A GBSAR image registration method based on differential aperture and error phase model

Abstract

This invention discloses a GBSAR image registration method based on differential aperture and error phase model, relating to the field of GBSAR deformation monitoring technology. The method includes the following steps: 1. Constructing primary and secondary full-aperture data matrices for the original GBSAR echo data obtained from two monitoring sessions; 2. Fixing the effective sub-aperture sequence in the primary data, and sliding to select sub-aperture sequences under different integer aperture offsets in the secondary data, determining the optimal aperture offset based on the maximum average coherence coefficient, and completing coarse registration at the integer array element level; 3. Based on the highly coherent and stable scattering points in the initial interferogram, fitting the error phase model caused by the residual offset of the guide rail, constructing a two-dimensional error phase screen, and performing pixel-by-pixel compensation. This invention, by combining differential aperture matching with error phase model compensation, achieves graded correction of aperture misalignment and residual phase errors caused by guide rail mechanical reset errors and trigger synchronization errors, improving the registration accuracy of GBSAR interferometric images and the reliability of micro-deformation inversion.

Description

Technical Field

[0001] This invention belongs to the field of ground-based synthetic aperture radar (GBSAR) deformation monitoring technology, specifically involving a GBSAR image registration method based on differential aperture and error phase model. It can be applied to the master-slave data interferometry processing of rail-mounted GBSAR, aiming to correct aperture misalignment and residual phase error caused by rail mechanical reset error and trigger synchronization error, thereby improving image registration accuracy and reliability of micro-deformation monitoring. Background Technology

[0002] Ground-based synthetic aperture radar interferometry (GBSAR-InSAR) technology offers advantages such as non-contact operation, high precision, and continuous monitoring, and has been widely applied to monitor the micro-deformation of targets such as slopes and bridges. Rail-mounted GBSAR typically requires strict overlap of the synthetic apertures from the primary and secondary observations to ensure that the interferometric phase primarily reflects the target's true deformation. However, in actual monitoring, factors such as mechanical reset errors of the rails, radar trigger synchronization errors, and minor platform vibrations can prevent the aperture positions from being perfectly aligned in the two acquisitions, leading to image mismatch and reduced coherence. For millimeter-wave GBSAR, due to the shorter wavelength, even small aperture offsets can be amplified into significant phase errors, severely impacting the accuracy of deformation inversion.

[0003] Existing GBSAR image registration methods mainly include registration methods based on image coherence coefficients, registration methods based on feature points, and phase compensation methods based on geometric error models. Among these, the coherence coefficient method achieves registration by searching for the maximum correlation position between the primary and secondary images; this method is simple and widely applicable. The feature point registration method establishes a spatial mapping relationship by extracting stable scattering points or image features. The geometric model method fits and compensates for the phase error introduced by baseline deviation based on the platform position error and the observation geometry. These methods can achieve certain results under conventional SAR or relatively stable GBSAR observation conditions.

