A co-polarization channel imbalance calibration method based on overall coupling components

Abstract

The present invention discloses a co-polarization channel imbalance calibration method based on overall coupling components, the method comprising the following steps: step 1: construct an uncalibrated covariance matrix through an uncalibrated backscattering matrix and divide the image into blocks along the azimuth and distance directions; step 2: solve the cross-polarization channel imbalance and crosstalk of the full polarization SAR based on the uncalibrated covariance matrix, and perform correction; step 3: solve OA by using DEM through threshold setting and select a region that meets the requirements; step 4: construct an equation based on the overall coupling component of the target direction deviation according to the selected region to determine the preliminary co-polarization channel imbalance result; step 5: correct the polarization SAR image according to the different results obtained in step 4, compare the polarization OA obtained by different results with DEMOA, and select the final co-polarization channel imbalance result. The present invention can reduce the error caused by selecting a region using uncalibrated parameters.

Description

Technical Field

[0001] The invention belongs to the field of radar detection, and in particular relates to a co-polarization channel unbalance calibration method based on overall coupling components. Background Art

[0002] Polarimetric synthetic aperture radar (SAR) can acquire, analyze and process polarization information carried by electromagnetic waves. By emitting and receiving electromagnetic waves in different bands that are perpendicular to each other, it can reflect the scattering characteristics of ground objects as completely as possible, and has priority for use in multiple scenarios and situations. There are two ways to represent polarization information: one is the backscattering matrix, and the other is the scattering power matrix of the target, also known as the Kennaugh matrix. Compared with the backscattering matrix, the Kennaugh matrix gives a clearer picture of the scattering characteristics of ground objects based on second-order statistics, but it is often overlooked in practical applications and has great mining value. The Kennaugh matrix can characterize symmetry, irregularity, depolarization components and coupling components of symmetrical and asymmetric targets. Among them, the overall coupling component of the target direction deviation is used in the Kennaugh matrix. It is related to the orientation angle (OA). For the area with an azimuth angle of 0, that is, the area where the radar line of sight of the target is parallel to the normal direction of the ground object in the target symmetry plane, the overall coupling component is also 0. The orientation angle is mainly solved by two methods, namely, based on polarization and based on digital elevation model (DEM). The two results of the orientation angle are called polarization OA and DEMOA respectively.

[0003] In polarimetric SAR, the accuracy of polarization echo is a prerequisite for its function. When polarimetric SAR is in orbit, it is usually affected by factors such as hardware loss and external environment, which will have a great restriction on polarization application. This restriction is reflected in whether the relationship between the amplitude and phase of different channels is correct, which is specifically quantified as channel crosstalk and imbalance between cross-polarization and co-polarization channels. The purpose of polarization calibration is to determine the severity of distortion between different channels and correct it, so as to characterize the correct ground object information in the polarimetric image.

[0004] For decades, corner reflectors have been considered the most accurate calibration method because of their clear relative relationships in each channel. In some application scenarios, such as multi-frequency polarization calibration and areas with rugged terrain, the use of corner reflectors will consume manpower and material resources and cause great difficulties for experimenters. Therefore, more research hotspots are now focused on using only distributed targets for calibration. Among them, co-polarization channel imbalance is usually calibrated using irrotational features, namely 1) using polarization distortion parameters to select irrotational areas; 2) performing preliminary solutions based on irrotational equations; 3) removing ambiguities in preliminary results based on DEM. Among them, because polarization distortion has a greater impact on the polarization parameters of the selected irrotational area, there are errors in irrotational calibration. Summary of the invention

[0005] To solve the above technical problems, the main purpose of the present invention is to provide a co-polarization channel unbalanced calibration method based on the overall coupling component, which can be based on DEM and reduce the influence of polarization distortion on polarization parameters, and accurately solve the external calibration polarization distortion parameters.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme:

[0007] A method for calibrating unbalance of co-polarization channels based on overall coupling components comprises the following steps:

[0008] Step 1: Construct an uncalibrated covariance matrix through the uncalibrated backscatter matrix and divide the image into blocks along the azimuth and range directions;

