A Robust Chinese Remainder Theorem-Based Multi-Baseline Phase Unwrapping Method Considering Branch Cut Lines

Abstract

The present invention discloses a multi-baseline phase unwrapping method of a robust Chinese Remainder Theorem considering branch cut lines. The method comprises the following steps: generating multi-baseline InSAR phase interferograms for the same region of interest; setting branch cut lines for the interferograms by using the branch cut method, and obtaining unwrapped phases in a large range by using the Chinese Remainder Theorem to avoid the branch cut lines through appropriate means; generating local branch cut lines for the branch cut line distribution region, and performing unwrapping again by using the Chinese Remainder Theorem; combining the phase unwrapping information of the large range region and the local region to obtain the final phase unwrapping information. This solution uses branch cut lines and an adaptive rectangular box to extract regions with strong noise, uses relevant filtering algorithms to weaken the interference of noise on interferometric phase information, uses the Chinese Remainder Theorem algorithm twice to unwrap the large area region with weak noise information and the local region with strong noise information respectively, and splices the unwrapped phase information to implement an anti-noise multi-baseline phase unwrapping algorithm.

Description

Technical Field

[0001] The present invention relates to a multi-baseline phase unwrapping method based on a robust Chinese Remainder Theorem considering branch tangents, belonging to the technical field of microwave remote sensing measurement, and is mainly applied to the fields of multi-baseline and high-precision phase information unwrapping and digital elevation model reconstruction. Background Art

[0002] In order to obtain elevation information of steep and undulating areas that are difficult to measure manually, synthetic aperture radar interferometry (InSAR) realizes the acquisition of a digital elevation model (DEM) for topographic mapping without on-site measurement in the form of a combination of signal processing and image processing. This is one of the modern new measurement technologies in the rapidly developing radar microwave remote sensing measurement. At the same time, relying on its all-weather, all-time imaging and strong penetration characteristics, it provides technical support and means for digital image processing. At present, this measurement method provides data support and process management for mountain mapping, environmental monitoring, geological exploration, climate analysis, etc., and has important value for its further research.

[0003] Phase unwrapping technology has always been the key in synthetic aperture radar interferometry. For the phase unwrapping technology in the inversion of elevation information, the performance of its unwrapping algorithm is directly related to the accuracy and precision of elevation information. In the research process of phase unwrapping in recent decades, it has developed into a unique and multi-disciplinary integrated technology. Especially with the development of deep learning algorithms, the improvement of traditional algorithms and the proposal of new algorithms will inevitably become a new trend in phase unwrapping technology.

[0004] Currently, the methods of phase unwrapping are mainly divided into single-baseline phase unwrapping algorithm and displacement multi-baseline phase unwrapping algorithm according to the number of baselines. The single-baseline phase unwrapping algorithm is limited by the assumption of phase continuity. For example, in the branch cut method, although branch tangents are set to effectively avoid areas containing residual points, the unwrapping effect on the phase undersampling area is not good. The multi-baseline phase unwrapping algorithm solves the ambiguity of the wrapped phase differential by introducing multiple phase interferograms and is widely applied to complex terrains with discontinuous phases. For example, the Chinese Remainder Theorem, by solving the congruence equations, accurately and simultaneously solves the ambiguity of multiple interferograms corresponding to a single point, with high accuracy. However, it is greatly affected by noise, and its poor noise robustness makes it unable to be widely used in measured data.

[0005] Based on the characteristics of these two methods, an effective way can be set with the help of the branch cut method to expand it into the Chinese Remainder Theorem phase unwrapping to unwrap the phase ambiguity with less noise in a large range area, then adopt the method of an adaptive rectangular frame to extract the points with larger noise information, filter them, and then use the Chinese Remainder Theorem unwrapping algorithm to process the phase ambiguity of this area again. Finally, the unwrapped phases of the large range area and each rectangular frame area are spliced. Summary of the Invention

