InSAR deformation precise monitoring method based on triangular net guiding and priori regularization

Abstract

The application discloses a method for InSAR deformation precision monitoring based on triangulation guidance and prior regularization, obtains an interference phase diagram, carries out high-coherent pixel screening, constructs a Delaunay triangulation with elevation and distance constraints, obtains residual terrain estimation values and deformation rate estimation values of all high-coherent pixels, constructs an optimization function of an augmented Lagrange function, solves residual terrain, differential deformation rate, Lagrange multiplier vector and deformation rate, and obtains accurate digital elevation model, deformation rate and elevation and deformation rate information of high-coherent pixels. The application constructs a regularization inversion model based on full-scene InSAR high-coherent pixels and full-scene high-coherent pixel triangulation equation constraints, can realize full-scene optimization of inversion results, effectively suppresses spatial transmission of phase noise, and further improves InSAR high-coherent pixel three-dimensional reconstruction and deformation rate inversion precision.

Description

Technical Field

[0001] The invention relates to an InSAR deformation precision monitoring method based on triangulation guidance and prior regularization, and belongs to the technical field of interferometric synthetic aperture radar data processing. Background Art

[0002] Synthetic Aperture Radar (SAR) is an all-day, all-weather, active microwave imaging radar with the advantages of wide swath width, high spatiotemporal resolution, and a certain degree of surface penetration. Developed on the basis of SAR, interferometric synthetic aperture radar (InSAR) is a highly efficient and cutting-edge geodetic technology that can acquire continuous, high-precision surface elevation and deformation information. Persistent scatterers InSAR (PS-InSAR) and distributed scatterers InSAR (DS-InSAR) are among the most advanced time-series InSAR technologies. They are capable of high-precision topographic measurement and deformation monitoring of artificial targets such as urban buildings, roads and bridges, as well as natural scenes such as exposed soil. Their height measurement accuracy can reach sub-meter levels, and their deformation monitoring accuracy can reach millimeters per year. They play a vital role in elevation measurement and precise ground subsidence measurement in cities, coastal areas, islands, and reefs.

[0003] After identifying robust, highly coherent targets, PS-InSAR and DS-InSAR technologies construct a Delaunay triangulation. The maximum set coherence criterion is used to calculate the differential residual elevation and differential deformation rate of connected triangle edges edge by edge. The residual elevation and deformation rate are then solved using a least-squares method weighted by the set coherence coefficient. However, the errors in the differential residual elevation and deformation rate calculated edge by edge can propagate spatially, resulting in a wide distribution range for the final solved residual elevation and deformation rate. Therefore, traditional edge-by-edge calculation followed by spatial resolution methods struggle to accurately invert the elevation terrain and deformation rate fields. A regularization model based on a spatial correlation prior for the deformation rate field offers the potential for suppressing spatially propagated noise. Therefore, a full-scene prior regularization model guided by a triangulation network can be constructed to suppress the spatial propagation effects of incoherent random phase noise, enabling a one-step, accurate inversion of the four-dimensional information of InSAR high-coherence pixels. Summary of the Invention

[0004] The purpose of this invention is to address the problem of large fluctuations in the difference between the elevation and deformation rate of adjacent scatterers caused by time-series InSAR decoherence noise. A triangulated network guided and prior regularized InSAR precision deformation monitoring method is proposed to effectively suppress the influence of random noise in low-coherence pixels on the estimation results, so as to achieve the purpose of accurately reconstructing three-dimensional terrain and deformation rate.

[0005] The above-mentioned purpose of the present invention is achieved by the following technical means:

[0006] The triangulated network guided and prior regularized InSAR deformation precision monitoring method includes the following steps:

[0007] Step 1: Acquire an interferometric phase image sequence, and use the reference topographic data to dereference the interferometric phase image sequence to obtain an interferometric phase map;

[0008] Step 2: Screening high coherence pixels in the interference phase image;

[0009] Step 3: Construct a Delaunay triangulation with elevation and distance constraints;

[0010] Step 4: Solve the maximum set coherence model to obtain the estimated value of the differential elevation between any connected high coherence pixels Estimation of differential deformation rate and the set coherence coefficient