[0009] Step 2: Based on the uncalibrated covariance matrix, solve the cross-polarization channel imbalance and crosstalk of the fully polarized SAR and perform correction;

[0010] Step 3: By setting a threshold, use DEM to select the area where OA meets the requirements; the DEM represents a digital elevation model, and the OA represents an orientation angle;

[0011] Step 4: Based on the regions in step 3, the equations are constructed using the overall coupling component of the target direction deviation to obtain three preliminary co-polarization channel imbalance results;

[0012] Step 5: Correct the polarimetric SAR image according to the three preliminary co-polarization channel imbalance results obtained in step 4, compare the obtained polarimetric OA with the DEMOA, and select the final co-polarization channel imbalance result; the DEMOA is the OA solved using the DEM.

[0013] Furthermore, in step 1, a calibration model is first constructed for the uncalibrated backscattering matrix, and the backscattering matrix is ​​used to characterize the relationship between the channels of each pixel to solve the covariance matrix of each pixel, and the covariance matrix of multiple pixels is averaged to reduce the influence of coherent speckle noise; the original image is divided into blocks along the azimuth and distance directions of the image.

[0014] Furthermore, in step 2, the high-order terms of the crosstalk value are retained through the averaged covariance matrix, and the cross-polarization channel imbalance and crosstalk are solved based on the Ainsworth algorithm considering reflection symmetry; and according to the calibration model, the solved cross-polarization channel imbalance and crosstalk are brought into the uncalibrated image and corrected.

[0015] Furthermore, in step 3, the DEM of the Space Shuttle Radar Topography Mission SRTM is forward geocoded, and the DEM is mapped from the World Geodetic System 84 coordinate system to the SAR image slant range coordinate system; then the DEM in the SAR image slant range coordinate system is converted to the DEM in the ground range coordinate system, and OA is solved and the area where OA is less than the threshold is obtained, so as to obtain the final area where OA meets the requirements.

[0016] Furthermore, in step 4, through the influence of OA on the Stokes reflection matrix and the equal relationship between polarization OA and DEMOA, utilizing the absence of an overall coupling component caused by the target azimuth deviation in the region where the final OA meets the requirements, an equation is constructed and solved to obtain three preliminary co-polarization channel imbalance results.

[0017] Furthermore, in step 5, the three preliminary co-polarization channel imbalance results are used to correct the uncalibrated backscattering matrix respectively, and the polarization OA is solved for the corrected backscattering matrix respectively, and then a correlation analysis is performed with DEMOA to find the co-polarization channel imbalance result corresponding to the maximum correlation as the final solution result.

[0018] Beneficial effects:

[0019] The present invention mainly calibrates the characteristic that the overall coupling component of the area with an azimuth angle of 0 is also 0 in the Kennaugh matrix. Considering that in actual calibration, the unbalance of the co-polarization channel is usually determined by the irrotation. However, the influence of the polarization distortion parameter on the selected area increases the calibration error, so the present invention proposes a method of calibrating the unbalance of the co-polarization channel by using the DEM to determine the ground feature information, thereby reducing the error caused by selecting the area using uncalibrated parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Schematic diagram of radar geometry;

[0021] Figure 2 It is a flow chart of the method for unbalanced calibration of co-polarization channels based on overall coupling components of the present invention;

[0022] Figure 3 This is a dataset of the satellite-borne Gaofen-3 (GF-3) simulation; (a) is the Pauli image and the block example, (b) is the DEM map, (c) is the data after adding crosstalk and channel imbalance, and (d) is the selected area result given by DEM, where white represents the selected area;

[0023] Figure 4 The following is a diagram of the actual solution result of the co-polarization channel imbalance; (a) is the amplitude solution result, and (b) is the phase solution result. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all 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 protection scope of the present invention.