[0006] The present invention aims to overcome the disadvantages of the existing phase unwrapping methods, namely, the island phenomenon in the phase unwrapping of the single-baseline branch-cut method algorithm for phase mutations and the weak anti-noise performance of the multi-baseline Chinese Remainder Theorem algorithm. A multi-baseline phase unwrapping method based on the robust Chinese Remainder Theorem considering branch-cut lines is provided. By adopting the technology of combining the single-baseline phase unwrapping algorithm with the multi-baseline phase unwrapping method, the problem of difficult unwrapping in the island area is solved, and the noise robustness of the algorithm is enhanced.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A multi-baseline phase unwrapping method based on the robust Chinese Remainder Theorem considering branch-cut lines of the present invention includes the following steps:

[0009] Generate multi-baseline InSAR phase interferograms for the same region of interest;

[0010] Use the branch-cut method to set branch-cut lines for the interferogram, and adopt the Chinese Remainder Theorem to avoid the branch-cut lines and obtain the unwrapped phase on a large scale through appropriate means;

[0011] For the region where the branch-cut lines are distributed in the interferogram, use the branch-cut method to generate branch-cut lines in a local range, and then use the Chinese Remainder Theorem for unwrapping to obtain the unwrapping information of the local region;

[0012] Combine the phase unwrapping information of the large-scale region and the unwrapping information of the local region to obtain the phase unwrapping information of all points in the interferogram.

[0013] Preferably, the generation of multi-baseline InSAR phase interferograms for the same region of interest further includes:

[0014] Step 1, obtain multiple multi-baseline SAR images for the same region, and select one master image and multiple slave images;

[0015] Step 2, perform differential interferometry processing on the master image and the slave images one by one, and arrange them according to the length of the baseline to generate multiple multi-baseline InSAR phase interferograms to be unwrapped.

[0016] Preferably, the use of the branch-cut method to set branch-cut lines for the interferogram and the adoption of the Chinese Remainder Theorem to avoid the branch-cut lines and obtain the unwrapped phase on a large scale through appropriate means further includes:

[0017] Step 3, preprocess the interferogram using the branch-cut method to generate branch-cut lines;

[0018] Step 4: Establish a multi-baseline InSAR geometric model based on the multiple interferograms, and build a Chinese Remainder Theorem mathematical model according to the multi-baseline InSAR geometric model;

[0019] Step 5: According to the preprocessed branch tangent distribution, avoid the branch tangents and adopt a reasonable approach to use the Chinese Remainder Theorem to unwrap the interferogram and obtain the phase unwrapping information of a large-scale area.

[0020] Preferably, for the branch tangent distribution area in the interferogram, the branch cut method is used to generate branch tangents in a local range, and the Chinese Remainder Theorem is used again for unwrapping to obtain the unwrapping information of the local area, which further includes:

[0021] Step 6: Use an adaptive rectangular box to extract the branch tangent distribution areas in the interferogram one by one;

[0022] Step 7: Re-select the starting point for unwrapping the interferometric phase within the box using the branch cut method to generate branch tangents in a local range;

[0023] Step 8: Filter the phase information of the points on the branch tangents in the local range;

[0024] Step 9: Use the Chinese Remainder Theorem again to unwrap the filtered interferometric phase within the rectangular box to obtain the unwrapping information of the local area.

[0025] Preferably, Step 4 further includes:

[0026] The parameter expression of the multi-baseline InSAR system geometric model is as follows:

[0027]

[0028] In the formula, h represents the height difference, λ represents the wavelength, θ represents the side view angle, α represents the baseline horizontal angle, B i represents different baseline lengths, R represents the slant range of the phase center relative to the target point, k i represents the ambiguity number of the interferometric phase at different points.

[0029] If B0 = [B1, B2, …, B n , where [] represents finding the least common multiple, and at the same time setting the modulus length m i = B0 / B i , the interferometric phase can be expressed by the height difference:

[0030]

[0031] If then the ambiguity numbers to be solved for the interferometric phases corresponding to each point in the interferograms under different baselines satisfy the following equation:

[0032] T = w i + k i ·m i

[0033] Corresponding to N interferograms formed by N groups of baselines, each corresponding point position consists of N groups of equations, and in each group of equations, m i is determined, w i is different, T is the same. According to mathematical theory, a system of congruence equations is constructed, and the ultimately required value, the fuzzy number k i is the process of solving the Chinese Remainder Theorem mathematical model.