[0011] Step 5: Use the weighted least squares method to obtain the residual terrain estimation value of all high coherence pixels and deformation rate estimates

[0012] Step 6: Construct the optimization function of the augmented Lagrangian function

[0013] Step 7: Use the alternating direction multiplier method to optimize and update the residual terrain h in the regularized inversion model of the full scene high coherence pixel triangulation equality constraint ε , differential deformation rate Lagrange multiplier vector y, and deformation rate v;

[0014] Step 8: Use the iteratively optimized residual terrain h obtained in step 7 ε and deformation rate v, as well as reference terrain data, to obtain an accurate digital elevation model (DEM), and obtain the elevation and deformation rate information of highly coherent pixels such as deformation rate.

[0015] As described above, the elevation and distance constraints in step 3 are: the connected edge of the elevation difference of adjacent high-coherence pixels is less than the corresponding set threshold, and the connected edge distance of adjacent high-coherence pixels is less than the corresponding set threshold.

[0016] The maximization set coherence model in step 4 above is based on the following formula:

[0017]

[0018] in, is the set coherence coefficient, is the difference operator between two adjacent pixels x1 and x2, where x1 and x2 refer to two adjacent pixels. represents the differential interference phase of the xth high coherence pixel in the i-th interference phase image generated in step 1, and They represent the residual topography and linear deformation interference phase of the x-th high coherence pixel in the i-th interference phase image generated in step 1, M represents the number of interference phase images, j is the imaginary unit, and The difference form of and for:

[0019]

[0020] Among them, b i With t i denote the vertical baseline and observation time of the i-th interferometric phase pattern, θ and λ denote the radar incident angle and wavelength, respectively. ε and v represent the residual terrain and deformation rate of the target respectively, and r is the distance from the satellite to the target.

[0021] In step 5 above: use the estimated value of the differential elevation of all connected high coherence pixels obtained in step 4 Arrange rows to form differential elevations The estimated differential deformation rate of all connected pixels obtained in step 4 Differential deformation rates are formed by row arrangement The coherence coefficient As the weight, the weighted least squares method is used to obtain the residual terrain estimation value of all high coherence pixels based on the following formula and deformation rate estimates

[0022]

[0023] in, A matrix representing the mapping of highly coherent pixels to connected pixel edges.

[0024] As mentioned above, the Lagrangian function in step 6 is augmented Based on the following formula:

[0025]

[0026] where W is a diagonal matrix formed by the set coherence coefficients of step 4 connected high-coherence pixels, the. behind W means the calculation of elements between the same size matrices, exp is the exponential operator, v is the deformation rate, h ε is the residual topography, diag{} means the vector diagonalization operator, Ψ represents the matrix formed by the differential interferometric phases of all high-coherence connected pixels, B represents the matrix formed by the baseline related beam domain elements of all high-coherence pixels, the baseline related beam domain element of high-coherence pixels is c represents the vector formed by the time related beam domain elements of all high-coherence pixels, the time related beam domain element of high-coherence pixels is A represents the triangular net mapping matrix formed by high-coherence connected pixels, η represents the regularization parameter, is the square of F-norm, ||1 is the 1-norm, μ represents the penalty parameter of constraint, y represents the Lagrange multiplier vector, <·,·> represents the inner product operation of vectors, represents the square of 2-norm.

[0027] The regularization inversion model of step 7 for updating the full-scene high-coherence pixel triangular net equation constraint by using the alternating direction multiplier method as described above comprises the following steps:

[0028] Step 7.1, taking the residual topography estimate value obtained in step 5 as the initial value of the residual topography h ε , taking the differential deformation rate obtained in step 5 as the initial value of the differential deformation rate , taking the deformation rate estimate value estimated in step 5 as the initial value of the deformation rate v, and taking 0 as the initial value of the Lagrange multiplier vector y;

[0029] Step 7.2, inputting the differential deformation rate obtained by the last iteration optimization and the Lagrange multiplier vector y into the following formula to solve the updated value of the residual topography estimate value

[0030]

[0031] and further obtaining the updated value of the estimated residual topography ;