[0025] According to one embodiment of the present invention, when the full polarimetric SAR system is not calibrated, the importance of using the overall coupling component of the target direction deviation to perform polarimetric calibration on the full polarimetric SAR system is:

[0026] The polarization calibration of the fully polarized SAR system mainly determines the amplitude and phase inconsistency between the four polarizations, which is mainly reflected in the imbalance of the cross-polarization and co-polarization channels and the leakage value, that is, the crosstalk value. The most accurate calibration method is to determine the polarization distortion parameters by using the relative relationship between the four channels of the corner reflector. However, considering that the corner reflector is not easy to carry and the different sizes of corner reflectors required for different bands, distributed targets are often used for calibration in practice. Nowadays, there are several algorithms for determining crosstalk and cross-polarization channel imbalance by distributed targets. The hot spots and difficulties of the research are mainly concentrated on the use of distributed targets for calibration. The main problems of distributed target calibration are how to set the calibration algorithm and how to select the area that meets the calibration algorithm when the co-polarization channel imbalance is not determined by the fully polarized data. The present invention uses the characteristic that when the orientation angle (OA) is 0, there is no overall coupling component caused by the target direction deviation for calibration for the first time. Furthermore, when selecting an area that complies with the calibration algorithm, considering that OA can be solved using an external digital elevation model (DEM), the selected area will not be affected by system distortion. Therefore, compared with selecting an irrotational area using distorted polarization parameters, the selection method proposed in the present invention is more accurate.

[0027] Furthermore, when the full polarization SAR system is not calibrated, the process of unbalanced calibration of the co-polarization channel based on the overall coupling component includes:

[0028] Firstly, the uncalibrated backscattering matrix is ​​used to construct the uncalibrated covariance matrix and the image is divided into blocks along the azimuth and range directions. Then, based on the uncalibrated covariance matrix, the cross-polarization channel imbalance and crosstalk values ​​of the full-polarization SAR are solved, and the image is corrected. By setting the threshold, the DEM is used to select the area where the OA meets the requirements. According to the selected area and the characteristic that the target direction deviation will cause the overall coupling component, an equation is constructed to determine the preliminary co-polarization channel imbalance result. According to the preliminary results of three different co-polarization channel imbalance solutions, the polarization OA obtained by solving different results is compared with the DEMOA to select the final result.

[0029] Regarding the applicability of the co-polarization channel imbalance calibration method based on the overall coupling component, for most known DEM models, the DEM is usually set to the minimum value or 0 in natural objects such as sea areas and waters, resulting in singular points when using DEM to solve polarization OA. In actual fine calibration, these areas should be removed based on single polarization data; considering that the presence of vegetation will affect the calculation of DEM and cause errors, multiple experiments can be performed for areas with dense vegetation such as forests to ensure that the impact of errors is eliminated; when using the algorithm proposed in the present invention to solve the co-polarization channel imbalance, the Newton iteration method is mainly used for solution. For the Newton iteration method, there may be non-convergence, and it is necessary to identify and remove the impact of singular values ​​during the solution.

[0030] Based on the above analysis, if Figure 2 As shown, according to an embodiment of the present invention, a method for calibrating the imbalance of a co-polarization channel based on an overall coupling component is proposed, and the method comprises the following steps:

[0031] Step 1: Construct an uncalibrated covariance matrix through the uncalibrated backscatter matrix and divide the image into blocks along the azimuth and range directions;

[0032] Step 2: Based on the uncalibrated covariance matrix, solve the cross-polarization channel imbalance and crosstalk value of the fully polarized SAR and perform correction;

[0033] Step 3: By setting the threshold, use DEM to select the area where OA meets the characteristics;

[0034] Step 4: Based on the region of step 3, an equation is constructed using the overall coupling component of the target direction deviation to determine the imbalance results of three different co-polarization channels;

[0035] Step 5: Correct the polarimetric SAR image according to the three different co-polarization channel imbalance results obtained in step 4, compare the polarimetric OA obtained with the DEMOA, and select the final result.

