An iterative unwrapping method for interferometric phase based on two-dimensional Gaussian model

Through the iterative unwrap method of the two-dimensional Gaussian model, the error problem of traditional phase unwrap in large gradient deformation monitoring of mining areas is solved, and high-precision phase unwrap in mining areas is achieved, which reduces the limitations of the use of the method and expands the application prospects.

CN115113202BActive Publication Date: 2024-12-03CENT SOUTH UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210589407.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-18
Publication Date
2024-12-03
Estimated Expiration
2042-07-18

AI Technical Summary

Technical Problem

Traditional phase disintegration methods are difficult to deal with interference graph errors caused by large gradient deformation in InSAR deformation monitoring in mining areas, and require external data or a large number of training samples, which limits the application scope and accuracy of the method.

Method used

Using the iterative interferometric phase unwrap method based on the two-dimensional Gaussian model, the data preprocessing, adaptive quadtree downsampling, the two-dimensional Gaussian function model is established and inverted, and the unwrap phase of the mining area is iteratively calculated, thereby avoiding underground mining. Dependence of parameters and training samples.

Benefits of technology

It achieves high-precision phase detangling results in the mining area without the need for underground mining parameters and training samples, overcomes the problem that traditional methods are difficult to untangle dense interference fringes, and broadens the application prospects of InSAR phase detangling in the mining area.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115113202B_ABST
    Figure CN115113202B_ABST
Patent Text Reader

Abstract

The invention provides an interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model, which is applied to InSAR monitoring of large gradient deformation in mining areas, comprising: the method is applied to InSAR monitoring of large gradient deformation in mining areas, comprising: step 1, collecting SAR images for data preprocessing, obtaining single-view complex SLC images, and obtaining an interferometric pattern by forming an interference combination, and calculating the interferometric pattern phase; step 2, performing adaptive quadtree downsampling processing on the interferometric pattern phase, and using a traditional phase unwrapping method to calculate the unwrapped phase at the edge of the mining area; step 3, establishing a two-dimensional Gaussian function model of the unwrapped phase at the edge of the mining area, calculating the model parameters and inverting the model parameters; step 4, obtaining the entangled residual interference phase by estimating the whole basin deformation phase of the mining area, and unwrapping the residual interference phase; step 5, obtaining the unwrapped phase of the coal mining area; step 6, obtaining the refined unwrapped phase of the mining area by iterating steps 3-5.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of mining area surface deformation monitoring, and in particular to an interference phase iterative unwrapping method based on a two-dimensional Gaussian model. Background Art

[0002] Mineral resources are an important material condition that is indispensable for human survival and a basic industry related to the national economy and people's livelihood. With the continuous exploitation of mineral resources in recent years, surface deformation may cause a series of potential disasters, such as surface collapse, house tilt, road damage, environmental pollution, etc. InSAR (Interferometric synthetic aperture radar) technology is a large-scale, high-temporal and spatial resolution ground monitoring technology developed in recent years, which can monitor the surface deformation caused by mining. A key step in InSAR deformation monitoring is phase unwrapping, the purpose of which is to recover the true deformation phase from the entangled phase. Traditional phase unwrapping methods, including path tracking method, optimization-based method and joint denoising and unwrapping method, all solve the ill-conditioned inversion problem based on the phase continuity assumption, while rapid mining in mining areas will lead to excessive deformation gradients and discontinuous phases. Therefore, the traditional phase unwrapping method is prone to errors in solving the entangled interferogram of mining areas, especially in areas with large deformation gradients.

[0003] In order to solve the application problem of traditional phase unwrapping methods in InSAR deformation monitoring in mining areas, the commonly used methods can be roughly divided into three categories: 1. Phase unwrapping assisted by external data such as ground GPS monitoring data and leveling data; 2. Phase unwrapping method based on deep learning; 3. Phase unwrapping method assisted by probability integral method model. Among them, the first type of method is difficult to promote and apply on a large scale due to the limitation of external data availability, while the second type of method relies on a large number of training samples and sample reliability, and it is difficult to ensure the accuracy of the solution results. The third type of method requires the field collection of geometric parameters of underground goafs.

[0004] Therefore, the InSAR phase unwrapping of large gradient deformation in mining areas (such as underground mining areas) is still very difficult, which seriously restricts the application prospects of InSAR technology in the mining field. Summary of the invention

[0005] The present invention provides an iterative interference phase unwrapping method based on a two-dimensional Gaussian model, the purpose of which is to overcome the limitation that the interference pattern fringes are dense and the aliasing is difficult to unwrap due to rapid mining in the mining area, and accurate phase unwrapping results of the mining area can be obtained without the need for underground mining parameters, thereby improving the accuracy of phase unwrapping in the mining area.

