data-to-scene evaluation threshold using the historical image data of the calibrated SAR and the SAR to be calibrated in the same target scene uniform area within a specified time interval includes:
[0021] S21: extracting the calibrated SAR amplitude information DN of the historical image data in the uniform area respectively cal , SAR amplitude information to be calibrated DN uncal , calibrated SAR backscatter coefficient information σ cal and the SAR backscatter coefficient information to be calibrated σ uncal ;
[0022] S22: according to the information extracted in S21, the backscatter coefficient difference Δσ and the amplitude difference ΔDN of the calibrated SAR and the SAR to be calibrated in the uniform area of the same target scene are obtained, and an empirical linear relationship between the backscatter coefficient difference Δσ and the amplitude difference ΔDN is constructed by fitting;
[0023] S23: Determine an amplitude difference value by substituting a specified backscatter coefficient difference value between the historical image data of the calibrated SAR in the uniform area of the same target scene and the historical image data of the SAR to be calibrated in the uniform area of the same target scene into the empirical linear relationship, and use the amplitude difference value as a data pair scene evaluation threshold; if the threshold is exceeded, perform incident angle correction compensation on the calibrated SAR, and use the corrected backscatter coefficient as the backscatter coefficient of the calibrated SAR in the uniform area in the cross-calibration data pair.
[0024] Preferably, the S23 uses an exponential cosine model to perform incident angle correction processing on the SAR to be calibrated.
[0025] Preferably, the overall uncertainty includes the uncertainty of the cross-calibration data pair caused by multiple calibration error influencing factors, and the overall uncertainty is obtained by using a square sum root calculation.
[0026] Preferably, the uncertainty of the cross-calibration data pair caused by multiple calibration error influencing factors includes any combination of the following:
[0027] Scattering stability uncertainty: obtain the time series of the historical image data of the calibrated SAR for the same cross data pair scene, calculate the standard deviation of the backscattering coefficient in the scene over time, and use it as the uncertainty of scattering stability;
[0028] Spatial matching uncertainty: The geographical location of the imaging scene of the SAR to be calibrated is taken as the standard value, and the longitude and latitude of the pixels are used to calculate the difference in longitude and latitude of the pixels of the calibrated SAR in the same scene, which is converted into the actual distance deviation as the uncertainty of spatial matching;
[0029] Data block uniformity uncertainty: Based on the changes in the surface scattering characteristics of the corresponding scene in the spatial latitude of the historical cross-calibration data, the ratio of the standard deviation of the surface scattering characteristic data of the target scene to the mean, that is, the coefficient of variation, is obtained, which is the uncertainty of the data block uniformity.
[0030] Preferably, the calibration coefficient model is:
[0031]
[0032] Among them, α is the calibration coefficient; σ correct calibrating the backscatter coefficient of the SAR for the cross-calibration data; is the square of the amplitude value of the cross-calibration data to the SAR to be calibrated, that is, the intensity value; W is the weight coefficient matrix corresponding to the cross-calibration data.
[0033] The synthetic aperture radar radiation cross calibration method based on weighted least squares proposed in the present invention has the following beneficial effects compared with the prior art:
[0034] The method for obtaining the calibration coefficient of the SAR satellite to be calibrated by the ordinary least squares model in the present invention ignores the quality differences of each data pair caused by radiation stability, spatiotemporal consistency and scene uniformity in the cross-calibration process, adopts the uncertainty analysis method to calculate the corresponding uncertainty of the data pair and assigns a weight coefficient, and then obtains the calibration coefficient of the satellite to be calibrated by the weighted least squares fitting method, which can further improve the cross-calibration accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without creative work.
[0036] Figure 1 A flow chart of a synthetic aperture radar radiation cross calibration method based on weighted least squares provided in an embodiment of the present invention;
[0037] Figure 2 A flow chart of SAR incident angle correction provided by an embodiment of the present invention;
[0038] FIG3 is a calibration coefficient fitting diagram provided by an embodiment of the present invention;
[0039] Figure 4 A comparison chart of the calibration accuracy of the method of the present invention and the traditional method in different scenarios provided by the embodiment of the present invention. DETAILED DESCRIPTION
[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0041] The present invention provides a synthetic aperture radar radiation cross calibration method based on weighted least squares, such as Figure 1 As shown, the following steps are included:
[0042] Extraction of cross-calibration data pairs: obtain two sets of image data of the uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval, namely, the calibrated SAR image and the SAR image to be calibrated. According to the selected backscatter coefficient description feature extraction method, extract the backscatter coefficient of the calibrated SAR in the uniform area and the amplitude value of the SAR to be calibrated in the uniform area as the cross-calibration data pair;
[0043] Calculation of weight coefficients: Obtain historical image data of the uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval, extract the backscattering coefficient of the calibrated SAR historical image data and the amplitude value of the SAR historical image data to be calibrated based on the historical image data, as the historical cross-calibration data pair; calculate the overall uncertainty of the historical cross-calibration data pair based on the calibration error influencing factors; calculate the weight coefficient of the cross-calibration data pair based on the overall uncertainty;
[0044] Generation of calibration coefficients: A calibration coefficient model is constructed based on a weighted least squares model, and the calibration coefficients are calculated by substituting the cross-calibration data pairs and weight coefficients into the calibration coefficient model.