[0036] Furthermore, the step 1 comprises:

[0037] In general, the uncalibrated backscatter matrix can be expressed as:

[0038] (1)

[0039] in, is the distorted ground scattering matrix, ( ) are the distorted four-channel ground object scattering values; is the absolute calibration parameter; is the receiver distortion matrix, The channel at the receiving end is unbalanced. is the crosstalk at the receiving end; is the undistorted ground scattering matrix, ( ) are the undistorted four-channel ground object scattering values; is the distortion matrix at the transmitting end, The channel at the transmitter is unbalanced. is the crosstalk at the transmitter; is the ground object scattering matrix, ( ) are the noise values ​​of the four channels respectively.

[0040] Since the noise can be measured by parameters such as water, and the present invention mainly focuses on crosstalk and channel imbalance, the absolute calibration parameters and noise can be ignored. In addition to equation (1), the uncalibrated backscattering model can also be expressed as:

[0041] (2)

[0042] in, is the crosstalk matrix, u, v, w, z are the crosstalk factors respectively; is the cross-polarization channel imbalance matrix, Cross-polarization channel imbalance; is the co-polarization channel imbalance matrix, The co-polarization channels are unbalanced.

[0043] Formula (1) and formula (2) can be converted to each other, and the conversion relationship is:

[0044] (3)

[0045] (4)

[0046] According to formula (2), the uncalibrated covariance matrix can be obtained as:

[0047] (5)

[0048] Among them, the upper right corner The meaning of is conjugate transpose, Represents the averaging process of multiple pixels. is the covariance matrix of the real target scattering, and its expression is:

[0049] (6)

[0050] In actual calculations, the time for satellite-borne SAR and airborne SAR to run in azimuth is relatively short, and the data at different distances downward is related to the angle of incidence, so the polarization distortion obtained for different distances upward is also different. In order to more accurately solve the different values ​​along the distance upward, the data along the distance upward should be divided into blocks. And in order to make the subsequent solution accurate, the data along the azimuth direction is also divided into blocks.

[0051] Step 2: Based on the uncalibrated covariance matrix, solve the cross-polarization channel imbalance and crosstalk values ​​of the fully polarized SAR and perform corrections.

[0052] Since the Ainsworth algorithm ignores the high-order terms of crosstalk, the iterative results are inaccurate or even divergent when solving for higher crosstalk. Therefore, the improved Ainsworth algorithm is used to solve the problem. First, the polarimetric SAR data is divided into blocks along the range direction, and the average uncalibrated covariance matrix is ​​calculated for each block of data. Then, The initial value of is:

[0053] (7)

[0054] in, represents the elements of the unscaled covariance matrix, and Represents the matrix OK, Represents the matrix List; It means to find the angle of a complex number, the range is , is a complex unit. Then according to formula (5), First, a preliminary correction is performed, and the corrected relationship is:

[0055] (8)

[0056] in, for Initial value correction matrix; then through the Quegan algorithm, the initial values ​​of u, v, w, and z are obtained as follows:

[0057] (9)

[0058] According to formula (8) and formula (9), we can get The crosstalk correction matrix ,for:

[0059] (10)

[0060] Without removing the high-order terms of crosstalk, expand the above formula to obtain:

[0061] (11)

[0062] (12)

[0063] (13)

[0064] (14)

[0065] Among them, the superscript represents the conjugation operation of complex numbers, and The matrices are and The element in the mth row and nth column in . Assuming that the co-polarization channel and the cross-polarization channel are unrelated after removing the crosstalk and cross-polarization channel imbalance, then:

[0066] (15)

[0067] According to formula (15), the real and imaginary parts of the four variables are 0, and there are eight variables in total. The Newton iteration method can be constructed to solve the eight unknown quantities u, v, w, and z (including the real and imaginary parts respectively). Finally, the crosstalk solution is obtained based on whether the residual is less than the system index. If it is not satisfied, the formula (10) is replaced by As the update value, replace Further iterations yield the final As well as u, v, w, z and put them into equation 5, the covariance matrix with only co-polarization channel imbalance is finally obtained:

[0068] (16)

[0069] Step 3: Use DEM to select the area where OA meets the characteristics by setting the threshold.