[0004] However, existing methods still have shortcomings in high-precision monitoring of rail-guided millimeter-wave GBSAR. On the one hand, traditional image domain registration usually searches directly on the imaging results, failing to fully utilize the physical constraints of the virtual aperture element sequence during continuous motion acquisition, making it difficult to accurately distinguish integer element-level aperture misalignment caused by frame triggering and rail reset errors. On the other hand, when relying solely on the error phase model for compensation, if the primary and secondary apertures have not yet been coarsely aligned, the model parameters are easily affected by large-scale misalignment and low-coherence regions, leading to unstable compensation. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a GBSAR image registration method based on differential aperture and error phase models. This method aims to search for the optimal aperture offset between primary and secondary data and to perform model-based compensation for residual phase errors, thereby achieving graded correction of aperture misalignment and phase errors, and ultimately improving the registration accuracy of GBSAR interferometric images and the reliability of micro-deformation monitoring.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: The GBSAR image registration method based on differential aperture and error phase model of the present invention is characterized by the following steps: Step 1: The rail-mounted GBSAR system performs two observations of the monitored target, and uses the raw echo matrix obtained from the first observation as the master data. The original echo matrix obtained from the second observation was used as auxiliary data. ; Based on the radar frame acquisition sequence, guide rail movement trajectory, and aperture sampling interval, the following is performed: and Preprocessing is performed to obtain the main full aperture data matrix. Hefu full aperture data matrix ; Step 2: Based on the set effective sub-aperture length L In the main full aperture data matrix By selecting a fixed effective sub-aperture sequence, the main sub-aperture data matrix is ​​obtained. ; Based on the offset of different apertures, construct The offset is obtained from the sequence of sliding sub-apertures. Candidate auxiliary aperture data matrix ; , K This indicates the preset maximum search offset; Step 3: Perform master-slave aperture data matrix analysis. and candidate auxiliary aperture data matrix GBSAR imaging processing was performed separately to obtain the corresponding main images. and offset The auxiliary image below And form the k-th pair of primary and secondary images; Calculate the average coherence coefficient for each pair of primary and secondary images to obtain the average coherence coefficient corresponding to the offset k. The aperture offset corresponding to the maximum average coherence coefficient is determined as the optimal aperture offset. and will The corresponding candidate auxiliary aperture data matrix is ​​used as the auxiliary aperture data matrix after coarse registration. ; Step 4, based on and Generate initial interferometric phase diagram and in A set of highly coherent and stable scattering points was selected from the data. This is used to establish the residual error phase model and solve for the parameters of the residual error phase model; Step 5: Based on the solved residual error phase model, construct a two-dimensional error phase screen, and use the two-dimensional error phase screen to... Pixel-by-pixel compensation is performed to obtain the compensated interferometric phase map.

[0007] The GBSAR image registration method based on differential aperture and error phase model described in this invention is characterized in that the average coherence coefficient is calculated in step 3 according to the following process. ; Calculate the principal complex image using equation (1) and auxiliary images In the i 1 pixel Local coherence coefficient at : (1) In equation (1), Indicates the first i 1 pixel A local estimation window centered on the center, p This represents any pixel within the local estimation window; express medium pixel p Complex values ​​at that location, Indicates offset Down medium pixel p Complex values ​​at that location, express The complex conjugate; Calculate the offset using equation (2) k The corresponding average image coherence coefficient : (2) In equation (2), express or The number of pixels in the image.

[0008] Furthermore, step 4 includes the following steps: Step 4.1, for and GBSAR imaging processing was performed separately to obtain the coarsely registered master and slave images. And the auxiliary image after coarse registration ; Step 4.2, using equation (3) to... and Perform complex conjugate multiplication and take the argument to obtain the initial interferogram. : (3) In equation (3), express The complex conjugate; Step 4.3, Calculation and In the i 1 pixel Local coherence coefficient at ; Step 4.4: Set the coherence coefficient threshold. Thus, equation (4) is used to filter the set of highly coherent pixels. : (4) Step 4.5, from After removing anomalous phase points and points within potential deformation regions, a set of highly coherent and stable scattering points is obtained for model fitting. and obtained Corresponding initial interferometric phase observation vector ,in, Indicates the first j A highly coherent and stable scattering point. n This indicates the number of highly coherent and stable scattering points; express The initial interferometric phase observation vector, T Indicates transpose; Step 4.6, according to Spatial location, calculation corresponding azimuth angle and pitch angle Therefore, the residual error phase model is established using equation (5). : (5) In equation (5), This represents a constant phase error term. This represents the scaling factor related to the residual offset of the guide rail; Step 4.7: Use the least squares method to... Solve the equation to obtain the parameter estimates of the residual error phase model; Step 4.7.1: Construct the error phase model and calculate the matrix. ; Step 4.7.2: Use equation (6) to obtain the parameter estimates of the residual error phase model. ,in, This represents the estimated value of the constant phase error term. This represents an estimated value of the scaling factor associated with the residual offset of the guide rail; (6).