[0045] In one embodiment, in the step of extracting the cross-calibration data pair, a feature extraction method is described according to the selected backscatter coefficient, and extracting the backscatter coefficient of the calibrated SAR in the cross-target scene area includes the following steps:
[0046] S11: Acquire a calibrated SAR image time series data set under the same orbit, the same incident angle, and VV polarization mode;
[0047] S12: After preprocessing the SAR image time series data group, select the same uniform scene in each SAR image data for slicing processing;
[0048] S13: calculating the backscatter coefficient of each slice according to the selected backscatter coefficient description feature extraction method;
[0049] S14: Calculate the error of the backscatter coefficient of the slice in the SAR image time series data group on the corresponding time series; select the backscatter coefficient with the smallest time series error to construct a cross-calibration data pair.
[0050] In one embodiment, the backscatter coefficient description feature extraction method includes a median method and an average method.
[0051] In one embodiment, the same uniform regions in the two sets of image data in the step of extracting the cross-calibrated data pair include: uniform regions selected from multiple same scene categories in each set of image data.
[0052] The extraction step of the cross-calibration data pair is specifically performed:
[0053] The image data used in this implementation is the Sentinel-1 series satellite images. A group of Sentinel-1A image data with the same orbit, the same incident angle, and VV polarization mode was obtained. The shooting interval of each image was about 3 months, and there were 5 images in total. The specific selected data parameters are shown in Table 1. Subsequently, the 5 image data were preprocessed by orbit correction, thermal noise removal, radiation calibration, coherent speckle filtering, geocoding, etc. using SNAP software. Then, the bare land scenes and urban scenes in the same longitude and latitude areas in the group of images were sliced, and the median and average values of the slices were calculated by the median method and the average method, and the stability of the backscatter coefficient values obtained by these two methods in the time series was compared. If the time series standard deviation of the backscatter coefficient is smaller, its time series stability is higher, and the corresponding value can be regarded as the descriptive feature of the backscatter coefficient under the target scene.
[0054] Table 1 Specific information of image dataset
[0055]
[0056]
[0057] The cross data pairs of urban area and bare land are extracted respectively by describing the feature extraction method through the determined backscatter coefficient:
[0058] First, two image data of the same target scene are obtained from the satellites of the cross-calibration, namely the calibrated Sentinel-1A image and the Sentinel-1B image data to be calibrated. Both are images in interferometric wide-band (IW) imaging mode and VV polarization mode, and the difference in incident angle between the two is 11°. In the case of radiometric cross-calibration, in order to avoid the influence of special terrain and rainfall changes, the actual data needs to be obtained in a similar time. The specific information of the data set is shown in Table 2.
[0059] Table 2 Specific information of cross-data pair image dataset
[0060] data Scene Type Polarization Get time Angle of incidence satellite Data to be calibrated Urban area + bare land VV 2022 / 02 / 12 44.6° Sentinel-1B Calibrated data Urban area + bare land VV 2022 / 02 / 11 33.1° Sentinel-1A
[0061] The reference target is roughly registered through the landmark objects, and the approximate position of the reference target in the two scene data is preliminarily located to complete the data slicing. In the registered target scene, the urban scene and the bare land scene are selected and sub-sliced respectively. The size of each sub-slice is 20 pixels × 20 pixels, and each scene has 20 slices. Then, the amplitude value (DN) and the backscatter coefficient (σ) value of each scene are extracted using the determined backscatter coefficient description feature as the cross data pair of the subsequent regression calibration coefficient.
[0062] In one embodiment, the step of correcting the incident angle is also included:
[0063] The historical image data of the uniform area of the target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval are used to determine the data pair scene evaluation threshold. The data pair scene evaluation threshold is used to correct the incident angle of the calibrated SAR to obtain the corrected backscatter coefficient of the calibrated SAR. Combined with the amplitude value of the SAR image data to be calibrated, a corrected cross-calibration data pair is constructed as the cross-calibration data pair.