[0070] The DEM of the Shuttle Radar Topography Mission SRTM is forward geocoded and the coordinate system of the DEM is converted to the coordinate system of the SAR image. Then the converted DEM is used to solve OA, and its expression is:

[0071] (17)

[0072] in, That is OA, is the angle of incidence, is the distance slope, is the azimuthal slope.

[0073] according to Figure 1 The schematic diagram of the radar geometric structure is shown in is the curve normal, then:

[0074] (18)

[0075] (19)

[0076] in, , , are the unit normal vectors of the curve normal in the azimuth, ground distance and altitude directions respectively; , , They are the azimuth resolution, the ground range resolution, and the elevation difference between adjacent pixels on the image.

[0077] By using equations (17), (18) and (19), the influence of the error is fully considered and the absolute value of OA is set to When the value is less than a certain threshold, a region is selected, which means that it is considered that there is no overall coupling component between symmetric targets and asymmetric targets in the region.

[0078] Step 4: Calibrate the selected area in step 3 to obtain the co-polarization channel imbalance The preliminary results.

[0079] The area where the overall coupling component is 0 due to the target azimuth deviation satisfies:

[0080] (20)

[0081] in, To take the real part operation, For the matrix The element in the mth row and nth column. The existence of is not 0. Combined with formula (16), The overall coupling component expression of the influence is:

[0082] (twenty one)

[0083] in, for The reciprocal of For the matrix The element in the mth row and nth column has a superscript represents the conjugate of the complex number. It is easy to see that equation (21) has a constant solution, namely In order to remove the error caused by this solution, the equation is divided as a whole , removing the difference caused by the 0 value, that is:

[0084] (twenty two)

[0085] Solved by Newton's iteration method , find , and then find the candidate value of co-polarization channel imbalance. Since in formula (20) The phase is close to 0° and ±180°, so the value obtained by equation (22) contains a phase difference of 90°±180°. Therefore, after solving, the whole is shifted by 90°, and then the phase value is taken between -180° and 180°. And considering that the actual Newton iteration method may have the problem of non-convergence, the obtained amplitude and phase are filtered, and then the least squares method is used to fit to obtain three preliminary co-polarization channel imbalance results.

[0086] Step 5: Correct the polarimetric SAR image according to the three different co-polarization channel imbalance results obtained in step 4, and compare the polarimetric OA and DEMOA obtained with different results. Select the co-polarization channel imbalance corresponding to the combination with the strongest correlation, which is the final result.

[0087] Substitute the obtained candidate values ​​into the corrected u, v, w, z and Then, the polarization OA is calculated using the selected polarization data and correlated with the OA calculated by DEM to find the value corresponding to the maximum correlation. , which is the final The technical solution of the present invention is further described in detail below in conjunction with specific embodiments.

[0088] Example 1

[0089] GaoFen-3 (GF-3) is a civilian microwave remote sensing imaging satellite and also my country's first C-band SAR satellite. It has the characteristics of high resolution, large imaging width, high radiation accuracy, multiple imaging modes and long-term operation, and can monitor global ocean and land information around the clock. Considering that artificially adding channel imbalance and crosstalk can clearly find the algorithm error under actual verification, the present invention uses artificially adding channel imbalance and crosstalk to verify the actual effect of the algorithm.

[0090] Figure 3 The data set of GF-3 simulation is displayed. The data is selected from the desert area to avoid the influence of water areas. Figure 3 (a) shows the Pauli diagram of the selected data, where the dotted lines divide the image into blocks in azimuth and range. Figure 3 (b) shows the result obtained by projecting the SRTM DEM onto the ground-based SAR image. Since the DEM is obtained using external data, it is not affected by polarization distortion. Figure 3(c) is the Pauli image obtained by simulation after adding different polarization distortions. By setting the threshold through equation (17), the area where the polarization OA meets the requirements can be obtained, such as Figure 3 The white dots in (d).