[0009] Furthermore, step 5 includes the following steps: Step 5.1: Calculate the first step using equation (7). i 1 pixel The residual error phase value at the location This allows us to obtain the error phase values ​​corresponding to all pixels and construct a two-dimensional error phase screen. ; (7) Step 5.2: Obtain the compensated interference phase diagram using equation (8). : (8).

[0010] The present invention provides an electronic device, including a memory and a processor, characterized in that the memory is used to store a program that supports the processor in performing the method described therein, and the processor is configured to execute the program stored in the memory.

[0011] The present invention discloses a computer-readable storage medium storing a computer program, characterized in that the computer program is executed by a processor to perform the steps of the method described thereon.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention effectively solves the problem of inaccurate aperture misalignment in rail-mounted GBSAR by introducing an aperture-domain differential matching mechanism. Existing methods typically perform correlation searches directly on the imaging image without considering matching using virtual aperture element sequences during continuous rail movement acquisition, making them susceptible to image noise, local deformation, and changes in scattering characteristics. To address this issue, this invention constructs primary and secondary full-aperture data matrices based on the rail movement trajectory and frame acquisition sequence. The primary data has a fixed effective sub-aperture sequence, while the secondary data slides to select sub-aperture sequences under different integer aperture offsets. The optimal aperture offset is determined by maximizing the average coherence coefficient. This method transforms traditional image-domain search into aperture-domain matching, enabling more accurate correction of integer element-level aperture misalignment caused by rail mechanical reset errors and trigger synchronization errors, thus improving the spatial consistency of the primary and secondary data.

[0013] 2. This invention achieves the correction of multi-scale guide rail offset errors through a differential aperture coarse registration combined with an error phase model fine registration mechanism. After achieving image alignment at the integer element scale, sub-element level residual offsets still exist, and these small errors can cause significant phase errors and spurious deformations in millimeter-wave GBSAR. Based on differential aperture coarse registration, this invention further filters out highly coherent and stable scattering points, excludes low-coherence regions and potential deformation regions, estimates parameters based on the residual error phase model, and constructs a two-dimensional error phase screen to compensate for the interference phase pixel by pixel. This hierarchical correction strategy can simultaneously handle large-scale aperture misalignment and small-scale residual phase errors, improving image registration accuracy while reducing the interference of offset errors on real micro-deformation signals, thus enhancing the reliability of GBSAR deformation monitoring results. Attached Figure Description

[0014] Figure 1 This is a flowchart of GBSAR image registration using the differential aperture and error phase model of the present invention. Figure 2 A simulated image of a point target; Figure 3 This is the deformation interference phase diagram without introducing guide rail offset; Figure 4 The interference phase diagram is shown when the offset is 2.2 mm. Figure 5 This is a curve showing the coherence coefficient of the differential aperture. Figure 6 Interference phase diagram after differential aperture matching; Figure 7 The final interferometric phase result diagram; Figure 8 A comparison chart of phase correction deviations under different guide rail offsets. Detailed Implementation

[0015] To verify the effectiveness of the GBSAR image registration method based on differential aperture and error phase model proposed in this invention, this embodiment constructs a W-band GBSAR point target simulation scenario and introduces guide rail aperture offset errors of different scales to verify the image registration method. In the simulation scenario, 100 point targets are randomly generated within a distance range of 8m to 12m in front of the radar to simulate stable scattering points in the GBSAR monitoring scenario.

[0016] Ten point targets in the left area of ​​the scene are selected as the simulated deformation area, and a slight displacement of the line of sight is set on them. The remaining point targets are used as static background targets. Figure 2 This embodiment demonstrates a point target simulation scenario. Figure 3 This is the ideal deformation interference phase diagram under conditions where guide rail offset is not introduced.

[0017] In this embodiment, a GBSAR image registration method based on a differential aperture and error phase model is a hierarchical registration method capable of simultaneously handling large-scale aperture misalignment and sub-element-level residual phase error, such as... Figure 1 As shown, the method includes the following steps: Step 1: Construct the main and auxiliary full aperture data matrix; Step 1.1: Acquire the raw echo data obtained from two observations using the rail-mounted GBSAR system. The raw echo data obtained from the first observation is taken as the master data and denoted as the master raw echo matrix. The raw echo data obtained from the second observation is used as auxiliary data, denoted as the auxiliary raw echo matrix. ; Step 1.2: Analyze the primary and secondary original echo matrices. and Preprocessing is performed, including data reading, abnormal frame removal, distance sampling point sorting, channel order correction, and complex echo matrix generation.