[0006] In order to achieve the above object, the present invention provides an interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model, which is applied to InSAR monitoring of large gradient deformation in mining areas, comprising:

[0007] Step 1, collect SAR images for data preprocessing, obtain single-view complex SLC images, and obtain the entangled interferogram by forming an interference combination, and calculate the interferogram phase;

[0008] Step 2, performing adaptive quadtree downsampling processing on the interference pattern phase, and using the traditional phase unwrapping method to calculate the unwrapped phase at the edge of the mining area;

[0009] Step 3, establishing a two-dimensional Gaussian function model of the unwrapped phase at the edge of the mining area, calculating the model parameters and inverting the model parameters;

[0010] Step 4, obtaining the entangled residual interference phase by estimating the basin-wide deformation phase of the mining area, and untangling the residual interference phase;

[0011] Step 5, obtaining the unwrapped phase of the coal mining area;

[0012] Step 6, obtain the refined mineral area unwrapping phase by iterating steps 3-5.

[0013] Wherein, step 1 comprises:

[0014] Data registration was performed using the first phase SLC images as reference, and the interferometric combination consisted of images at each adjacent time after registration;

[0015] The interference pattern phase is expressed as:

[0016]

[0017] Where φ(i) is the winding phase of the i-th pixel, is the unwrapped phase of the ith pixel, and k is the number of cycles.

[0018] Wherein, step 3 comprises:

[0019] The two-dimensional Gaussian function model is expressed as:

[0020] f(i,j,m)=Sexp(-a(ix) 2 -2b(ix)(jy)-c(jy) 2 )

[0021]

[0022]

[0023]

[0024] Where S is the amplitude, x, y are the coordinates of the center point, and R x , R y is the standard deviation, θ is the clockwise rotation angle, and the initial value of the model parameter m0=S0,R is determined according to the position of the deformation zone. x0 , R y0 ,x0,y0,θ0] T , the two-dimensional Gaussian function model is inverted by the Levenberg–Marquardt algorithm to obtain the inversion parameter m, m = [S, R x , R y ,x,y,θ] T .

[0025] Wherein, step 4 comprises:

[0026] Estimate the basin-wide deformation phase of the mining area based on the inverted model parameters;

[0027] The whole basin deformation phase is subtracted from the interference pattern phase to obtain the entangled residual interference phase;

[0028] The entangled residual interference phase is phase unwrapped using a traditional phase unwrapping algorithm.

[0029] Wherein, step 5 comprises:

[0030] The residual interference phase after phase unwrapping is added to the basin-wide deformation phase to obtain the surface unwrapping phase of the mining area.

[0031] Among them, step 6 includes: refining the unwrapping phase results by iterating the steps 3-5 until the difference between the last unwrapping phase and the previous unwrapping phase is less than a threshold, the threshold being 0.05; stopping the iteration to obtain the unwrapping phase of the mining area.

[0032] For any pixel (i, j), the relationship between the unwrapped phase at the edge of the mining area and the two-dimensional Gaussian function model is:

[0033]

[0034] Among them, ε(i,j) is the residual of any pixel, and the weighted residual phase is obtained by weighting:

[0035]

[0036] Where σ(i,j) represents the variance determined by the image coherence.

[0037] The above scheme of the present invention has the following beneficial effects:

[0038] Compared with the existing large deformation gradient InSAR phase unwrapping method, the present invention does not require underground mining parameters, field observation data of the mining area and training samples. It only performs phase unwrapping through iterative modeling and parameter inversion of the interference phase, which greatly reduces the limitations in the use of the method. It effectively overcomes the problem that traditional phase unwrapping methods are difficult to unwrap dense and aliased interference fringes in mining areas, has an important impact on the subsequent LOS deformation solution accuracy of the mining area, and broadens the application prospects of InSAR phase unwrapping in mining areas.