[0032] Step 7.3, taking the newly obtained estimated residual topography as the updated residual topography h ε ​, together with the deformation rate v and Lagrange multiplier vector y obtained from the previous iterative optimization, are substituted into the following formula to solve the estimated value of the differential deformation rate: Updated value of:

[0033]

[0034] Step 7.4: Substitute the estimated differential deformation rate obtained in step 7.3 into The updated value of is used as the differential deformation rate of the current iteration The differential deformation rate of the current iteration And the Lagrange multiplier vector y obtained from the previous iterative optimization is substituted into the following formula to obtain the updated value of the Lagrange multiplier vector As the Lagrange multiplier vector y for the current iteration optimization:

[0035]

[0036] Step 7.5: Substitute the latest estimate of the differential deformation rate obtained in step 7.3 into As the differential deformation rate of the current iteration optimization The differential deformation rate of the current iteration optimization Substitute the following formula and use the least squares method to solve the deformation rate The updated value of is used as the deformation rate v of the current iterative optimization.

[0037]

[0038] Step 7.6, iterative optimization executes steps 7.2-7.5 until the maximum number of iterations is reached or the residual topography h obtained from the two iterations is reached. ε , deformation rate v, Lagrange multiplier vector y, differential deformation rate The difference between the two iterations is less than the corresponding threshold, and the residual terrain h ε , Lagrange multiplier vector y, differential deformation rate Substitute the augmented Lagrangian function If the values ​​of are all less than the corresponding set threshold, the iteration is stopped and the residual terrain h after iterative optimization is output. ε and deformation rate v.

[0039] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned InSAR deformation precision monitoring method when executing the computer program.

[0040] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned InSAR deformation precision monitoring method.

[0041] A computer program product includes a computer program, which implements the steps of the above-mentioned InSAR deformation precision monitoring method when executed by a processor.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] 1. Accuracy. This invention creatively constructs a regularized inversion model based on full-scene high-coherence InSAR pixels and a triangulated network equality constraint for full-scene high-coherence pixels. This model can achieve full-scene optimal inversion results and effectively suppress the spatial transmission of phase noise, thereby improving the accuracy of InSAR high-coherence pixel 3D reconstruction and deformation rate inversion.

[0044] 2. Effectiveness. The method provided by this invention has been compared with traditional methods using measured satellite data and analyzed for robustness of key parameters, verifying its convergence, effectiveness, and superiority, demonstrating its good application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Flowchart of the present invention.

[0046] Figure 2 The four-dimensional information inversion results of InSAR high-coherence pixels are shown in Figure 2. (a) is a schematic diagram of the DEM obtained by traditional edge-by-edge differencing followed by weighted least squares inversion; (b) is a schematic diagram of the deformation rate estimate obtained by traditional edge-by-edge differencing followed by weighted least squares inversion; (c) is a schematic diagram of the DEM inverted based on Example 2 of the present invention; and (d) is a schematic diagram of the deformation rate estimate inverted based on Example 2 of the present invention.

[0047] Figure 3 The uncertainty results of the present invention are shown in Figure 2. (a) is the uncertainty diagram of the elevation estimation of Example 2 of the present invention, and (b) is the uncertainty diagram of the deformation rate estimation of Example 2 of the present invention.

[0048] Figure 4 The following are the robustness analysis and runtime analysis diagrams of the present invention. (a) The augmented Lagrange function shows the linear variation of the residual terrain and differential deformation rate with the number of iterations; (b) The augmented Lagrange function shows the linear variation of the runtime with the number of iterations. DETAILED DESCRIPTION

[0049] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below with reference to the embodiments. The embodiments described herein are only used to illustrate and explain the present invention and are not intended to limit the present invention.

[0050] Example 1:

[0051] like Figure 1 As shown in FIG, the triangulated network guided and prior regularized InSAR deformation precision monitoring method includes the following steps:

[0052] Step 1: Perform image registration on the multi-temporal SAR images to obtain precisely registered primary and secondary phase images, which are then conjugate multiplied to generate an interferometric phase image sequence. The interferometric phase image sequence is dereferenced to the terrain phase using coarse-precision reference terrain data, thereby obtaining differential interferometric stack data of the residual terrain and surface deformation components, i.e., the interferometric phase map.