[0064] In one embodiment, the step of determining the data-to-scene evaluation threshold using historical image data of a uniform area of a target scene in a specified time interval from a calibrated SAR and a SAR to be calibrated includes:
[0065] S21: Extract the calibrated SAR amplitude information DN of the historical image data in the uniform area respectively cal , SAR amplitude information to be calibrated DN uncal , calibrated SAR backscatter coefficient information σ cal and the SAR backscatter coefficient information to be calibrated σ uncal ;
[0066] S22: according to the information extracted in S21, the backscatter coefficient difference Δσ and the amplitude difference ΔDN of the calibrated SAR and the SAR to be calibrated in the uniform area of the same target scene are obtained, and an empirical linear relationship between the backscatter coefficient difference Δσ and the amplitude difference ΔDN is constructed by fitting;
[0067] S23: Determine an amplitude difference value by substituting a specified backscatter coefficient difference value between the historical image data of the calibrated SAR in the uniform area of the same target scene and the historical image data of the SAR to be calibrated in the uniform area of the same target scene into an empirical linear relationship, and use the amplitude difference value as a data pair scene evaluation threshold; if the threshold is exceeded, perform incident angle correction compensation on the calibrated SAR, and use the corrected backscatter coefficient as the backscatter coefficient of the calibrated SAR in the uniform area in the cross-calibration data pair.
[0068] When executing:
[0069] The historical image data of the calibrated and uncalibrated SAR in the target area are obtained, and their specific information is shown in Table 3.
[0070] Table 3 Specific parameters of SAR historical image data that have been calibrated and those to be calibrated
[0071]
[0072]
[0073] They have different incident angle information in the same target scene, and then extract their amplitude information DN on the uniform scene respectively. cal DN uncal and backscatter coefficient information σ cal , σ uncal . Then, the backscatter coefficient difference Δσ and amplitude difference ΔDN of the uniform object type in the same target scene are obtained, and their units are both dB. Construct an empirical linear relationship between Δσ and ΔDN, that is, ΔDN = aΔσ + b. Determine the intensity value difference value based on the backscatter coefficient difference value (0.3dB), and then use the intensity value difference value as the evaluation threshold before the subsequent image needs to be corrected for the incident angle.
[0074] According to historical data analysis, when the difference in backscatter coefficient caused by the incident angle change is within 0.3dB, no calibration compensation is required. If the DN difference before and after the incident angle correction of the target scene exceeds (0.3·a+b), the incident angle needs to be corrected, otherwise it is not necessary to avoid over-correction of the data.
[0075] In one embodiment, for the cross data pairs that need incident angle correction, S23 uses an exponential cosine model to perform correction processing. The correction index n in different scenarios can be obtained by analyzing the historical image data of the target area.
[0076] Through historical data analysis, it is found that when n = 5, cosine correction can effectively solve the scattering difference caused by the incident angle of the bare land scene in the target area. When n = 2, cosine correction can also effectively solve the scattering difference caused by the incident angle of the urban scene in the target area.
[0077] In one embodiment, the overall uncertainty includes the uncertainty of the cross-calibration data pair caused by multiple calibration error influencing factors. When the error influencing factors are independent of each other, the overall uncertainty is calculated using the root of the sum of squares.
[0078] In one embodiment, the quality differences of the cross-calibration data pairs during the screening process mainly come from the following points: instability of data block scattering, errors caused by spatial position matching, and lack of uniformity of data blocks. Therefore, the uncertainty of the cross-calibration data pairs caused by multiple calibration error influencing factors includes any combination of the following:
[0079] Scattering stability uncertainty: Obtain the time series of the historical image data of the calibrated SAR for the same cross-data pair scene, and analyze the fluctuation of the backscattering coefficient of each 20 data pairs in the bare land scene and the urban scene over time. Define the standard deviation of the scattering stability of each cross-data pair as the uncertainty of its scattering stability;
[0080] Spatial matching uncertainty: Factors such as geometric correction may cause pixel point deviation, which is why a site with as uniform scattering as possible is selected during calibration. The geographical location of the imaging scene of the SAR to be calibrated is used as the standard value, and the longitude and latitude of the pixel points are used for analysis. The longitude and latitude differences of the pixels of the calibrated SAR in the same scene are analyzed and then converted into actual geographic distance deviations. The relative distance deviation between the actual geographic space of the imaging scene of the SAR data to be calibrated and the actual geographic space of the imaging scene of the calibrated SAR data is defined as the uncertainty of spatial matching of cross-data pairs;
[0081] Data block uniformity uncertainty: Analyze the changes in the surface scattering characteristics of the scene corresponding to the cross data pair in the spatial latitude, and obtain the spatial discreteness of the target scene. Define the ratio of the standard deviation of the historical scattering data of the data pair to the average value of the historical scattering data (i.e., coefficient of variation, CV) as the uncertainty of the uniformity of the cross data pair.