[0039] Other beneficial effects of the present invention will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the process of the present invention;

[0041] Figure 2 (a)-(c) are simulation results of winding interference patterns with deformation phase of 20 rad and noise phases of 0.5 rad, 1 rad and 1.5 rad respectively, simulated by probability integral method in an embodiment of the present invention; Figure 2 (d)-(f) in the figure are simulation results of winding interference patterns with deformation phase of 60 rad and noise phase of 0.5 rad, 1 rad and 1.5 rad respectively, simulated by probability integral method in an embodiment of the present invention; Figure 2 (g)-(i) are simulation results of winding interference patterns with deformation phase of 100 rad, noise phase of 0.5 rad, 1 rad and 1.5 rad by probability integral method in an embodiment of the present invention;

[0042] Figure 3 (a)-(i) in the figure are the simulation results of the two-dimensional Gaussian function model with different deformation gradients and different noise phases, corresponding to Figure 2 (a)-(i) Simulated entanglement interference patterns;

[0043] Figure 4 (a)-(i) in the figure are the phase unwrapping results of the mining area with different deformation gradients and different noise phases calculated by the present invention, corresponding to Figure 2 (a)-(i) Simulated entanglement interference patterns of the mining area;

[0044] Figure 5 (a)-(i) in the figure are statistical histogram curves of the difference between different deformation gradients and different noise levels calculated by the embodiment of the present invention, corresponding to Figure 4 Phase unwrapping results of the mining area (a)-(i). DETAILED DESCRIPTION

[0045] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0047] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a locking connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0048] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0049] The present invention aims at the existing problems and provides an iterative unwrapping method of interference phase based on a two-dimensional Gaussian model. First, the entangled interference pattern is obtained by data processing of SAR images. Second, the entangled interference pattern is subjected to adaptive quadtree downsampling processing, and a relatively accurate unwrapping phase of the mining area edge is obtained by a traditional phase unwrapping method. Third, a two-dimensional Gaussian function model for describing the surface deformation of the mining area is established, and the model parameters of the two-dimensional Gaussian function model are inverted by a nonlinear method LM (Levenberg-Marquardt). Fourth, the full basin deformation phase of the mining area is calculated by using a two-dimensional Gaussian function model, and the entangled residual interference phase is obtained by subtracting the phase from the interference pattern, and the residual interference phase is unwrapped by a traditional phase unwrapping method. Fifth, the unwrapped residual interference phase is added to the full basin deformation phase to obtain the surface unwrapping phase of the mining area. Finally, the third, fourth, and fifth processes are iterated until the difference between the last unwrapping phase and the previous unwrapping phase is less than a certain threshold, and the iteration is stopped, thereby obtaining the final unwrapping phase of the mining area.

[0050] like Figure 1 As shown, an embodiment of the present invention provides an interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model, which is applied to InSAR monitoring of large gradient deformation in mining areas, including:

[0051] The single-view complex SLC images are obtained by preprocessing the collected SAR images, and the data are aligned with the first-phase SLC images as reference. The images at each adjacent time after alignment are combined into an interference combination. The entangled interferogram is obtained by forming the interference combination and the phase of the interferogram is calculated.

[0052] The interference pattern phase is expressed as:

[0053]

[0054] Where φ(i) is the winding phase of the i-th pixel, is the unwrapped phase of the ith pixel, and ik is the number of cycles.

[0055] Specifically, this embodiment first uses the probability integral method to simulate the three-dimensional large-scale time-series deformation phase (in rad) of the surface of the mining area caused by the mining of an underground working face at four time points. According to the incident angle and heading angle of the TerraSAR-X SAR satellite data, the three-dimensional deformation is projected to the LOS direction to obtain the LOS direction time-series deformation phase. After subtracting and twisting the LOS deformation phases at adjacent times, three interference patterns with different deformation gradients are obtained. In order to make the simulation closer to the actual situation of mining in the mining area, the atmospheric phase (maximum phase change is 2 rad) and different noise phases (0.5 rad, 1 rad and 1.5 rad) are added, and a total of 9 mining area interference pattern results are simulated, as shown in FIG. Figure 2 shown.

[0056] The interferogram phase is processed by adaptive quadtree downsampling, and the traditional phase unwrapping method is used to calculate the unwrapped phase at the edge of the mining area.

[0057] Specifically, an embodiment of the present invention performs adaptive quadtree downsampling processing on the simulated mining area interference pattern results, so that more points are retained in the deformation area and fewer points are retained in the non-deformation area, and then calculates the mining area edge unwrapping phase results in the high coherence area according to the traditional MCF phase unwrapping method.