[0091] Figure 4 The actual solution results of the co-polarization channel imbalance are given, where Figure 4 (a) is the amplitude solution result, Figure 4 (b) is the phase solution result. Compared with the actual value represented by the realization, the fitting value of the dotted line has a higher degree of fit between amplitude and phase. Through calculation, the actual amplitude error and phase error are 0.0496dB and 0.5278°, which can meet the actual GF-3 calibration requirements.

[0092] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for unbalanced calibration of co-polarization channels based on overall coupling components, characterized in that: Based on DEM and reducing the influence of polarization distortion on polarization parameters, the external calibration polarization distortion parameters are accurately solved, including the following steps: Step 1: Construct an uncalibrated covariance matrix through the uncalibrated backscatter matrix and divide the image into blocks along the azimuth and range directions; Step 2: Based on the uncalibrated covariance matrix, solve the cross-polarization channel imbalance and crosstalk of the fully polarized SAR and perform correction; Step 3: By setting a threshold, use DEM to select the area where OA meets the requirements; the DEM represents a digital elevation model, and the OA represents an orientation angle; Step 4: Based on the regions in step 3, the equations are constructed using the overall coupling component of the target direction deviation to obtain three preliminary co-polarization channel imbalance results; pass The overall coupling component expression of the influence is: (21) in, for The reciprocal of For the matrix The element in the mth row and nth column has a superscript represents the conjugate of a complex number, and equation (21) has a constant solution, namely , divide the equation as a whole by , remove the difference caused by 0 value; Indicates the selected area. For co-polarization channel imbalance, is the real part operation; Step 5: Correct the polarimetric SAR image according to the three preliminary co-polarization channel imbalance results obtained in step 4, compare the obtained polarimetric OA with the DEMOA, and select the final co-polarization channel imbalance result; the DEMOA is the OA solved using the DEM.

2. The method for unbalanced calibration of co-polarization channels based on overall coupling components according to claim 1, characterized in that: The step 1 specifically includes: Firstly, a calibration model is constructed for the uncalibrated backscattering matrix, and the backscattering matrix is ​​used to characterize the relationship between each pixel channel to solve the covariance matrix of each pixel, and the covariance matrix of multiple pixels is averaged to reduce the influence of coherent speckle noise; the original image is divided into blocks along the azimuth and distance directions of the image.

3. The method for unbalanced calibration of co-polarization channels based on overall coupling components according to claim 2, characterized in that: In the step 2, the high-order terms of the crosstalk value are retained through the averaged covariance matrix, and the cross-polarization channel imbalance and crosstalk are solved based on the Ainsworth algorithm considering reflection symmetry; and according to the calibration model, the solved cross-polarization channel imbalance and crosstalk are brought into the uncalibrated image and corrected.

4. The method for unbalanced calibration of co-polarization channels based on overall coupling components according to claim 3, characterized in that: In the step 3, the DEM of the Space Shuttle Radar Topography Mission SRTM is forward geocoded, and the DEM is mapped from the World Geodetic System 84 coordinate system to the SAR image slant range coordinate system; then the DEM in the SAR image slant range coordinate system is converted to the DEM in the ground range coordinate system, and OA is solved and the area where OA is less than the threshold is obtained, so as to obtain the final area where OA meets the requirements.

5. The method for unbalanced calibration of co-polarization channels based on overall coupling components according to claim 4, characterized in that: In the step 4, through the influence of OA on the Stokes reflection matrix and the relationship between polarization OA and DEMOA, and by utilizing the fact that there is no overall coupling component caused by the target azimuth deviation in the area where the final OA meets the requirements, an equation is constructed and solved to obtain three preliminary co-polarization channel imbalance results.

6. The method for unbalanced calibration of co-polarization channels based on overall coupling components according to claim 5, characterized in that: In step 5, the three preliminary co-polarization channel imbalance results are used to correct the uncalibrated backscattering matrix respectively, and the polarization OA is solved for the corrected backscattering matrix respectively, and then a correlation analysis is performed with DEMOA to find the co-polarization channel imbalance result corresponding to the maximum correlation as the final solution result.

Images