[0082] The overall uncertainty is calculated using the root sum of squares:
[0083]
[0084] In the formula, Represents the uncertainty of the above three items.
[0085] In one embodiment, the weight coefficient corresponding to the cross-calibration data pair is generally expressed by the inverse of the square of its uncertainty. If the uncertainty corresponding to the nth group of cross-calibration data pairs is u n (σ 0 ), the corresponding weight is There are 40 groups of cross data pairs, and their weights can be expressed by the weight coefficient matrix W:
[0086]
[0087] In one embodiment, based on the data pairs of urban and bare land scenes, the calibration coefficients of the Sentinel-1B data to be calibrated are obtained through the calibrated Sentinel-1A reference data, where the urban data provides high backscatter coefficient values and the bare land data provides low backscatter coefficient values. The weighted least squares model (WLS) is used to model it, and the calibration coefficient model is:
[0088]
[0089] Among them, α is the calibration coefficient; σ correct The backscatter coefficient of the calibrated SAR is centered for the cross-calibration data; is the square of the amplitude value of the cross-calibration data to be calibrated SAR, that is, the intensity value; W is the weight coefficient matrix corresponding to the cross-calibration data.
[0090] As shown in FIG3 , FIG3( a ) is the fitting result of the calibration coefficient by the ordinary least squares regression method, and FIG3( b ) is the fitting result of the calibration coefficient by the weighted least squares regression method according to an embodiment of the present invention.
[0091] The calibration formula obtained using the ordinary least squares regression method is as follows:
[0092]
[0093] The calibration formula obtained using the weighted least squares regression method is as follows:
[0094]
[0095] It can be seen that the calibration coefficient obtained by the ordinary least squares regression method is 3.1651×10 -6 The calibration coefficient obtained by the weighted least squares regression method is 3.0534×10 -6 , the difference in weight coefficients will provide a more accurate fit in some areas.
[0096] Depend on Figure 4 It can be seen that from top to bottom, the comparison effect diagrams of the backscattering coefficients obtained by the method of the present invention and the traditional method are shown in the three verification areas of urban area, farmland and bare land. The left side is the effect diagram of the traditional method, and the right side is the effect diagram of the method of the present invention. The backscattering coefficient data obtained by the method of the present invention are closer to the central axis as a whole, which is more obvious in the bare land area. From the quantitative analysis, the calibration root mean square error (RMSE) and relative error (ε) of the method of the present invention are generally lower than those of the traditional method. The relative error of the calibration decreased by an average of 27.7%, and the root mean square error of the calibration decreased by an average of 24.1%. The results show that the method of the present invention can further reduce the cross-calibration error and improve its calibration accuracy.
[0097] The above is a detailed introduction to a synthetic aperture radar radiation cross-calibration method based on weighted least squares provided by the present invention. In this article, specific examples are used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as limiting the present invention.
[0098] In this article, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
Claims
1. A synthetic aperture radar radiation cross calibration method based on weighted least squares, characterized in that: The steps include: Extraction of cross-calibration data pairs: obtain two sets of image data of the uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval, namely, the calibrated SAR image and the SAR image to be calibrated. According to the selected backscatter coefficient description feature extraction method, extract the backscatter coefficient of the calibrated SAR in the uniform area and the amplitude value of the SAR to be calibrated in the uniform area as the cross-calibration data pair; Calculation of weight coefficients: acquiring historical image data of the calibrated SAR and the SAR to be calibrated of the same target scene uniform area within a specified time interval, extracting the backscattering coefficient of the calibrated SAR historical image data and the amplitude value of the SAR historical image data to be calibrated according to the historical image data, as a historical cross-calibration data pair; calculating the overall uncertainty of the historical cross-calibration data pair according to the calibration error influencing factors; calculating the weight coefficient of the cross-calibration data pair according to the overall uncertainty; Generation of calibration coefficients: constructing a calibration coefficient model based on a weighted least squares model, and substituting the cross-calibration data pair and the weight coefficient into the calibration coefficient model to calculate and obtain the calibration coefficient.
2. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: In the step of extracting the cross-calibration data pair, a feature extraction method is described according to the selected backscatter coefficient, and extracting the backscatter coefficient of the calibrated SAR in the cross-target scene area includes the following steps: S11: Acquire a calibrated SAR image time series data set under the same orbit, the same incident angle, and VV polarization mode; S12: After preprocessing the SAR image time series data group, select the same uniform scene in each SAR image data for slicing processing; S13: calculating the backscatter coefficient of each slice according to the selected backscatter coefficient description feature extraction method; S14: calculating the error of the backscatter coefficient of the slice in the SAR image time series data group on the corresponding time series; The backscatter coefficient with the smallest time series error is selected to construct the cross-calibration data pair.
3. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: The backscatter coefficient description feature extraction method includes a median method and an average method.
4. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: The same uniform regions in the two groups of image data in the step of extracting the cross-calibration data pair include: uniform regions selected from multiple same scene categories in each group of image data.
5. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: It also includes the steps of correction of the angle of incidence: A data pair scene evaluation threshold is determined by using historical image data of a uniform area of the same target scene of the calibrated SAR and the SAR to be calibrated within a specified time interval. The data pair scene evaluation threshold is used to perform incident angle correction on the calibrated SAR to obtain a corrected backscatter coefficient of the calibrated SAR. Combined with the amplitude value of the SAR image data to be calibrated, a corrected cross-calibration data pair is constructed as the cross-calibration data pair.
6. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 5, characterized in that: The step of determining the data-to-scene evaluation threshold using the historical image data of the calibrated SAR and the SAR to be calibrated in the same target scene uniform area within a specified time interval includes: S21: extracting the calibrated SAR amplitude information DN of the historical image data in the uniform area respectively cal , SAR amplitude information to be calibrated DN uncal , calibrated SAR backscatter coefficient information σ cal and the SAR backscatter coefficient information σ to be calibrated uncal ; S22: according to the information extracted in S21, the backscatter coefficient difference Δσ and the amplitude difference ΔDN of the calibrated SAR and the SAR to be calibrated in the uniform area of the same target scene are obtained, and an empirical linear relationship between the backscatter coefficient difference Δσ and the amplitude difference ΔDN is constructed by fitting; S23: Determine an amplitude difference value by substituting a specified backscatter coefficient difference value between the historical image data of the calibrated SAR in the uniform area of the same target scene and the historical image data of the SAR to be calibrated in the uniform area of the same target scene into the empirical linear relationship, and use the amplitude difference value as a data pair scene evaluation threshold; if the threshold is exceeded, perform incident angle correction compensation on the calibrated SAR, and use the corrected backscatter coefficient as the backscatter coefficient of the calibrated SAR in the uniform area in the cross-calibration data pair.
7. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 6, characterized in that: The S23 uses an exponential cosine model to perform incident angle correction processing on the SAR to be calibrated.
8. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: The overall uncertainty includes the uncertainty of the cross-calibration data pair caused by multiple calibration error influencing factors, and the overall uncertainty is obtained by using the root formula of the sum of squares to calculate the uncertainty.
9. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 8, characterized in that: The uncertainty of the cross-calibration data pair caused by multiple calibration error factors includes any combination of the following: Scattering stability uncertainty: obtain the time series of the historical image data of the calibrated SAR for the same cross data pair scene, calculate the standard deviation of the backscattering coefficient in the scene over time, and use it as the uncertainty of scattering stability; Spatial matching uncertainty: The geographical location of the imaging scene of the SAR to be calibrated is taken as the standard value, and the longitude and latitude of the pixels are used to calculate the difference in longitude and latitude of the pixels of the calibrated SAR in the same scene, which is converted into the actual distance deviation as the uncertainty of spatial matching; Data block uniformity uncertainty: Based on the changes in the surface scattering characteristics of the corresponding scene in the spatial latitude of the historical cross-calibration data, the ratio of the standard deviation of the surface scattering characteristic data of the target scene to the mean, that is, the coefficient of variation, is obtained, which is the uncertainty of the data block uniformity.
10. The method for cross-calibration of synthetic aperture radar radiation based on weighted least squares according to claim 1, characterized in that: The calibration coefficient model is: Among them, α is the calibration coefficient; σ correct calibrating the backscatter coefficient of the SAR for the cross-calibration data; is the square of the amplitude value of the cross-calibration data to the SAR to be calibrated, that is, the intensity value; W is the weight coefficient matrix corresponding to the cross-calibration data.