[0053] Step 2: Using high coherence pixel screening criteria such as amplitude deviation and coherence threshold, the interference phase image is screened to robustly identify unevenly and sparsely distributed high coherence pixels;

[0054] Step 3: Using the unevenly and sparsely distributed high-coherence pixels in the identified scene, a Delaunay triangulation with elevation and distance constraints is constructed. The constraint criteria are: the elevation difference between adjacent high-coherence pixels must be less than 50 meters, and the distance between adjacent high-coherence pixels must be less than 1 kilometer.

[0055] Step 4: Perform phase difference on the highly coherent pixels connected by the Delaunay triangulation network, and obtain the set coherence coefficient of adjacent highly coherent pixels based on the maximum set coherence model. Estimated differential elevation by

[0056] and the estimated value of the differential deformation rate The maximum set coherence model is based on the following formula:

[0057]

[0058] In formula (1), is the set coherence coefficient, is the difference operator between two adjacent pixels x1 and x2, where x1 and x2 refer to two adjacent pixels. represents the differential interference phase of the xth high coherence pixel in the i-th interference phase image generated in step 1, then They are and They represent the residual topography and linear deformation interference phase of the x-th high coherence pixel in the i-th interference phase image generated in step 1, M represents the number of interference phase images, j is the imaginary unit, and The difference form of and It can be expressed as:

[0059]

[0060] Among them, b i With t i denote the vertical baseline and observation time of the i-th interferometric phase pattern, θ and λ denote the radar incident angle and wavelength, respectively. ε and v represent the residual terrain and deformation rate of the target respectively, and r is the distance from the satellite to the target. The parameters of the maximum set coherence model can be quickly solved using the grid search method under the CPU parallel and GPU computing strategy, and then the estimated value of the differential elevation between any connected high coherence pixels is obtained. Estimation of differential deformation rate and the set coherence coefficient

[0061] Step 5: Use the estimated values ​​of the differential elevation of all connected high-coherence pixels obtained in step 4 Arrange rows to form differential elevations The estimated differential deformation rate of all connected pixels obtained in step 4 Differential deformation rates are formed by row arrangement The coherence coefficient As the weight, the weighted least squares method of formula (3) is used to obtain the residual terrain estimation value of all high coherence pixels. and deformation rate estimates

[0062]

[0063] in, Represents the mapping matrix from high coherence pixels to connected pixel edges. The residual terrain estimation value of all high coherence pixels can be obtained by formula (3): and deformation rate estimates

[0064] Step 6: Based on the prior information of the spatial correlation of the deformation rates of adjacent pixels, a regularized inversion model with equality constraints for the triangulated network of highly coherent pixels in the entire scene is constructed:

[0065]

[0066] Where W is the diagonal matrix formed by the set coherence coefficients calculated by the connected high coherence pixels in step 4. The * after W indicates the calculation of elements between matrices of the same size. exp is the exponential operator, v is the deformation rate, and h is the εis the residual terrain, diag{} represents the vector diagonalization operator, Ψ represents the matrix composed of the differential interference phase of all highly coherent connected pixels, Represents the mapping matrix from highly coherent pixels to connected pixel edges Respectively with deformation rate v, residual topography h ε , matrix Ψ performs matrix multiplication operation, B represents all high coherence pixel baseline correlation beam domain elements (i.e. ), c represents the time-correlated beam domain elements of all high-coherence pixels (i.e., ), A represents the triangulated network mapping matrix composed of highly coherent connected pixels, η represents the regularization parameter, is the square of the F norm,

[0067] 1 is the 1 norm. The equality constraint optimization problem (4) can be transformed into the augmented Lagrange function The optimization function

[0068]

[0069] Among them, μ represents the penalty parameter of the constraint, y represents the Lagrange multiplier vector, and <·,·> represents the inner product operation of the vector. represents the 2-norm squared.

[0070] Step 7: The residual terrain estimate obtained in step 5 is As the residual terrain h ε The initial value of the differential deformation rate obtained in step 5 As the differential deformation rate The initial value of the deformation rate estimated in step 5 is As the initial value of the deformation rate v, 0 is used as the initial value of the Lagrange multiplier vector y, and then the alternating direction multiplier method is used to optimize and update the residual terrain h in the regularized inversion model of the full scene high coherence pixel triangulation equality constraint. ε , differential deformation rate Lagrange multiplier vector y, and deformation rate v.