[0034] Preferably, step 5 further includes: according to the preprocessed branch cut line distribution, for the terrain mutation areas with many derivatives of branch cut lines, the Chinese Remainder Theorem is used to unwrap the interferogram to obtain the phase unwrapping information of a large area.

[0035] Preferably, in step 6, the size of the interferogram is M×N, the size of the rectangular frame is u×v, and the size of the adaptive rectangular frame satisfies the formula:

[0036] u = Floor[0.1×M + Ω]

[0037] v = Floor[0.1×N + Ω]

[0038] where Ω is the signal-to-noise ratio of the interferogram.

[0039] Preferably, the filtering process in step 8 uses median filtering.

[0040] A single baseline phase unwrapping algorithm provided by the present invention combines with a multi-baseline phase unwrapping method, considering the limitation of the island phenomenon of the branch cut method and the weak anti-noise ability of the Chinese Remainder Theorem algorithm. The branch cut line and the adaptive rectangular frame are used to extract the areas with strong noise, the relevant filtering algorithm is used to weaken the interference of the noise on the interferometric phase information, the Chinese Remainder Theorem algorithm is used twice to unwrap the large area with weak noise information and the local area with strong noise information respectively, and the unwrapped phase information is spliced to realize the anti-noise multi-baseline phase unwrapping algorithm. Description of the Drawings

[0041] Figure 1 is the differential processing of the interferogram to be unwrapped in Embodiment 3 of the present application.

[0042] Figure 2 is the multi-baseline InSAR geometric model in Embodiment 3 of the present application.

[0043] Figure 3 is the three-dimensional and two-dimensional reference map corresponding to the real DEM data of the American Isolation Peak National Park in Embodiment 3 of the present application.

[0044] Figure 4 It is the differential interferogram corresponding to the three baselines in Embodiment 3 of the present application, where they are sorted by the length of the baseline.

[0045] Figure 5 It is the flowchart of Embodiment 3 of the present application.

[0046] Figure 6 It is the residual points searched in the global range by using the branch cut method to process three interferograms in Embodiment 3 of the present application.

[0047] Figure 7 It is the branch cut lines formed in the global range by using the branch cut method to process three interferograms in Embodiment 3 of the present application.

[0048] Figure 8 It is the large - scale unwrapped phase obtained by using the Chinese Remainder Theorem to avoid the branch cut lines and taking appropriate paths in Embodiment 3 of the present application.

[0049] Figure 9 It is the local branch cut lines formed within the rectangular frame by using the branch cut method to process any undersampled area in Embodiment 3 of the present application.

[0050] Figure 10 It is the unwrapped phase integrated by using the Chinese Remainder Theorem again after median filtering the phase information within all rectangular frames in Embodiment 3 of the present application.

[0051] Figure 11 It is the unwrapped phase of the entire image stitched together from the large - scale unwrapped phase and the local area in Embodiment 3 of the present application.

[0052] Figure 12 It is the error distribution diagram of the difference between the unwrapped phase obtained in Embodiment 3 of the present application and the reference phase.

[0053] Figure 13 It is the error histogram of the difference between the unwrapped phase obtained in Embodiment 3 of the present application and the reference phase. Detailed implementation manners

[0054] The present invention will be further described below in conjunction with the accompanying drawings and detailed implementation manners.

[0055] Embodiment 1

[0056] As Figure 1 shown, a multi - baseline phase unwrapping method based on a robust Chinese Remainder Theorem considering branch cut lines of the present invention includes the following steps:

[0057] Step S1, generating multi - baseline InSAR phase interferograms for the same region of interest.

[0058] Step S2: Set branch cut lines for the interferogram using the branch cut method, and use the Chinese Remainder Theorem to avoid the branch cut lines and obtain the unwrapped phase over a large range through a suitable approach.

[0059] Step S3: For the area where branch cut lines are distributed in the interferogram, generate branch cut lines within a local range using the branch cut method, and perform unwrapping again using the Chinese Remainder Theorem to obtain the unwrapping information for the local area.

[0060] Step S4: Combine the phase unwrapping information for the large range area and the unwrapping information for the local area to obtain the phase unwrapping information for all points in the interferogram.

