Interference processing method for dual-antenna fully polarized SAR system

Through the interference processing method under the dual-antenna fully polarized SAR system, backscatter coefficient simulation and multi-view optimization algorithm are used to solve the problem of interference processing in lunar SAR observation, realize the effective application of non-polarized data, and improve the The scientific mission capability and measurement accuracy of the lunar microwave imaging radar.

CN118759524BActive Publication Date: 2024-11-01AEROSPACE INFORMATION RES INST CAS
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411215684.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-01
Estimated Expiration
2044-09-02

AI Technical Summary

Technical Problem

In the field of lunar observation, the current lunar SAR observation function lacks the conditions to realize more mature SAR topographic mapping solutions such as dual-antenna interference and multi-star formations, which makes it difficult to apply SAR interference processing technology and affects the accuracy of elevation measurement.

Method used

A interference processing method under the dual-antenna fully polarized SAR system is proposed. Through backscattering coefficient simulation and multi-view optimization algorithm, the polarization phase error of non-polarization interference is reduced, and the polarization phase similarity index is introduced to achieve non-polarization. Interference processing of homopolarized data expands the data application dimension.

Benefits of technology

It effectively reduces traditional noise, improves the possibility of homopolarization interference processing, increases the application dimension of scientific tasks of lunar microwave imaging radar, and improves the accuracy and reliability of interference processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118759524B_ABST
    Figure CN118759524B_ABST
Patent Text Reader

Abstract

The present invention discloses an interference processing method under a dual-antenna fully polarized SAR system, and belongs to the field of satellite-borne synthetic aperture radar interference processing technology. The method comprises: step 1, simulating the backscattering coefficient in combination with satellite system parameters and scattering characteristics of different types of objects; step 2, obtaining full polarization data with a baseline, selecting the co-polarization data pair with the smallest difference in polarization characteristics according to the difference in polarization characteristics of different channels, and determining the scene terrain slope range for interference processing according to the threshold of the backscattering coefficient difference; step 3, using a multi-view optimization algorithm to reduce the polarization phase error of non-co-polarization interference, and introducing a polarization phase similarity index to evaluate interference performance and analyze its application. The present invention proposes an index of polarization similarity, which can effectively evaluate the possibility of interference of non-co-polarization data, and is conducive to the promotion and application of interference processing technology using full polarization data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of spaceborne synthetic aperture radar interferometry processing technology, and involves multiple key technologies such as spaceborne synthetic aperture radar, dual antennas, full polarization, and interferometry. The present invention can be applied to a dual-antenna fully polarized SAR system and interferometry processing and analysis of fully polarized data that meets baseline constraints, and specifically relates to an interferometry processing method under a dual-antenna fully polarized SAR system. Background Art

[0002] Earth observation technology is an important way to achieve global environmental and dynamic change monitoring. Spaceborne Synthetic Aperture Radar (SAR) uses satellites and other spacecraft as a mobile platform, has all-day, all-weather, and global observation capabilities, and has become an indispensable means of earth observation. Interferometric SAR (InSAR) technology, which has evolved from SAR technology, continues the advantage of SAR being able to work all day and all weather. It uses the phase information in SAR images to measure surface elevation and is widely used in terrain mapping. The interference of incoherence factors in the interference pattern is the key to affecting the accuracy of terrain phase extraction. Among them, temporal incoherence is the only incoherence factor that can be completely removed by hardware design.

[0003] Dual-antenna interferometry and multi-satellite formation are the two most commonly used observation schemes that can remove temporal incoherence, and are also the mainstream SAR altimetry schemes currently recognized. The current SAR mapping application design for earth observation also revolves around dual-antenna interferometry and multi-satellite formation, and is developing towards higher accuracy and higher efficiency.

[0004] In the field of lunar observation, lunar altitude measurement is also an important part of lunar exploration. In the exploration of the moon, several exploration plans are equipped with laser altimeters or optical cameras, and more than ten types of elevation data with different resolutions, accuracy and coverage have been obtained. However, there is still a blank in obtaining elevation data through SAR payload interferometry technology. This is mainly because the current lunar monitoring is mainly based on microwave image observations of special areas, and there are few SAR payloads in orbiters, so the acquisition of microwave image data is relatively limited.