[0071] In step 7 above, the alternating direction multiplier method is used to optimize and update the residual terrain h in the regularized model guided by the triangulation network. ε , differential deformation rate The Lagrange multiplier vector y and the deformation rate v parameter are as follows:

[0072] Step 7.1: The residual terrain estimate obtained in step 5 is As the residual terrain h εThe initial value of the differential deformation rate obtained in step 5 As the differential deformation rate The initial value of the deformation rate estimated in step 5 is As the initial value of the deformation rate v, 0 is used as the initial value of the Lagrange multiplier vector y and input into the optimization problem of the following steps 7.2 and 7.3 to start the first parameter iterative optimization.

[0073] Step 7.2: The differential deformation rate obtained from the previous iterative optimization is The Lagrange multiplier vector y is used as the input to the optimization problem (6) to solve the estimated value of the residual terrain Updated value of:

[0074]

[0075] Then we get the estimated residual terrain The updated value of .

[0076] Step 7.3: The newly obtained estimated residual terrain As the residual terrain h updated in the current iteration ε , the deformation rate v and Lagrange multiplier vector y obtained in the previous iterative optimization are used as the input of the optimization problem of formula (7) to solve the estimated value of the differential deformation rate Updated value of:

[0077]

[0078] Then we can get the estimated value of differential deformation rate The updated value of .

[0079] Step 7.4: Substitute the estimated differential deformation rate obtained in step 7.3 into The updated value of is used as the differential deformation rate of the current iteration The differential deformation rate of the current iteration And the Lagrange multiplier vector y obtained in the previous iterative optimization is used as the input of formula (8) to obtain the updated value of the Lagrange multiplier vector As the Lagrange multiplier vector y for the current iteration optimization:

[0080]

[0081] Step 7.5: Substitute the latest estimate of the differential deformation rate obtained in step 7.3 into As the differential deformation rate of the current iteration optimization The differential deformation rate of the current iteration optimization As the input of formula (9), the least square method is used to solve the updated value of the deformation rate as the deformation rate v of the current iteration optimization.

[0082]

[0083] Step 7.6, iteration optimization is performed on steps 7.2-7.5 until the maximum number of iterations is reached or the residual terrain h ε , the deformation rate v, the Lagrange multiplier vector y, the differential deformation rate obtained by the previous two iterations are all less than the corresponding set threshold, and the value of the augmented Lagrange function ε , the Lagrange multiplier vector y, the differential deformation rate calculated by formula (5) are all less than the corresponding set threshold, then the iteration is stopped and the residual terrain h ε and the deformation rate v of the iteration optimization are output.

[0084] Step 8, using the residual terrain h ε and the deformation rate v of the iteration optimization obtained in step 7, and adding the coarse-precision reference terrain data input in step 1, the accurate height and deformation rate information of the high-coherent pixels are obtained.

[0085] In one embodiment, a computer device is also provided, including a memory and a processor, the memory stores a computer program, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0086] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in the above-mentioned method embodiments.

[0087] In one embodiment, a computer program product is provided, which includes a computer program, and the computer program is executed by a processor to implement the steps in the above-mentioned method embodiments.

[0088] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned method embodiments.

[0089] Embodiment 2:

[0090] ​The effectiveness of the triangulated network-guided and a priori regularized InSAR deformation precision monitoring method described in Example 1 was demonstrated using 31 real-world data images from the Beijing Capital International Airport area acquired using TerraSAR-X / TanDEM-X. The acquisition time and baseline parameter information for the 31 TerraSAR-X / TanDEM-X images are shown in Table 1.