[0061] This solution takes into account the limitations of the branch cut method's island phenomenon and the weak anti-noise ability of the Chinese Remainder Theorem algorithm. It uses branch cut lines and an adaptive rectangular box to extract areas with strong noise, uses relevant filtering algorithms to reduce the interference of noise on the interferometric phase information, uses the Chinese Remainder Theorem algorithm twice to unwrap the large area with weak noise information and the local area with strong noise information respectively, and splices the unwrapped phase information to implement an anti-noise multi-baseline phase unwrapping algorithm.

[0062] Embodiment 2

[0063] This embodiment is an optimized solution based on Embodiment 1. In this embodiment, steps S1, S2, S3, and S4 of Embodiment 1 are further expanded respectively.

[0064] Step S1 specifically further includes the following steps:

[0065] Step 1: Obtain multiple multi-baseline SAR images for the same area, and select one master image and multiple slave images.

[0066] Step 2: Perform differential interferometric processing on the master image and the slave images one by one, and arrange them according to the length of the baseline to generate multiple multi-baseline InSAR phase interferograms to be unwrapped.

[0067] Step S2 specifically further includes the following steps:

[0068] Step 3: Preprocess the interferogram using the branch cut method to generate branch cut lines.

[0069] Step 4: Establish a multi-baseline InSAR geometric model based on the multiple interferograms, and build a Chinese Remainder Theorem mathematical model according to the multi-baseline InSAR geometric model;

[0070] Step 5: According to the distribution of the preprocessed branch cut lines, avoid the branch cut lines and use a reasonable approach, and use the Chinese Remainder Theorem to unwrap the interferogram to obtain the phase unwrapping information for the large range area.

[0071] Specifically, the generation of branch cutting lines in step 3 involves the two-dimensional phase unwrapping theory and residual points.

[0072] First of all, two-dimensional phase unwrapping is for a rectangular space of size M×N. Among them, a point in the space is selected as the starting point to expand the phase in different directions. Then, the two-dimensional phase unwrapping path is divided into horizontal and vertical directions. Among them, the interference phase φ at the starting point position and the unwrapped phase Have the following relationship:

[0073]

[0074] Among them, W[φ(i,j)] is defined as the phase wrapping function. In the above-mentioned path, for the phase gradients in the horizontal and vertical directions, they can be defined as:

[0075] △ i φ(i,j) = φ(i,j) - φ(i - 1,j) (2)

[0076] △ j φ(i,j) = φ(i,j) - φ(i,j - 1) (3)

[0077] According to the different paths, using the phase wrapping function, the wrapped phase gradients in the vertical and horizontal directions of adjacent pixels are:

[0078] △ i W[φ(i,j)] = △ i φ(i,j) + 2π△ i k(i,j) (5)

[0079] △ j W[φ(i,j)] = △ j φ(i,j) + 2π△ j k(i,j) (6)

[0080] Using the phase wrapping function again for the wrapped phase gradients in the above different directions, the wrapped phases in the vertical and horizontal directions of adjacent pixels can be obtained:

[0081] W2[△ i W1[φ(i,j)] = △ i φ(i,j) + 2π[△ i k1(i,j) + △ i k2(i,j)] (7)

[0082] W3[△ j W1[φ(i,j)] = △ j φ(i,j) + 2π[△ j k3(i,j) + △ jk4(i,j)] (8)

[0083] To ensure that the winding phase is between -π and π, the corresponding parameters in the formula should satisfy the following relational expressions:

[0084] π ≤ △ i φ(i,j) < π (9)

[0085] π ≤ △ j φ(i,j) < π (10)

[0086] △ i k1(i,j) + △ i k2(i,j) = 0 (11)

[0087] △ j k3(i,j) + △ j k4(i,j) = 0 (12)

[0088] Next, a reliable path integral will be adopted to achieve unwrapping, first integrating vertically and then horizontally, or first integrating horizontally and then vertically.

[0089] Integrating vertically first and then horizontally:

[0090] First, perform pixel unwrapping on the first column in the vertical direction:

[0091]

[0092] After unwrapping the pixels in the first column, use it as the new starting column. At this time, for each column, perform horizontal gradient integration on the previous column in sequence. When the processing of column j is completed, the unwrapping process is finished.