[0005] SAR remote sensing resources for the moon are relatively scarce. The current development direction of lunar SAR observation functions is high-precision and multi-polarization observation. It is not possible to achieve more mature SAR terrain mapping solutions such as dual-antenna interferometry and multi-satellite formations in a short time. Therefore, how to use the existing lunar SAR observation conditions to achieve SAR interferometry processing will provide support for subsequent lunar interferometric mapping and has great application value. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides an interference processing method under a dual-antenna fully polarized SAR system. For the dual-antenna fully polarized SAR system of the lunar microwave imaging radar, a method of using a single fully polarized data to realize interference processing is studied, aiming to increase the application dimension of limited data on the basis of the existing hardware system and enrich the scientific tasks of the lunar microwave imaging radar. The present invention proposes an interference processing method for non-co-polarized data, and proves the effectiveness of the method for interference processing by comparing the interference processing results with the co-polarized data; and analyzes the non-co-polarized data of the fully polarized SAR system, studies the difference in polarization phase and the influencing factors, and proposes the scope of application of the method. The invention can be applied to dual-antenna fully polarized SAR systems and fully polarized systems that meet baseline constraints, expanding the data range of existing interference applications such as deformation and height measurement.

[0007] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0008] An interference processing method in a dual-antenna fully polarized SAR system comprises the following steps:

[0009] Step 1: simulate the backscatter coefficient by combining satellite system parameters and scattering characteristics of different types of ground objects, and calculate the difference of backscatter coefficients obtained under different polarization modes based on the simulated backscatter coefficient;

[0010] Step 2: Obtain full polarization data with a baseline, select the co-polarization data pair with the smallest polarization characteristic difference according to the polarization characteristic difference of different channels, and determine the scene terrain slope range for interference processing according to the threshold of the backscatter coefficient difference;

[0011] Step 3: For data that meets the terrain slope range of the scene, a multi-look optimization algorithm is used to reduce the polarization phase error of non-co-polarization interference, and a polarization phase similarity index is introduced to evaluate the interference performance and conduct application analysis.

[0012] The beneficial effects of the present invention are:

[0013] The present invention uses the dual-scale model to simulate the lunar surface polarization scattering coefficient results to evaluate the confidence of different co-polarization interference results, and proposes a polarization similarity index based on this. Based on the simulation results and the multi-look processing method, while reducing traditional noise, the polarization phase difference between HH and VV is effectively reduced, and the possibility of different co-polarization interference processing is improved.

[0014] Based on the hardware system of the existing dual-antenna fully polarized SAR system of the lunar microwave imaging radar, the application dimension of fully polarized data is increased. It can also be applied to other dual-antenna fully polarized SAR systems and fully polarized systems that meet baseline constraints. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of the full polarization operation of the Ku-band antenna;

[0016] Figure 2 This is a flowchart of an interference processing method in a dual-antenna fully polarized SAR system according to the present invention;

[0017] Figure 3 Schematic diagram of the backscattering difference between HH and VV polarizations of the lunar surface simulated based on the dual-scale model;

[0018] Figure 4 Schematic diagram of the calculation of HH and VV polarization backscattering coefficients, where (a) is the relative phase under the KA model at a small incident angle Schematic diagram, (b) is the relative phase under the SPM model with large incident angle Schematic diagram;

[0019] Figure 5 is the phase difference distribution histogram between HH and VV polarizations;

[0020] Figure 6 is a graph showing the standard deviation changing with the number of views;

[0021] Figure 7 It is the curve of standard deviation changing with the number of azimuth views;

[0022] Figure 8 Schematic diagrams of the statistical results of coherence coefficients of different interference pairs, where (a) is a schematic diagram of the statistical results of the 20181221 HH polarization-20181221 VV polarization interference pair, (b) is a schematic diagram of the statistical results of the 20190119 HH polarization-20190119 VV polarization, (c) is a schematic diagram of the statistical results of the 20181221 HH polarization-20190119 HH polarization, and (d) is a schematic diagram of the statistical results of the 20181221HH polarization-20190119 VV polarization;

[0023] Fig. 9 Schematic diagram of polarization phase difference after multi-look processing, where (a) is the result without multi-look processing; (b) is the result of 2 2 Schematic diagram of multi-view processing, (c) is 3 2 Schematic diagram of multi-view processing, (d) is 4 3. Schematic diagram of multi-view processing;

[0024] Fig.10 is the curve of polarization phase error standard deviation changing with the number of multiple looks;