[0018] Step 1.3: Based on the radar frame acquisition sequence, assign each frame of echo data to the monitoring value of a virtual aperture element. Let the number of virtual aperture elements in the main data be... The number of virtual aperture elements in the auxiliary data is The number of sampling points is [number] Then the original echo matrix can be organized into a full aperture data matrix. and .

[0019] In this embodiment, two sets of raw echo simulation data corresponding to GBSAR monitoring are first generated. The first set of monitoring data is used as the master data, and the second set of monitoring data is used as the auxiliary data. 400 array data elements are selected from the master data as zero-error baseline data to construct... Similarly, generate the auxiliary data. To verify the adaptability of this method to guide rail offset errors of different scales, this embodiment sets up bidirectional guide rail offset sequences. (The mechanical offset of the guide rail is set during the simulation echo signal generation stage. That is, in the second monitoring after the target displacement is introduced, the array position is shifted as a whole to simulate the mechanical offset of the guide rail in actual monitoring); Among them, the data without the introduction of guide rail offset is used as the true value reference, and the data with the introduction of guide rail offset is used as the data to be registered.

[0020] Step 2: Construct a differential aperture sequence; Step 2.1: Set the effective sub-aperture length required for GBSAR imaging as... L Using equation (1) The fixed length is selected as LMaster aperture index set: (1) In equation (1), These represent the selected virtual aperture array element indices in the master data.

[0021] Step 2.2: Based on the master-slave aperture index set Using equation (2) from Select the master aperture data matrix: (2) Step 2.3: Set the integer aperture offset search range as follows. , K This indicates the preset maximum search offset; the search range can be determined based on the guide rail mechanical reset error range, radar frame sampling interval, and virtual aperture element spacing.

[0022] Step 2.4: For any integer aperture offset Using equation (3) Constructing a set of auxiliary sliding sub-aperture indexes: (3) Using equation (4) from Extract the corresponding offset The candidate auxiliary aperture data matrix is ​​as follows: (4) In this embodiment, the main full aperture data matrix The middle 300 consecutive aperture elements are selected as the effective imaging aperture. That is, from the 400 full-aperture array elements, the array element sequence with an index range of [51, 350] is selected as the fixed sub-aperture sequence of the main data, denoted as... Based on the above fixed sub-aperture sequence, from Extract the master aperture data matrix .

[0023] for A sequence of candidate auxiliary apertures is constructed using a sliding sub-aperture method. Let the integer aperture offset be... k Then the corresponding candidate sub-aperture index set in the auxiliary data is Since the total number of aperture elements in this embodiment is 400, and the selection range of the primary aperture is [51, 350], the secondary aperture must still satisfy the requirement that its index does not exceed the valid range of [1, 400] during the sliding process. For any offset k, from the secondary full aperture data matrix... Extract the corresponding candidate auxiliary aperture data matrix ; Step 3: Search for the optimal aperture offset; Step 3.1: Perform master-slave aperture data matrix analysis. and candidate auxiliary aperture data matrix Imaging was performed separately to obtain main and auxiliary image pairs. and ; Step 3.2: Using equation (5), let the set of pixels in the image be: (5) In equation (5), Indicates the first i 1 pixel N This indicates the number of pixels in the image.

[0024] Step 3.3: For any integer aperture offset In pixels Construct a local estimation window centered on The local coherence coefficient of the master-complex image pair at that pixel is calculated using equation (6): (6) In equation (6), Represented by pixels A local estimation window centered on the center, p denoted as any pixel within the window, and  denotes the complex conjugate operation. The local estimation window is used to perform statistical averaging using neighboring pixels to reduce the impact of speckle noise on the coherence coefficient estimation.