[0093]

[0094] Integrating horizontally first and then vertically:

[0095] First, perform pixel unwrapping on the first column in the horizontal direction:

[0096]

[0097] After unwrapping the pixels in the first row, use it as the new starting row. At this time, for each row, perform horizontal gradient integration on the previous row in sequence. When the processing of row i is completed, the unwrapping process is finished.

[0098]

[0099] Ideally, in the case of continuous phase, according to any one of the above two path integrals, the true phase of the entire image can be extended from the wrapped phase at one point. However, in actual situations, due to the presence of noise, the selection of the path will deviate, and even the point position error will accumulate and expand into an error in a larger interval.

[0100] Based on the above situation, the concept of residual points needs to be introduced, that is, the phase continuity is determined by solving the azimuth gradients of adjacent pixels:

[0101] △1 = φ(i + 1, j) - φ(i, j) (17)

[0102] △2 = φ(i + 1, j + 1) - φ(i + 1, j) (18)

[0103] △3 = φ(i, j + 1) - φ(i + 1, j + 1) (19)

[0104] △4 = φ(i, j) - φ(i, j + 1) (20)

[0105] The sum of the azimuth gradients solved is the residual:

[0106]

[0107] The continuity of the phase information is determined using the solved residual, that is: T = 0, this point is determined as a phase continuous point; T ≠ 0, this point is determined as a residual point. Usually, it is defined as: T > 0, this point is determined as a positive residual point; T < 0, this point is determined as a negative residual point.

[0108] The introduction of residual points has a guiding effect on the path tracking algorithm, especially for the single - baseline unwrapping algorithm, which ensures the assumption of phase continuity in a large - range area.

[0109] Specifically, step 4 further includes:

[0110] For the parameter expressions of the geometric model of the multi - baseline InSAR system are as follows:

[0111]

[0112] In the formula, h represents the height difference, λ represents the wavelength, θ represents the side - view angle, α represents the baseline horizontal angle, B i represents different baseline lengths, R represents the slant range of the phase center relative to the target point, k i represents the ambiguity number of the interference phase at different point positions.

[0113] If B0 = [B1, B2, …, B n , where [] represents finding the least common multiple, and at the same time setting the modulus length m i = B0 / Bi , the interference phase can be expressed by the height difference:

[0114]

[0115] If it is set then the ambiguity numbers to be solved corresponding to the interference phases at each point in the interferograms with different baselines satisfy the following equation:

[0116] T = w i + k i ·m i (24)

[0117] Corresponding to the N interferograms formed by N groups of baselines, each corresponding point is composed of N groups of equations. And in each group of equations, m i is determined, w i is different, T is the same. According to mathematical theory, a system of congruence equations is constructed, and the finally required value, the ambiguity number k i is the process of solving the Chinese Remainder Theorem mathematical model.

[0118] Specifically, step 5 further includes, according to the preprocessed branch cut line distribution, for the terrain mutation areas with more derivatives of branch cut lines, using the Chinese Remainder Theorem to unwrap the interferogram and obtain the phase unwrapping information of a large range of areas.

[0119] Step S3 specifically further includes the following steps:

[0120] Step 6: Use an adaptive rectangular box to extract the branch cut line distribution areas in the interferogram one by one;

[0121] Step 7: Re-select the unwrapping points for the interference phase within the box using the branch cut method to generate branch cut lines within a local range;

[0122] Step 8: Perform filtering processing on the phase information of the points on the branch cut lines within the local range. The filtering processing in step 8 uses median filtering processing;

[0123] Step 9: Use the Chinese Remainder Theorem again to unwrap the filtered interference phase within the rectangular box to obtain the unwrapping information of the local area.

[0124] Specifically, in step 6, the size of the interferogram is M×N, the size of the rectangular box is u×v, and the size of the adaptive rectangular box satisfies the formula:

[0125] u = Floor[0.1×M + Ω] (25)

[0126] v = Floor[0.1×N + Ω] (26)

[0127] where, Ω is the signal-to-noise ratio of the interferogram.

[0128] Example 3

[0129] This example applies the method of Example 2 and uses the data of IsolationPeak National Park in the United States as a specific operation example to elaborate on the solution of this application in detail.