[0025] Fig.11 An intensity map of a region selected for an embodiment of the present invention;

[0026] Fig.12 An elevation map of a selected area for an embodiment of the present invention;

[0027] Fig.13 The interferometric DEM inversion results and regional magnification results are shown, where (a) is the 20181221 HH-20190119HH interferometric DEM inversion result map, and (b) is the 20181221 HH-20190119 VV interferometric DEM inversion result map;

[0028] Fig.14 Schematic diagram of interferometric DEM inversion error, where (a) is the schematic diagram of 20181221 HH-20190119 HH interferometric DEM inversion error, and (b) is the schematic diagram of 20181221 HH-20190119 VV interferometric DEM inversion error;

[0029] Fig.15 Schematic diagram of the incident angle in the selected area radar coordinate system;

[0030] Fig.16 Schematic diagram of backscattering coefficient under different simulated incident angles;

[0031] Fig.17 Schematic diagram of polarization phase similarity. DETAILED DESCRIPTION

[0032] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0033] The existing lunar imaging microwave imaging radar has comprehensively considered the scientific mission requirements, designed a Ku and L dual-band payload scheme, and has the function of full polarization. The antenna array is an active phased array system, and the phased array antenna consists of 1 set of L-band fully polarized antennas and 2 sets of Ku-band linear antennas. The three sets of antenna structures share a mounting plate and a frame, with the L-band fully polarized antenna in the center, and the Ku-band H-polarized antenna and the Ku-band V-polarized antenna on both sides. The two sets of Ku-band H-polarized and V-polarized antennas only have different antenna radiation arrays, and the architectural layout of other modules is exactly the same. The two sets of Ku-band antennas are separated on the left and right, and a spatial baseline of nearly 1m is generated between the phase centers of the two antennas. In addition to the different polarization modes, the basic conditions of the dual-antenna interference scheme are met. Based on this, the present invention studies the single-track interference processing method for the dual-antenna fully polarized system of the lunar microwave imaging radar.

[0034] like Figure 1The figure shows the full polarization operation diagram of the Ku-band antenna. The full polarization SAR adopts the "alternating transmission and simultaneous reception" operation mode. The system alternately transmits horizontal (H) and vertical (V) orthogonal linear polarization pulses, and uses H and V linear polarization channels to simultaneously receive echo signals to obtain four images of HH / HV / VH / VV. According to the working mode design of the lunar microwave imaging radar, the Ku-band H-polarization antenna and the Ku-band V-polarization antenna are separated, and the paths of the electromagnetic signals are represented as r1 and r2 respectively. The electromagnetic wave processes of the four images obtained can be represented as follows:

[0035] (1)

[0036] In the formula, R represents the process, the left subscript of R represents the polarization of the transmitted electromagnetic wave, and the right subscript of R represents the polarization of the received electromagnetic wave.

[0037] Path difference is the basic condition for interference. Therefore, there are three types of image pairs with path difference: HH and VV, HH and HV / VH, and VV and HV / VH. HH and VV are both co-polarized, while HV and VH are cross-polarized. The difference in scattering characteristics between co-polarized and cross-polarized is greater than the difference in scattering characteristics between different co-polarized images. Therefore, when co-polarized interference is not satisfied, different co-polarized pairs are preferentially selected for interference, and then the difference in scattering characteristics between different co-polarized data is minimized as much as possible. At the same time, the difference between different co-polarized data needs to be considered. According to the system and scene constraints, the target range that meets the error limit of the interference application is analyzed, and then the accuracy of the interference processing is improved through interference processing analysis.

[0038] like Figure 2 The figure shows the entire algorithm flow chart of the interference processing method under the dual-antenna fully polarized SAR system of the present invention. The backscattering coefficient is simulated by combining the satellite system parameters and the scattering characteristics of different types of objects; the full polarization data with a baseline is obtained, and the co-polarization data pair with the smallest difference in polarization characteristics is selected according to the difference in polarization characteristics of different channels, and the scene terrain slope range suitable for interference processing is determined according to the threshold of the backscattering coefficient error; the multi-view optimization algorithm is used to reduce the polarization phase error of non-co-polarization interference, and the polarization phase similarity index is introduced to evaluate the interference performance and analyze the application. The specific processes are introduced below.