[0091] Table 1

[0092]

[0093]

[0094] Under the same PS-InSAR data processing flow, the processing results of the traditional edge-by-edge difference calculation followed by weighted least squares method are compared with the results of the triangulated network guidance and prior regularized full-scene inversion method of the present invention, illustrating the advantages of the method proposed in the present invention. Figure 2 The four-dimensional information inversion results of InSAR high-coherence pixels are shown in Figure 1. (a) is a schematic diagram of the traditional DEM using edge-by-edge differential calculation followed by weighted least squares inversion, (b) is a schematic diagram of the deformation rate estimation using the traditional edge-by-edge differential calculation followed by weighted least squares inversion; (c) is a schematic diagram of the DEM inverted based on Example 2 of the present invention, and (d) is a schematic diagram of the deformation rate estimation based on Example 2 of the present invention. The triangulated network-guided and prior-regularized full-scene inversion method of the present invention has a smaller elevation and deformation rate distribution range, indicating that spatial error propagation is effectively suppressed. Figure 3 Figures 2 and 3 show the uncertainty results of the present invention. (a) shows the uncertainty diagram for elevation estimation in Example 2 of the present invention, and (b) shows the uncertainty diagram for deformation rate estimation in Example 2 of the present invention. (a) shows that 99% of the uncertainty for elevation estimation using the present invention's method is within 1 meter, and (b) shows that the uncertainty for deformation rate estimation using the present invention is all within 1 mm / year, demonstrating the effectiveness of the present invention. Figure 4 The following are the robustness analysis and runtime analysis of the number of iterations of the present invention. (a) The augmented Lagrange function with respect to the residual terrain and differential deformation rate varies linearly with the number of iterations; (b) The augmented Lagrange function with respect to the runtime varies linearly with the number of iterations. It can be seen that as the number of iterations increases, the augmented Lagrange function with respect to the differential residual elevation and differential deformation rate first slows down and then remains constant, while the runtime varies linearly with the number of iterations. This demonstrates the iterative convergence of the triangulated network-guided and a priori regularized InSAR deformation precision monitoring method, with rapid convergence achieved after only two iterations.

[0095] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A triangulated network guided and prior regularized InSAR deformation precision monitoring method, characterized by: The following steps are involved: Step 1: Acquire an interferometric phase image sequence, and use the reference topographic data to dereference the interferometric phase image sequence to obtain an interferometric phase map; Step 2: Screening high coherence pixels in the interference phase image; Step 3: Construct a Delaunay triangulation with elevation and distance constraints; Step 4: Solve the maximum set coherence model to obtain the estimated value of the differential elevation between any connected high coherence pixels Estimation of differential deformation rate and the set coherence coefficient Step 5: Use the weighted least squares method to obtain the residual terrain estimation value of all high coherence pixels and deformation rate estimates Step 6: Construct the optimization function of the augmented Lagrangian function Step 7: Use the alternating direction multiplier method to optimize and update the residual terrain h in the regularized inversion model of the full scene high coherence pixel triangulation equality constraint ε , differential deformation rate Lagrange multiplier vector y, and deformation rate v; Step 8: Use the iteratively optimized residual terrain h obtained in step 7 ε and deformation rate v, as well as reference terrain data, to obtain an accurate digital elevation model (DEM), and obtain the elevation and deformation rate information of highly coherent pixels such as deformation rate. The augmented Lagrangian function in step 6 Based on the following formula: Where W is the diagonal matrix formed by the set coherence coefficients calculated by the connected high coherence pixels in step 4. The * after W indicates the calculation of elements between matrices of the same size. exp is the exponential operator, v is the deformation rate, and h is the ε is the residual terrain, diag{} represents the vector diagonalization operator, Ψ represents the matrix composed of the differential interference phase of all high coherent connected pixels, B represents the matrix composed of the elements of the baseline correlation beam domain of all high coherent pixels, and the elements of the baseline correlation beam domain of high coherent pixels are c represents the vector composed of all high coherence pixel time correlation beam domain elements. The high coherence pixel time correlation beam domain elements are A represents the triangulated network mapping matrix composed of highly coherent connected pixels, η represents the regularization parameter, is the square of the F norm, ||1 is the 1 norm, μ represents the penalty parameter of the constraint, y represents the Lagrange multiplier vector, <·,·> represents the inner product operation of the vector, represents the 2-norm square, j is the imaginary unit, A matrix representing the mapping of highly coherent pixels to connected pixel edges.