[0130] As Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 shown, the data used in this example is one main image and three auxiliary images. According to the multi-baseline geometric model and the differential principle, three sets of interferograms corresponding to different baselines are generated as the data to be unwrapped. As Figure 5 shown, the following is a further description of the specific implementation process of the method in this example.

[0131] Step 1: Distinguish the main image and the auxiliary images from the four images, perform differential interferometry processing, and arrange them according to the length of the baseline to form a multi-baseline InSAR phase interferogram.

[0132] Step 2: For the accuracy of the data and experimental comparative analysis, Figure 3 three-dimensional and two-dimensional reference maps of the real DEM data of Isolation Peak National Park in the United States are given.

[0133] Step 3: For each of the three interferograms, the branch-cut method is used to process the positive and negative residual points searched in the global range, and the formula is used for calculation, where the positive residual points are represented by 1 and the negative residual points are represented by -1.

[0134] Step 4: Use the residual points extracted in Step 3 to set branch-cut lines for the global phase, which serves as a path tracking guidance.

[0135] Step 5: Based on the branch-cut lines already generated in the interferogram, the Chinese Remainder Theorem is used to avoid the branch-cut lines and obtain the large-range unwrapped phase by taking appropriate paths.

[0136] Figure 6 shows the residual points searched in the global range for each of the three interferograms processed by the branch-cut method, Figure 7 shows the branch-cut lines formed in the global range for each of the three interferograms processed by the branch-cut method.

[0137] Figure 8 shows the large-range unwrapped phase obtained by taking appropriate paths to avoid the branch-cut lines using the Chinese Remainder Theorem.

[0138] Step 6: For the generation of the unwrapped phase globally, it is necessary to consider the regions of local undersampling and phase mutations. The branch-cut method is used to process the regions of undersampling and phase mutations one by one to form local branch-cut lines within the rectangular frame and mark this position. Among them, there are many phase mutation points in one undersampling region during the experimental process. To ensure 60% stable phase information, the size of the adaptive rectangular frame is expanded to 98*52, as shown in Figure 9 shown.

[0139] Step 7: Median filtering is performed on the phase information of the points on the branch-cut line. If there are only 1-2 phase mutation points within the rectangular frame, it is defaulted that the unwrapping effect within the rectangular frame is good and no filtering and re-unwrapping are required.

[0140] Step 8: The Chinese Remainder Theorem is used again to unwrap the interferometric phase after filtering within the rectangular frame. The result is as shown in Figure 10 shown.

[0141] Step 9: By combining the overall phase unwrapping information and the small-region unwrapping information, the phase unwrapping information of all points in the interferogram is obtained; as shown in Figure 11 shown is the unwrapped phase of the large-scale unwrapped phase and the entire image stitched by the local region.

[0142] Step 10: To verify the anti-noise performance and the unwrapping performance at the phase mutation points of this algorithm, an error distribution diagram and an error histogram of the difference between the unwrapped phase and the reference phase are made.

[0143] As can be seen from Figure 11 it, the multi-baseline phase unwrapping map integrated by this algorithm has a relatively high overall coincidence degree with the original DEM, and no island phenomenon appears.

[0144] As can be seen from Figure 12 it, in the error distribution diagram of the difference between the unwrapped phase and the reference phase of this algorithm, the overall phase error value is relatively low. Only a few bright spots are the phase mutation points that are not filtered within the rectangular frame and have 1-2, and their error values are relatively large.

[0145] As can be seen from Figure 13 it, the error histogram distribution is also around 0rad, and the frequency of the mutated points is very low, so the robustness of this algorithm is improved and the anti-noise ability is verified.