[0039] 1. Backscattering coefficient simulation

[0040] In order to study the differences in the characteristics of different co-polarization data, especially the characteristics of polarized electromagnetic scattering on the lunar surface, it is necessary to understand the electromagnetic scattering mechanism of the rough surface of the lunar surface. Many researchers have conducted research on the scattering coefficient of the rough surface of the lunar surface. The research team of Fudan University constructed the lunar surface topography from the digital elevation model (DEM) measured by the Lunar Orbiter Laser Altimeter (LOLA), and divided the lunar surface into many small patches. The side length of these patches is less than the SAR resolution. A dual-scale model is used to simulate the electromagnetic scattering of the rough surface of the lunar surface. When the local incident angle is small, the Kirchhoff approximation (KA) is used to calculate the backscattering coefficient; when the local incident angle is large, the backscattering coefficient is calculated using the perturbation method (SPM). The backscattering coefficient simulated by this method is compared with the L-band DFSAR data of the Chandrayaan-2. When the incident angle is less than 35°, the simulation results are consistent with the data. In the Ku-band full polarization mode of the lunar microwave imaging radar studied in the present invention, the near-end wave position is also dominant, which is basically within the high confidence interval of the dual-scale model. Therefore, subsequent simulation and analysis are carried out based on this.

[0041] The backscattering coefficient of the lunar microwave imaging radar Ku band is simulated based on the dual-scale model. Typical lunar surface parameters are selected. The difference between the backscattering coefficients of HH and VV polarizations is as follows: Figure 3 As shown, when the incident angle is approximately less than 12.1°, the backscattering coefficients of HH and VV are the same. Then, as the incident angle increases, the difference in the backscattering coefficients of HH and VV gradually increases. When the incident angle is approximately less than 30.25°, the difference between the two is less than 2dB. When the incident angle increases to 51°, the difference between the two reaches 5dB.

[0042] In addition, relative phase is also the focus of subsequent interference. By analyzing the distance and azimuth compression of synthetic aperture radar signals, it is found that when the incident angle is small, KA scattering is dominant and HH polarization scattering echo signal S HH and VV polarization scattered echo signal S VV The amplitudes of are calculated by KA, and there is an equal relationship. The backscattering coefficient of HH polarization and the backscattering coefficient for VV polarization is obtained by weighted summation of the backscattering coefficients of multiple facets. and The impact is the same, so and The relative phase between them is 0°. When the incident angle is large, the co-polarization S HH and S VV The amplitude is calculated by SPM. The difference between the HH and VV backscatter coefficients increases with the increase of the incident angle, making the relative phase Deviating from 0°, such as Figure 4 As shown, (a) is the relative phase under the small incident angle KA model Schematic diagram, (b) is the relative phase under the SPM model with large incident angle Schematic diagram.

[0043] Therefore, the polarization phase difference between HH and VV is basically consistent with the simulation law of their backscattering, that is, within a limited small incident angle range, the polarization phases of HH and VV are the same, and then as the incident angle increases, the polarization phase difference between the two gradually increases. Here we define a polarization phase similarity index between HH and VV:

[0044] (2)

[0045] in, represents the angle of incidence, and represent the backscattering coefficients of HH polarization and VV polarization respectively, The acceptable backscatter coefficient error for interferometric processing can be determined by the specific scenario. When the value is 0, it means that the polarization phase similarity between HH and VV is 0, which is not suitable for subsequent interference processing; when the value is 1, it means that there is no difference in the polarization phase between HH and VV; Other values ​​can also be used as the confidence of the interferometric processing at the incident angle and as an evaluation index of the relevant processing accuracy. According to the simulation of the lunar surface backscatter coefficient, in the subsequent application where different co-polarized data will be interferometrically processed, the acceptable backscatter coefficient difference of the interferometric processing can be determined according to the system parameters and scene characteristics, and the confidence of the interferometric processing results at different incident angles can be evaluated within the acceptable polarization error range.

[0046] 2. Feature analysis and multi-view interference processing

[0047] The smaller the incident angle, the smaller the backscatter difference between HH and VV. Even when the incident angle is small enough, there will be no difference between the two. Therefore, in order to expand the application scenarios of different co-polarization interference and improve the accuracy of subsequent interference processing, an algorithm can be used to reduce the incident angle as much as possible. This incident angle is directly related to the surface where polarization scattering occurs, and it combines terrain information and satellite parameters. , which can be expressed by the following formula:

[0048] (3)

[0049] in, represents the downward viewing angle of satellite observation, is the terrain slope.