[0025] Step 3.4: Calculate the offset using equation (7). The image below shows the average coherence coefficient. : (7) Step 3.5: Search within the aperture offset range Within, the optimal aperture offset that maximizes the average coherence coefficient is determined using equation (8). : (8) In equation (8), arg max represents the independent variable that makes the objective function reach its maximum value.

[0026] Step 3.6: Based on the optimal aperture offset The auxiliary aperture data matrix after coarse registration is determined using equation (9): (9) Step 3 completes coarse registration of the master and slave data at the integer element level within the aperture domain, which is used to eliminate aperture misalignment at the integer multiple virtual aperture element scale caused by guide rail mechanical reset error or radar trigger synchronization error.

[0027] In this embodiment, the experimental group with a 2.2mm guide rail offset is taken as an example. Figure 4 The interferometric phase diagram after offset is introduced. It can be seen that due to the influence of the guide rail offset, obvious phase fringes appear in the interferogram, and the deformed area is masked by the phase of the systematic error. Figure 5 The coherence coefficient curve obtained from differential aperture scanning is used to determine the optimal aperture offset by searching for the maximum coherence coefficient. After coarse registration is performed based on this offset, the desired result is obtained. Figure 6 The interference phase diagram after aperture matching is shown. At this point, the dense phase fringes have been largely eliminated, but some phase fringes still exist, indicating that sub-element-level phase errors remain after differential aperture coarse registration. Step 4: Generate the initial interferogram and fit the residual error phase model; Step 4.1, for and GBSAR imaging processing was performed to obtain coarsely registered main and auxiliary images. and ; Step 4.2: Compare the coarsely registered master and slave images. and Perform complex conjugate multiplication, generate the initial interferogram using equation (10), and take the argument angle to obtain the initial interferometric phase diagram. : (10) In equation (10), express The complex conjugate, This indicates the operation of taking the argument angle.

[0028] Step 4.3: Based on the coarsely registered master and slave images and The coherence coefficient of the image pair is calculated using equation (6).

[0029] Step 4.4: Set the coherence coefficient threshold. The set of high coherence pixels in the image pair is selected using equation (11). : (11) In equation (11), Point The coherence coefficient value at that location.

[0030] Step 4.5: From the set of highly coherent pixels By removing anomalous phase points and pixels located within potential deformation regions, a set of highly coherent and stable scattering points for fitting the error phase model is obtained using Equation (12). : (12) In equation (12), Indicates the first j A highly coherent and stable scattering point. n This indicates the number of highly coherent and stable scattering points.

[0031] The initial interferometric phase observation vector corresponding to the highly coherent stable scattering point is obtained using equation (13). : (13) In equation (13), Indicates the first j A highly coherent and stable scattering point. n This indicates the number of highly coherent and stable scattering points; express The initial interferometric phase observation vector, T This indicates transpose.

[0032] Step 4.6, Set of highly coherent scattering points Any stable scattering point in Calculate the corresponding azimuth and elevation angles based on its spatial location, and denot them as follows: and Then, regarding the obtained... and Establish using equation (14) Error phase model caused by residual offset of the guide rail: (14) In equation (14), This represents a constant phase error term. This represents the scaling factor associated with the residual offset of the guide rail. This model is used to represent the residual phase error caused by the residual aperture offset at the subarray level after differential aperture coarse registration. Step 4.7: Use the least squares method to... Solve the equation to obtain the parameter estimates of the residual error phase model; Step 4.7.1: Construct the error phase model using equation (15) and calculate matrix H: (15) Let the model parameter vector be Then, using equation (16), the error phase model at the highly coherent stable scattering point can be written as: (16) In equation (16), This represents the model residual term.

[0033] Step 4.7.2: Solve for the model parameters using equation (17). When reversible, the estimated parameters of the residual error phase model. ,in, This represents the estimated value of the constant phase error term. The estimated value of the scaling factor associated with the residual offset of the guide rail is: (17) Step 4 allows for the fitting of the residual error phase model using the coarsely registered, highly coherent, stable scattering points, to obtain the residual phase caused by the subarray-level residual aperture shift.