[0058] Among them, adaptive quadtree downsampling processing is an image segmentation algorithm based on uniformity detection, in order to obtain the mean of each pixel with the least amount of data, so as to obtain the best approximation to the original image. The basic process of quadtree downsampling processing for images is: first, the dimension of the image is expanded to a power of 2 by adding zero values, and then the image is divided into 4 quadrants of equal size, and the variance of the unwrapped value in each quadrant is calculated respectively, and compared with the pre-set variance threshold. If the variance of a quadrant exceeds the threshold, the quadrant is further decomposed into 4 new quadrants, and the above process is repeated until the variance of each quadrant is less than the given threshold. Finally, the mean or median of each decomposed quadrant is assigned to the central coordinate of the quadrant, and only the modified quadrant center coordinate point is retained.

[0059] A two-dimensional Gaussian function model is established for the unwrapped phase, and the model parameters are calculated and inverted.

[0060] Specifically, the two-dimensional Gaussian function model is expressed as:

[0061] f(i,j,m)=Sexp(-a(ix) 2 -2b(ix)(jy)-c(jy) 2 )

[0062]

[0063]

[0064]

[0065] Where S is the amplitude, x, y are the coordinates of the center point, and R x , R y is the standard deviation, θ is the clockwise rotation angle, and the initial value of the model parameter m0=S0,R is determined according to the position of the deformation zone. x0 , R y0 ,x0,y0,θ0] T , the two-dimensional Gaussian function model is inverted by the Levenberg–Marquardt algorithm to obtain the inversion parameter m, m = [S, R x , R y ,x,y,θ] T .

[0066] For any pixel (i, j), the relationship between the unwrapped phase at the edge of the mining area and the two-dimensional Gaussian function model is:

[0067]

[0068] Among them, ε(i,j) is the residual of any pixel, and the weighted residual phase is obtained by weighting:

[0069]

[0070] Where σ(i,j) represents the variance determined by the image coherence.

[0071] The two-dimensional Gaussian function model is used to calculate the basin-wide deformation phase of the mining area, and then the entangled residual phase is calculated and unwrapped to obtain the unwrapped phase of the mining area surface.

[0072] Specifically, the whole basin deformation phase is subtracted from the interference pattern phase to obtain the entangled residual interference phase; and a traditional phase unwrapping algorithm is used for phase unwrapping; the embodiment of the present invention adopts the minimum cost flow method to phase unwrap the entangled residual interference phase, and the residual interference phase after phase unwrapping is added to the whole basin deformation phase to obtain the surface unwrapping phase of the mining area.

[0073] The calculated unwrapped phase results are refined through iteration to obtain the unwrapped phase of the mining area, and the condition for stopping the iteration is whether the root mean square difference between two adjacent unwrapped phases is less than a certain threshold. The threshold is set to 0.05. The simulation results of the two-dimensional Gaussian function model calculated by the last iteration are as follows: Figure 3 As shown, Figure 3 (a)-(i) are the simulation results of the two-dimensional Gaussian function model with different deformation gradients and different noise phases, corresponding to Figure 2 (a)-(i) Simulated entanglement interference patterns.

[0074] The phase unwrapping result of the mining area is obtained by adding the two-dimensional Gaussian model result of the last iteration calculation to the residual unwrapping phase, such as Figure 4 shown. Figure 4 (a)-(i) are the phase unwrapping results of the mining area with different deformation gradients and different noise phases calculated by the present invention, corresponding to Figure 2 (a)-(i) Simulated entanglement interference patterns of the mining area; Figure 4 It can be seen that when the noise level is small, even if the deformation gradient of the mining area is large, the present invention can obtain a high-precision phase unwrapping result of the mining area interference pattern; when the noise level is large, even if the deformation gradient is small, the obtained mining area unwrapping phase will have a certain error, but the error is within an acceptable range. The results show that within a certain noise error level, the embodiment of the present invention can solve the problem that the interference pattern is dense and difficult to unwrap due to the large gradient deformation of the mining area, and obtain accurate phase unwrapping results.