2. The triangulated network guided and prior regularized InSAR deformation precision monitoring method according to claim 1, characterized in that: The elevation and distance constraints in step 3 are: the connected edge of the elevation difference of adjacent high-coherence pixels is less than the corresponding set threshold, and the connected edge distance of adjacent high-coherence pixels is less than the corresponding set threshold.

3. The triangulated network guided and prior regularized InSAR deformation precision monitoring method according to claim 1, characterized in that: The maximization set coherence model in step 4 is based on the following formula: in, is the set coherence coefficient, is the difference operator between two adjacent pixels x1 and x2, where x1 and x2 refer to two adjacent pixels. represents the differential interference phase of the xth high coherence pixel in the i-th interference phase image generated in step 1, and They represent the residual topography and linear deformation interference phase of the x-th high coherence pixel in the i-th interference phase image generated in step 1, M represents the number of interference phase images, j is the imaginary unit, and The difference form of and for: Among them, b i With t i denote the vertical baseline and observation time of the i-th interferometric phase pattern, θ and λ denote the radar incident angle and wavelength, respectively. ε and v represent the residual terrain and deformation rate of the target respectively, and r is the distance from the satellite to the target.

4. The triangulated network guided and prior regularized InSAR deformation precision monitoring method according to claim 3 is characterized in that: In step 5: the estimated value of the differential elevation of all connected high coherence pixels obtained in step 4 is used Arrange rows to form differential elevations The estimated differential deformation rate of all connected pixels obtained in step 4 Differential deformation rates are formed by row arrangement The coherence coefficient As the weight, the weighted least squares method is used to obtain the residual terrain estimation value of all high coherence pixels based on the following formula and deformation rate estimates in, A matrix representing the mapping of highly coherent pixels to connected pixel edges.

5. The triangulated network guided and prior regularized InSAR deformation precision monitoring method according to claim 4, characterized in that: In step 7, the regularized inversion model for optimizing and updating the equality constraints of the full-scene high-coherence pixel triangulation network using the alternating direction multiplier method includes the following steps: Step 7.1: The residual terrain estimate obtained in step 5 is As the residual terrain h ε The initial value of the differential deformation rate obtained in step 5 As the differential deformation rate The initial value of the deformation rate estimated in step 5 is As the initial value of the deformation rate v, 0 is used as the initial value of the Lagrange multiplier vector y; Step 7.2: The differential deformation rate obtained from the previous iterative optimization is The Lagrange multiplier vector y is input to the following formula to solve for the estimated residual terrain Updated value of: Then we get the estimated residual terrain The updated value of Step 7.3: The newly obtained estimated residual terrain As the residual terrain h updated in the current iteration ε , together with the deformation rate v and Lagrange multiplier vector y obtained from the previous iterative optimization, are substituted into the following formula to solve the estimated value of the differential deformation rate: Updated value of: Step 7.4: Substitute the estimated differential deformation rate obtained in step 7.3 into The updated value of is used as the differential deformation rate of the current iteration The differential deformation rate of the current iteration And the Lagrange multiplier vector y obtained from the previous iterative optimization is substituted into the following formula to obtain the updated value of the Lagrange multiplier vector As the Lagrange multiplier vector y for the current iteration optimization: Step 7.5: Substitute the latest estimate of the differential deformation rate obtained in step 7.3 into As the differential deformation rate of the current iteration optimization The differential deformation rate of the current iteration optimization Substitute the following formula and use the least squares method to solve the deformation rate The updated value of is used as the deformation rate v of the current iterative optimization. Step 7.6, iterative optimization executes steps 7.2-7.5 until the maximum number of iterations is reached or the residual topography h obtained from the two iterations is reached. ε , deformation rate v, Lagrange multiplier vector y, differential deformation rate The difference between the two iterations is less than the corresponding threshold, and the residual terrain h ε , Lagrange multiplier vector y, differential deformation rate Substitute the augmented Lagrangian function If the values ​​of are all less than the corresponding set threshold, the iteration is stopped and the residual terrain h after iterative optimization is output. ε and deformation rate v.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the InSAR deformation precision monitoring method according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the InSAR deformation precision monitoring method according to any one of claims 1 to 5 are implemented.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the InSAR deformation precision monitoring method according to any one of claims 1 to 5 are implemented.

Images