[0034] In this embodiment, the main and auxiliary images generated based on the optimal aperture offset are paired. and The obtained initial interferogram is as follows Figure 6 As shown, then, highly coherent stable scattering points are selected from the initial interferometric phase diagram, and anomalous phase points and points in potential deformation regions are removed to obtain a set of highly coherent stable scattering points for model fitting. For any scattering point in the set, the error phase model caused by the residual migration shown in Equation (14) is established. The observation vector is constructed using the initial interferometric phase observations on the highly coherent stable scattering points, and the model parameters are solved using the least squares method to obtain the estimated values ​​of the residual error phase model parameters.

[0035] Step 5: Construct an error phase screen and compensate for the interference phase; Step 5.1: Perform initial interference phase diagram analysis. any pixel Using the residual error phase model parameter estimates obtained in step 4 The residual error phase value at the pixel is calculated using equation (18), and the initial interference phase map is then processed. The residual error phase value is calculated for each pixel in the array, and a two-dimensional error phase screen is constructed from the residual error phase values ​​corresponding to all pixels. .

[0036] (18) Step 5.2: Obtain the two-dimensional error phase screen From the initial interferometric phase diagram Pixel-by-pixel removal yields the compensated interferometric phase map: (19) In equation (19), This is the interferometric phase diagram after image registration is completed.

[0037] In this embodiment, a two-dimensional error phase screen is used. from Figure 6 The initial interferometric phase map is removed pixel by pixel to obtain the compensated interferometric phase map, as shown below. Figure 7 As shown, the phase of the background region tends to be flat after compensation.

[0038] Reference points can be selected P Quantitative analysis was performed at (−1.66, 8.01). The true deformation interference phase at this point was approximately -1.61 rad. After introducing a 2.2 mm guide rail offset, the phase at this point deviated to -3.06 rad; after differential aperture coarse registration, the phase was corrected to approximately -1.77 rad; after error phase model compensation, the final phase recovered to approximately -1.58 rad, with a deviation from the true value of approximately 0.03 rad.

[0039] Figure 8 The comparison results of phase correction deviations under different guide rail offsets are presented. The comparison results show that when only differential aperture coarse registration is used, sub-element level residual errors still exist due to the limitation of virtual element spacing; when only error phase model compensation is used, the model fitting stability is insufficient under large aperture misalignment conditions. This invention combines differential aperture coarse registration with error phase model compensation, achieving small phase correction deviations under different offsets, verifying the method's ability to hierarchically correct large-scale aperture misalignment and residual phase errors.

[0040] Through the above implementation methods, this invention first utilizes differential aperture matching to determine the optimal integer aperture offset of the primary and secondary data, and then fits a residual error phase model based on highly coherent stable scattering points and constructs a two-dimensional error phase screen for compensation. This method achieves hierarchical correction of aperture misalignment and residual phase error, which can improve the registration accuracy and deformation inversion reliability of rail-mounted millimeter-wave GBSAR images.

[0041] In this embodiment, an electronic device includes a memory and a processor. The memory stores a program that supports the processor in executing the above-described method, and the processor is configured to execute the program stored in the memory.

[0042] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the above method.