[0050] There are large and small craters and rocks widely distributed on the rough surface of the lunar surface. Its landform determines the complexity of its terrain slope. It is also divided into many small patches as described above. When the steep terrain dominates in the resolution unit of the image, the large incident angles in each patch will also account for the majority, which is not conducive to subsequent interference processing. Therefore, it is considered to reduce the incident angle based on the resolution unit calculation. As the resolution unit continues to expand, the extremely steep terrain will be smoothed to achieve the purpose of reducing the incident angle, thereby reducing the polarization noise caused by non-co-polarization interference (HH and VV interference).

[0051] The dual-antenna full polarization system of the lunar microwave imaging radar is equivalent to superimposing polarization noise on the traditional interference data. If the polarization noise is small enough, or is at the same level as other interference noise after processing, it can be considered feasible to use non-co-polarization data for interference. Therefore, the SAR co-polarization image data can be processed by statistical smoothing to improve the confidence of different co-polarization interference processing results. The following is an analysis of the smoothing method in a statistical sense based on the full polarization data obtained by the Chandrayaan-2 orbiter. A scene of L-band full polarization data is selected, and the specific data information is shown in Table 1 below.

[0052] Table 1

[0053]

[0054] From the dual-scale model simulation, it can be seen that when the local incident angle is less than 30.25°, the difference in the backscatter coefficients of HH and VV is less than 2dB. It is calculated that the corresponding terrain slope constraint should be less than 56.25°, which satisfies most cases of lunar surface terrain. It can be considered that the similarity of the SAR data of HH polarization and VV polarization in this scene is within an acceptable range and has the possibility of interference processing. The following uses statistical smoothing to further reduce its slope and verify the effect of multi-look processing on the polarization phase. The phase difference between the HH and VV data in the polarization data is shown in Figure 5 As shown in the figure. Since the pixel resolution of the data is quite different in the range and azimuth directions, the azimuth and range directions are processed according to the downlink parameters at a ratio of 18:1, making it close to the visual observation resolution effect. After multi-view processing according to integer multiples, the mean and standard deviation of the phase are statistically calculated, and the results are shown in Table 2 below.

[0055] Table 2 (Unit: degrees)

[0056]

[0057] Among them, 1 Range Representative It can be seen from Table 2 that the phase difference standard deviation of HH and VV is significantly reduced after multi-look processing, and the larger the number of multi-looks, the smaller the polarization phase standard deviation. After multi-view processing, the polarization phase is 1 The value is less than 10°, which means that the interferometric polarization phase after multi-look processing can be reduced to the general interferometric phase error level and can meet the constraints of interferometric processing. To further refine the change law, the multi-look processing value is optimized and the standard deviation changes under different multi-look numbers are statistically analyzed, as shown in Figure 2. Figure 6 shown.

[0058] As the number of multi-looks increases, the interference polarization phase of the image gradually decreases, and the decreasing acceleration rate (within a limited range) gradually decreases and approaches 0. For this data, when the number of multi-looks is greater than 40, the interference polarization phase is 1 The value is less than 10°. After multi-view processing according to the nominal resolution of the system product 25m ( or ), the interference polarization phase is 1 The value distribution can be reduced to around 8.5°.

[0059] However, the above analysis does not take into account the difference between the azimuth and range views, and there may be local abnormal points. Therefore, further analysis will be conducted below. Taking the HH-VV interferometric polarization phase data as an example, the standard deviation difference is analyzed when the overall views are the same but the azimuth and range views are different. Taking the overall view 72 as an example, the azimuth and range views are adjusted, and the phase standard deviation is as follows: Figure 7 shown.

[0060] When the overall number of multi-looks is the same, the closer the azimuth and range views are, the larger the standard deviation is. In the test case, when the overall number of multi-looks is the same and the difference between the azimuth and range views is large, a larger azimuth view can obtain a lower standard deviation. It is speculated that this is related to the difference in the original resolution of the data in the azimuth and range directions, and the azimuth resolution is higher. When the overall number of views is 72, the standard deviation under different azimuth multi-looks is between 7.4° and 9.4°, and the overall difference does not exceed 2°. Therefore, for the processing of polarization phase, generally speaking, the larger the number of multi-looks, the smaller the phase standard deviation. When the overall number of multi-looks is the same, the standard deviation caused by the difference in azimuth and range views is not much different.