Claims

1. A multi-baseline phase unwrapping method based on a robust Chinese Remainder Theorem considering branch cut lines, characterized in that The method includes the following steps: Generate multi-baseline InSAR phase interferograms for the same region of interest; Use the branch cut method to set branch cut lines for the interferograms, and use the Chinese Remainder Theorem to avoid the branch cut lines and obtain the unwrapped phase over a large range through appropriate means; For the region where branch cut lines are distributed in the interferogram, use the branch cut method to generate branch cut lines within a local range, and use the Chinese Remainder Theorem again for unwrapping to obtain the unwrapping information of the local region; Combine the phase unwrapping information of the large-range region and the unwrapping information of the local region to obtain the phase unwrapping information of all points in the interferogram; The step of using the branch cut method to generate branch cut lines within a local range for the region where branch cut lines are distributed in the interferogram, and using the Chinese Remainder Theorem again for unwrapping to obtain the unwrapping information of the local region further includes: Step 6: Extract one by one using an adaptive rectangular box for the region where branch cut lines are distributed in the interferogram; Step 7: Re-select the unwrapping points for the interferometric phase within the box using the branch cut method to generate branch cut lines within a local range; Step 8: Filter the phase information of points on the branch cut lines within the local range; Step 9: Use the Chinese Remainder Theorem again to unwrap the interferometric phase after filtering within the rectangular box to obtain the unwrapping information of the local region.

2. The multi-baseline phase unwrapping method of the robust Chinese Remainder Theorem considering branch tangent lines according to claim 1, characterized in that, The step of generating multi-baseline InSAR phase interferograms for the same region of interest further includes: Step 1, Obtain multiple multi-baseline SAR images for the same region, and select one master image and multiple slave images; Step 2, Perform differential interferometric processing on the master image and the slave images one by one, and arrange them according to the length of the baselines to generate multiple multi-baseline InSAR phase interferograms to be unwrapped.

3. The multi-baseline phase unwrapping method of the robust Chinese Remainder Theorem considering branch tangent lines according to claim 2, characterized in that, The step of using the branch cut method to set branch cut lines for the interferogram, and using the Chinese Remainder Theorem to avoid the branch cut lines and obtain the unwrapped phase over a large range through appropriate means further includes: Step 3, Preprocess the interferogram using the branch cut method to generate branch cut lines; Step 4, Establish a multi-baseline InSAR geometric model based on the multiple interferograms, and build a Chinese Remainder Theorem mathematical model according to the multi-baseline InSAR geometric model; Step 5, According to the distribution of the preprocessed branch cut lines, avoid the branch cut lines and use the Chinese Remainder Theorem to unwrap the interferogram through reasonable means to obtain the phase unwrapping information of the large-range region.

4. The multi-baseline phase unwrapping method based on the robust Chinese Remainder Theorem considering branch tangent lines according to claim 3, characterized in that The step 4 further includes: The parameter expressions for the geometric model of the multi-baseline InSAR system are as follows: Wherein, h represents the height difference, λ represents the wavelength, θ represents the side viewing angle, α represents the baseline horizontal angle, B i represents different baseline lengths, R represents the slant range of the phase center relative to the target point, k i represents the ambiguity number of the interference phase at different positions; If we set B0 = [B1, B2, …, B n , where [] represents finding the least common multiple, and at the same time set the modulus length m i = B0 / B i , the interference phase can be expressed by the height difference: If set Then the ambiguity numbers to be solved corresponding to the interference phases at each point in the interference patterns under different baselines satisfy the following equation: T = w i + k i · m i Corresponding to the N interferograms formed by N groups of baselines, each corresponding point position consists of N groups of equations, and in each group of equations, m i is determined, w i is different, T is the same. According to mathematical theory, a system of congruence equations is constructed, and the value to be finally obtained, the fuzzy number k i is the process of solving the Chinese Remainder Theorem mathematical model.

5. The multi-baseline phase unwrapping method of the robust Chinese Remainder Theorem considering branch tangent lines according to claim 3, characterized in that, The step 5 further includes: According to the distribution of the preprocessed branch cut lines, for the terrain mutation regions with more derived branch cut lines, use the Chinese Remainder Theorem to unwrap the interferogram to obtain the phase unwrapping information of the large-range region.

6. The multi-baseline phase unwrapping method of the robust Chinese Remainder Theorem considering branch tangent lines according to claim 1, characterized in that, In the step 6, the size of the interferogram is M×N, the size of the rectangular box is u×v, and the size of the adaptive rectangular box satisfies the formula: u = Floor[0.1×M + Ω] v = Floor[0.1×N + Ω] where Ω is the signal-to-noise ratio of the interferogram.

7. The multi-baseline phase unwrapping method of the robust Chinese Remainder Theorem considering branch tangent lines according to claim 1, characterized in that, The filtering process in the step 8 uses median filtering.

Images