Claims

1. A GBSAR image registration method based on differential aperture and error phase model, characterized in that, Includes the following steps: Step 1: The rail-mounted GBSAR system performs two observations of the monitored target, and uses the raw echo matrix obtained from the first observation as the master data. The original echo matrix obtained from the second observation was used as auxiliary data. ; Based on the radar frame acquisition sequence, guide rail movement trajectory, and aperture sampling interval, the following is performed: and Preprocessing is performed to obtain the main full aperture data matrix. Hefu full aperture data matrix ; Step 2: Based on the set effective sub-aperture length L In the main full aperture data matrix By selecting a fixed effective sub-aperture sequence, the main sub-aperture data matrix is ​​obtained. ; Based on the offset of different apertures, construct The offset is obtained from the sequence of sliding sub-apertures. Candidate auxiliary aperture data matrix ; , K This indicates the preset maximum search offset; Step 3: Perform master-slave aperture data matrix analysis. and candidate auxiliary aperture data matrix GBSAR imaging processing was performed separately to obtain the corresponding main images. and offset The auxiliary image below And form the kth group of primary and secondary image pairs; Calculate the average coherence coefficient for each pair of primary and secondary images to obtain the average coherence coefficient corresponding to the offset k. The aperture offset corresponding to the maximum average coherence coefficient is determined as the optimal aperture offset. and will The corresponding candidate auxiliary aperture data matrix is ​​used as the auxiliary aperture data matrix after coarse registration. ; Step 4, based on and Generate initial interferometric phase diagram and in A set of highly coherent and stable scattering points was selected from the data. This is used to establish the residual error phase model and solve for the parameters of the residual error phase model; Step 5: Based on the solved residual error phase model, construct a two-dimensional error phase screen, and use the two-dimensional error phase screen to... Pixel-by-pixel compensation is performed to obtain the compensated interferometric phase map.

2. The GBSAR image registration method based on differential aperture and error phase model according to claim 1, characterized in that, Step 3 involves calculating the average coherence coefficient using the following procedure. ; Calculate the principal complex image using equation (1) and auxiliary images In the i 1 pixel Local coherence coefficient at : (1) In equation (1), Indicates the first i 1 pixel A local estimation window centered on the center, p This represents any pixel within the local estimation window; express medium pixel p Complex values ​​at that location, Indicates offset Down medium pixel p Complex values ​​at that location, express The complex conjugate; Calculate the offset using equation (2) k The corresponding average image coherence coefficient : (2) In equation (2), express or The number of pixels in the image.

3. The GBSAR image registration method based on differential aperture and error phase model according to claim 1, characterized in that, Step 4 includes the following steps: Step 4.1, for and GBSAR imaging processing was performed separately to obtain the coarsely registered master and slave images. And the auxiliary image after coarse registration ; Step 4.2, using equation (3) to... and Perform complex conjugate multiplication and take the argument to obtain the initial interferogram. : (3) In equation (3), express The complex conjugate; Step 4.3, Calculation and In the i 1 pixel Local coherence coefficient at ; Step 4.4: Set the coherence coefficient threshold. Thus, equation (4) is used to filter the set of highly coherent pixels. : (4) Step 4.5, from After removing anomalous phase points and points within potential deformation regions, a set of highly coherent and stable scattering points is obtained for model fitting. and obtain Corresponding initial interferometric phase observation vector ,in, Indicates the first j A highly coherent and stable scattering point. n This indicates the number of highly coherent and stable scattering points; express The initial interferometric phase observation vector, T Indicates transpose; Step 4.6, according to Spatial location, calculation corresponding azimuth angle and pitch angle Therefore, the residual error phase model is established using equation (5). : (5) In equation (5), This represents a constant phase error term. This represents the scaling factor related to the residual offset of the guide rail; Step 4.7: Use the least squares method to... Solve the equation to obtain the parameter estimates of the residual error phase model; Step 4.7.1: Construct the error phase model and calculate the matrix. ; Step 4.7.2: Use equation (6) to obtain the parameter estimates of the residual error phase model. ,in, This represents the estimated value of the constant phase error term. This represents an estimated value of the scaling factor associated with the residual offset of the guide rail; (6)。 4. The GBSAR image registration method based on differential aperture and error phase model according to claim 1, characterized in that, Step 5 includes the following steps: Step 5.1: Calculate the first step using equation (7). i 1 pixel The residual error phase value at the location This allows us to obtain the error phase values ​​corresponding to all pixels and construct a two-dimensional error phase screen. ; (7) Step 5.2: Obtain the compensated interference phase diagram using equation (8). : (8)。 5. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports a processor in executing the method of any one of claims 1-4, the processor being configured to execute the program stored in the memory.

6. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program is executed by the processor to perform the steps of the method according to any one of claims 1-4.

Images