[0061] According to the statistical results of the data, after multi-view processing of the system product with a nominal resolution of 25m, the magnitude of the polarization phase error can be reduced to the same magnitude as the traditional interferometric phase error. Based on the slope information provided by the external DEM information, the scene can be subdivided according to the results of the backscatter coefficient simulation, and the confidence of the interferometric processing results can be evaluated based on the value of the polarization similarity index.

[0062] Example

[0063] Since the lunar microwave imaging radar has not been launched yet and the existing full polarization data of the moon does not have an interferometric baseline, in order to verify the effect of the present invention, the full polarization data of the GF-3 satellite obtained by two scene re-orbits are selected to verify the method. The basic information of the two scene data is shown in Table 3:

[0064] Table 3

[0065]

[0066] Both scene data are fully polarized data, which have data from four channels of HH+HV+VH+VV, that is, four single-view complex images. Traditional interference processing generally uses image pairs with the same polarization mode, such as two single-view complex images with a certain baseline that are both obtained by HH (or VV) polarization mode. This experiment simultaneously analyzes the interference results of single-view complex images obtained by different polarization modes at the same time and different polarization modes at different times.

[0067] (1) Polarization decoherence factor

[0068] Three types of data, namely, different polarization modes at the same time, the same polarization mode at different times, and different polarization modes at different times, are used for interference processing to analyze the influence of polarization decoherence on the interference coherence coefficient. 5 The results after coherent processing are as follows Figure 8 The corresponding statistical means and variances are shown in Table 4.

[0069] Table 4

[0070]

[0071] Figure 8 In the figure, (a) and (b) are the coherence coefficients after HH polarization and VV polarization interference at the same time. Since there is no baseline for the corresponding interference pair, the difference is only caused by different polarization modes. Comparing the statistical results of the coherence coefficients of the two interference pairs in Table 4, it is found that the difference between the two is very small. The difference may come from the change of the scattering characteristics of the ground objects at different times. However, since the time difference between the two is very small, the difference between the ground objects is not obvious. Compared with the decoherence caused by the polarization difference, the difference can be ignored. Figure 8In the figure, the statistical results of the coherence coefficients of (c) and (d) are generally similar. The mean coherence coefficient of the 20181221 HH polarization-20190119 VV polarization interference pair in Table 4 is slightly lower than that of the 20181221 HH polarization-20190119 HH polarization interference pair. The difference is mainly caused by polarization decoherence. In this case, the calculated polarization decoherence factor is 0.9173 (=0.3115 / 0.3396). The results show that in the presence of other decoherence factors, the influence of polarization decoherence on the interference coherence coefficient is small. Therefore, starting from the coherence coefficient, the data interference of different polarization modes that meet the basic interference conditions has little change in coherence compared with the data interference of the same polarization mode, and has the ability of interference processing.

[0072] (2) Polarization phase analysis

[0073] Generally speaking, the phase difference between the co-polarization interference data at different times and the non-co-polarization interference data at different times is a phase error caused by different polarization interferences, which can be called polarization phase error. The experiment selected the phase difference between the 20181221HH polarization-20190119HH polarization interference pair and the 20181221HH polarization-20190119VV polarization interference pair as the polarization phase error for analysis. Multi-view processing with different view numbers is performed to analyze the polarization phase changes caused by non-co-polarization interference, such as Fig. 9 As shown, (a) is a schematic diagram of the result without multi-view processing; (b) is a schematic diagram of 2 2 Schematic diagram of multi-view processing, (c) is 3 2 Schematic diagram of multi-view processing, (d) is 4 3 Schematic diagram of multi-look processing. The statistical mean and standard deviation of the polarization phase difference are shown in Table 5.

[0074] Table 5

[0075]

[0076] It can be seen from the data in Table 5 that with the increase of the number of multi-looks, the polarization phase difference gradually decreases, the statistical mean rapidly drops to near 0, and the standard deviation also decreases. In order to study the reduction law of the standard deviation, the number of multi-looks is further increased, the relationship between the number of multi-looks and the standard deviation is analyzed, and the distribution of the polarization phase difference under different multi-looks is calculated. The data results are expressed in the form of a statistical curve as shown below: Fig.10 shown.