[0075] In order to evaluate the unwrapping effect of the embodiment of the present invention on the dense interference pattern fringes in the mining area, the statistical histogram curve of the difference between the phase unwrapping result and the simulation value is calculated, as shown in FIG. Figure 5 shown. Figure 5(a)-(i) are statistical histogram curves of the difference between different deformation gradients and different noise levels calculated by the embodiment of the present invention, corresponding to Figure 4 (a)-(i) The phase unwrapping results of the mining area; By comparing the simulated large gradient LOS deformation phase with the LOS unwrapping phase of the mining area calculated by the embodiment of the present invention, it is found that the two are in good agreement. Under the condition of adding random noise of 0.5 rad, the 95% probability of the difference between the deformation gradients of 20 rad, 60 rad and 100 rad and the simulated values ​​is 0.16 rad, 0.18 rad and 0.18 rad respectively. Under the condition of adding random noise of 1.0 rad, the 95% probability of the difference between the deformation gradients of 20 rad, 60 rad and 100 rad and the simulated values ​​is 0.21 rad, 0.28 rad and 0.3 rad respectively. Under the condition of adding random noise of 2.0 rad, the 95% probability of the difference between the deformation gradients of 20 rad, 60 rad and 100 rad and the simulated values ​​is 1.6 rad, 1.8 rad and 3.0 rad respectively. Even for a large gradient deformation of 100 rad, under different noise conditions (0.5 rad, 1.0 rad and 1.5 rad), the difference accounts for nearly 1.6%, 1.8% and 3% of the total deformation gradient respectively, indicating the reliability of the phase unwrapping results of the mining area calculated by the embodiment of the present invention, and meets the accuracy requirements of the phase unwrapping of a large number of magnitude deformation interference patterns in the mining area.

[0076] The embodiments of the present invention illustrate that compared with the existing large deformation gradient InSAR phase unwrapping methods, the present invention does not require underground mining parameters, field observation data in the mining area, and training samples. It only performs phase unwrapping through iterative modeling and parameter inversion of the interference phase, which greatly reduces the limitations in the use of the method, and effectively overcomes the problem that traditional phase unwrapping methods are difficult to unwrap dense and aliased interference fringes in mining areas. It has an important impact on the accuracy of subsequent surface LOS deformation solution in mining areas, and broadens the application prospects of InSAR phase unwrapping in mining areas.

[0077] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model, applied to InSAR monitoring of large gradient deformation in mining areas, characterized in that: include: Step 1, collect SAR images for data preprocessing, obtain single-view complex SLC images, and obtain the entangled interferogram by forming an interference combination, and calculate the interferogram phase; Step 2, performing adaptive quadtree downsampling processing on the interference pattern phase, and using a traditional phase unwrapping method to calculate the unwrapped phase at the edge of the mining area; Step 3, establishing a two-dimensional Gaussian function model of the unwrapped phase at the edge of the mining area, calculating model parameters and inverting the model parameters; Step 4, obtaining the entangled residual interference phase by estimating the basin-wide deformation phase of the mining area, and untangling the residual interference phase; Step 5, obtaining the unwrapped phase of the coal mining area; Step 6, obtain the refined mineral area unwrapping phase by iterating steps 3-5.

2. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 1, characterized in that: The step 1 comprises: The data is registered with reference to the first phase SLC image, and the interference combination is composed of images at each adjacent time after registration; The interference pattern phase is expressed as: Where φ(i) is the winding phase of the i-th pixel, is the unwrapped phase of the ith pixel, and k is the number of cycles.

3. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 1, characterized in that: The step 3 comprises: The two-dimensional Gaussian function model is expressed as: f(i,j,m)=Sexp(-a(i-x) 2 -2b(i-x)(j-y)-c(j-y) 2 ) Where S is the amplitude, x, y are the coordinates of the center point, and R x , R y is the standard deviation, θ is the clockwise rotation angle, and the initial value of the model parameter m0=[S0,R x0 , R y0 ,x0,y0,θ0] T , the two-dimensional Gaussian function model is inverted by the Levenberg–Marquardt algorithm to obtain the inversion parameter m, m = [S, R x , R y ,x,y,θ] T .

4. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 1, characterized in that: The step 4 comprises: Estimate the basin-wide deformation phase of the mining area based on the inverted model parameters; The whole basin deformation phase is subtracted from the interference pattern phase to obtain a tangled residual interference phase; The entangled residual interference phase is phase unwrapped using a traditional phase unwrapping algorithm.

5. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 1, characterized in that: The step 5 comprises: adding the residual interference phase after phase unwrapping to the whole basin deformation phase to obtain the mining area surface unwrapping phase.

6. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 1, characterized in that: The step 6 comprises: refining the unwrapping phase result by iterating the steps 3-5 until the difference between the last unwrapping phase and the previous unwrapping phase is less than a threshold, stopping the iteration, and obtaining the mining area unwrapping phase.

7. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 3, characterized in that: For any pixel (i, j), the relationship between the unwrapped phase of the mining area edge and the two-dimensional Gaussian function model is: Among them, ε(i,j) is the residual of any pixel, and the weighted residual phase is obtained by weighting: Where σ(i,j) represents the variance determined by the image coherence.

8. The interferometric phase iterative unwrapping method based on a two-dimensional Gaussian model according to claim 6, characterized in that: The threshold is 0.05.