[0077] Depend on Fig.10It can be seen that the polarization phase error decreases with the increase of the number of multiple looks. Compared with the previous analysis results, the change law of the polarization phase error with the presence of a spatial baseline has not changed much. It also decreases with the increase of the number of multiple looks. The decline speed is fast at first, then slows down and gradually stabilizes. Therefore, the polarization phase error can be reduced by statistical smoothing.

[0078] (3) Interference processing results

[0079] The following uses elevation inversion to verify the results of the interference processing. Some typical target scenes in the above GF-3 data are intercepted for interference processing, and the fuzzy height of the interference system is calculated based on the known parameters. The basic interference parameters are shown in Table 6 below:

[0080] Table 6

[0081]

[0082] The fuzzy height of the system is 5.4m. Too small a fuzzy height is not conducive to the untangling of the interference pattern, especially in steeper areas with large height changes. Therefore, the elevation error of the interferometric height measurement test is large. It is only used here for the polarization influence analysis of the full polarization interference test and the comparison of the results of HH-HH interference and HH-VV interference, without considering the optimization of its accuracy.

[0083] The data of the same polarization mode at different times and the data of different polarization modes are selected for processing to compare and evaluate the ability of the interference of different polarization modes. From the previous analysis and research, it can be seen that multi-viewing will improve the effect of interferometry, and the horizontal grid of the elevation data selected from ALPSMLC30 is 30 meters. In order to make the interference results as close as possible, the data are processed 6 times. 6. Interference processing. According to the characteristics of the data, a relatively flat area in the scene is selected for interference result analysis. The intensity map and elevation map DEM (source: ALPSMLC30) of the selected area are shown in the figure. Fig.11 and Fig.12 shown.

[0084] The DEM information is used to establish ground control points and re-estimate the baseline of the interference processing to improve the estimation accuracy of the baseline as much as possible. The 20181221 HH data is used as the main image, and the 20190119 HH and 20190119 VV data are used as auxiliary images for interference processing. After conventional interference processing steps such as registration, resampling, interference, deflating, filtering, unwrapping, phase-elevation conversion, etc., the inversion results of DEM are obtained, such as Fig.13 As shown, (a) is the 20181221 HH-20190119 HH interference DEM inversion result, and (b) is the 20181221 HH-20190119 VV interference DEM inversion result.

[0085] Depend on Fig.13 The results show that the processing results of non-co-polarization interference at different times are basically consistent with the results of co-polarization processing. However, due to data resolution issues, the details of the inversion results are greatly lost, but this is a load and data problem and has little to do with the selection of non-co-polarization data. From the magnified results of the red frame area, the contour lines of the DEM can be clearly seen in both sets of processing results, indicating that compared with co-polarization interference data, non-co-polarization data also has the ability to perform interference processing.

[0086] like Fig.14 As shown in the figure, (a) is a schematic diagram of the inversion error of the 20181221 HH-20190119 HH interferometric DEM, and (b) is a schematic diagram of the inversion error of the 20181221 HH-20190119 VV interferometric DEM. In most areas of the scene, the inversion error of the interferometric DEM is small, and its error is mainly concentrated in the building area in the upper left corner of the scene. The non-co-polarization interferometric error is higher than the co-polarization interferometric error, and it is more obvious in the building area. The difference is small in the bare land area, which is also directly related to the greater difference in images of different polarization modes in the building area. From the statistics of the interferometric DEM inversion error, it can be seen that the mean error of the 20181221 HH-20190119 HH interferometric is 3.8395m, and the standard deviation is 3.4069; the mean error of the 20181221 HH-20190119 VV interferometric is 4.5485m, and the standard deviation is 3.8922. There is a difference in error between co-polarization interference and non-co-polarization interference, but the difference is not large and is of the same order of magnitude. Therefore, in special circumstances, non-co-polarization interference can be used as a means of height measurement.

[0087] (4) Confidence Assessment

[0088] Based on the simulated backscatter coefficient, the interferometric processing method proposed in this invention is quantitatively evaluated for its ability to process different co-polarized data. Since the scene is small, the downward viewing angle of the satellite is uniformly approximated to be 18.9°. The terrain slope is obtained based on the external DEM data, and the local incident angle information of the area is calculated in combination with formula (3), as follows: Fig.15 As shown in the figure, the pixels with a local incident angle of less than 12° account for 59.27%, the pixels with a local incident angle of less than 15° account for 81.25%, and the pixels with a local incident angle of less than 20° account for 95.14%.

[0089] First, the backscatter coefficients of HH and VV of the above GF-3 data are simulated. Since the earth scene is selected, the backscatter coefficients of HH and VV of the SAR image are simulated according to the Urabi model (dry snow) to obtain the backscatter coefficients of HH and VV at different incident angles, as shown in Fig.16 shown.

[0090] The maximum difference between the backscatter coefficients of HH and VV obtained by simulation occurs near the incident angle of 46°, and the difference is about 1.3 dB. The maximum error is taken as the allowable backscatter coefficient difference, and the polarization phase similarity of the scene is calculated according to formula (2), as follows: Fig.17 shown.

[0091] Depend on Fig.17 It can be seen that after multi-view smoothing, the polarization similarity of most areas of the scene is higher than 0.8, and the average polarization phase similarity is 0.7738. The polarization phase similarity index map is consistent with the inversion error. Fig.14 (b) in the figure is basically consistent, which once again confirms the reliability of the polarization phase similarity index.

[0092] In general, in special load design situations, when the conditions for co-polarization interferometry are not met, non-co-polarization interferometry can be used to invert the interferometric DEM. The accuracy will decrease, but the measurement error is within an acceptable range for relatively flat areas. In practical applications, the threshold of polarization similarity can be further determined to select scenarios with different confidence requirements and obtain the elevation measurement results of the area.

[0093] Therefore, for the dual-antenna fully polarized SAR system of the lunar microwave imaging radar, the baseline advantage achieved by the existing dual antennas can be utilized, and the method of non-identical polarization (HH and VV) interference can be used in combination with multi-view smoothing to perform interferometric measurement of the lunar surface and acquire elevation information. Based on the existing external (coarse) DEM information and satellite information, the effect of the processing method can be pre-evaluated through the backscatter coefficient simulation method, and the processed results can also be roughly screened to select application scenarios with high reliability. This method can expand the application scope of the lunar microwave imaging radar and provide more possibilities for its scientific applications.

[0094] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An interference processing method in a dual-antenna fully polarized SAR system, characterized in that: The steps include: Step 1: simulate the backscatter coefficient by combining satellite system parameters and scattering characteristics of different types of ground objects, and calculate the difference of backscatter coefficients obtained under different polarization modes based on the simulated backscatter coefficient; Step 2: Obtain full polarization data with a baseline, select the co-polarization data pair with the smallest polarization characteristic difference according to the polarization characteristic difference of different channels, and determine the scene terrain slope range for interference processing according to the threshold of the backscatter coefficient difference; Step 3: For data within the range of the scene terrain slope, a multi-view optimization algorithm is used to reduce the polarization phase error of non-co-polarization interference, and a polarization phase similarity index is introduced to evaluate the interference performance and conduct application analysis; The step 2 includes selecting two channel data with the smallest difference in polarization characteristics in the full polarization data to form an interference pair based on the simulated backscattering coefficient, and considering the allowable backscattering coefficient difference as a threshold to determine the incident angle that meets the conditions. Range and the corresponding scene terrain slope range and incident angle The relationship with terrain information and satellite parameters is as follows: (3) in, represents the downward viewing angle of satellite observation, is the terrain slope; The step 3 includes determining the number of multi-looks for interference processing according to the resolution ratio of the image azimuth and range, reducing the polarization phase error of non-co-polarization interference, and evaluating the polarization phase similarity index at different incident angles. The confidence level of the interference processing results under the condition of defining the polarization phase similarity index between the horizontal co-polarization HH and the vertical co-polarization VV : (2) in, and represent the backscattering coefficient of horizontal co-polarization HH and vertical co-polarization VV respectively, The acceptable backscatter coefficient error for interferometric processing, polarization phase similarity index When the value is 0, it means that the polarization phase similarity between the horizontal co-polarization HH and the vertical co-polarization VV is 0, and no subsequent interference processing is performed; when the polarization phase similarity index When the value is 1, it means that there is no difference in polarization phase between horizontal co-polarization HH and vertical co-polarization VV.

2. The interference processing method in a dual-antenna fully polarized SAR system according to claim 1, characterized in that: The step 1 includes simulating the electromagnetic scattering coefficient of the rough surface of the lunar surface using a dual-scale model, wherein when the local incident angle is small, the Kirchhoff approximation method is used to calculate the backscattering coefficient; when the local incident angle is large, the perturbation method is used to calculate the backscattering coefficient.