InSAR Phase Filtering Method Based on Complex Interferograms
By using an InSAR phase filtering method based on complex interferograms and constructing an adaptive filter using K-means clustering and Kaiser window function, the problem of low noise removal efficiency in existing InSAR filtering methods is solved, achieving more efficient noise filtering and signal preservation, and improving the deformation monitoring accuracy and computational efficiency of InSAR technology.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NORTHWESTERN POLYTECHNICAL UNIV
- Filing Date
- 2025-05-28
- Publication Date
- 2026-05-26
AI Technical Summary
Existing InSAR phase filtering methods are inefficient in removing noise, and the design flaws of existing filters mean that noise in the interferogram cannot be completely eliminated, affecting the accuracy of unwrapping and spatial resolution.
A phase filtering method for InSAR based on complex interferograms is designed. An adaptive two-dimensional low-pass filter is constructed using K-means clustering and Kaiser window function. By analyzing the difference between signal and noise in the frequency domain, noise is directly filtered out in the frequency domain while the signal is preserved, avoiding the iterative process.
It improves the deformation monitoring accuracy of InSAR technology, simplifies the calculation process, effectively removes noise in interferograms, enhances filtering effect, and reduces computational complexity.
Smart Images
Figure CN120669240B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of time-series data processing technology in synthetic aperture radar interferometry, and particularly to an InSAR phase filtering method based on complex interferograms. Background Technology
[0002] Interferometric synthetic aperture radar (InSAR), as an active microwave sensor, acquires both ground object scattering information and surface elevation information. The cloud and fog penetration capabilities of microwaves have enabled it to achieve significant success and contributions in the field of remote sensing Earth observation. In particular, its interferometric phase measurement is considered a crucial feature for current topographic mapping and deformation monitoring. InSAR measurements are the difference in phase observed from two slightly different locations of the same ground object, inevitably introducing noise components. The presence of phase noise severely affects the accuracy of unwrapping, making phase denoising a necessary step in the interferometric processing. This noise can be broadly categorized into three parts: 1) system noise, such as thermal noise and speckle noise; 2) noise caused by the observation scattering mechanism, including baseline, time, atmospheric, and volume decorrelation; and 3) noise caused by signal processing, primarily including co-registration errors. Phase noise on the interferogram is the result of the combined effect of one or more of these factors.
[0003] Since InSAR interferometric phase was first used to extract seismic deformation fields in 1993, phase denoising has been a hot topic in InSAR processing research. The goal of phase denoising has always been to remove as much interferometric phase noise as possible without reducing spatial resolution; these methods can be roughly divided into four categories: traditional spatial domain filters, non-local (NL) filters, transform domain filters, and filters based on new methods.
[0004] In summary, frequency domain filtering algorithms based on Fourier transform are simple in principle and easy to implement. However, existing research is limited to the Goldstein filtering algorithm framework and does not discuss the principles and functions behind choosing the filter shape S{H(u,v)} in the Goldstein method. Instead, it chooses to improve filtering efficiency by modeling the filter factor. In practical applications, simply choosing the filter factor cannot completely compensate for the problems caused by filter shape defects and also reduces filtering efficiency. Finally, this manifests as some unavoidable noise in the interferogram.
[0005] Therefore, based on the spectral differences between noise and signal in the frequency domain, a new filter shape and filtering method need to be designed to extend the Goldstein filter into a two-dimensional ideal low-pass filter. Interferograms with strong noise signals no longer require multiple iterations to obtain better phase filtering results, thus optimizing the filtering results while simplifying the calculation process of existing interferogram phase filtering methods. Summary of the Invention
[0006] In view of this, this invention analyzes the difference in spectral amplitude between noise and signal after Fourier transform, and explores a frequency-domain-based, computationally convenient, unrestricted by external conditions, and universal InSAR phase filtering method based on complex interferograms, thereby improving the deformation monitoring capability of time-series InSAR technology.
[0007] To achieve the above objectives, the present invention employs the following technical solution:
[0008] InSAR phase filtering methods based on complex interferograms include:
[0009] The acquired primary and secondary SAR images are preprocessed to obtain complex InSAR interferograms; the complex InSAR interferograms are then divided into square-shaped interferogram patches along the azimuth and range directions.
[0010] Each interferogram patch is converted from the spatial domain to the frequency domain signal, and the corresponding spectral amplitude diagram is constructed.
[0011] The K-means clustering method is used to extract the initial low-pass filter from the spectral amplitude plot, and the phase filter is constructed together with the Kaiser window function.
[0012] The frequency domain signal of the interferogram patch is passed through a phase filter. The power of the signal frequency band is preserved, the noise at high frequency positions is filtered out, and the noise at low frequency positions is filtered out by a smoothing operation, resulting in a filtered frequency domain signal.
[0013] The filtered frequency domain signal is converted to the spatial domain to obtain the filtered interferogram patches;
[0014] By stitching together the interferogram patches, a filtered InSAR interferogram is obtained.
[0015] Furthermore, the complex InSAR interferogram slides in the azimuth and range directions with a preset step size to divide the interferogram patches; there is a preset overlap between adjacent interferogram patches.
[0016] Furthermore, the K-means clustering method divides the spectral amplitude map into two categories, specifically:
[0017] First, determine that the number of clusters is 2, and then randomly select 2 data points in the spectrum amplitude diagram as cluster centers;
[0018] Associate all data points in the spectrum graph with the nearest cluster center and use this as the basis for clustering; move the cluster center to the center of the current cluster containing all data points; repeat this process until the number of iterations or the change in cluster centers is less than a preset value, and finally the spectrum amplitude graph can be divided into two categories; assign the category with the larger amplitude value to 1 and the category with the smaller amplitude value to 0.
[0019] Furthermore, based on the clustering results of the spectral amplitude map, and combining geometric graph theory and statistical principles, the shape with the largest area and closed geometric shape in the class assigned a value of 1 is extracted; the geometric shape is used as the shape of the low-pass filter, with the pixel value inside the shape being 1, indicating that the signal passes through; the remaining pixels are assigned a value of 0, indicating that the signal is filtered out; the obtained geometric shape is used as the initial low-pass filter for adaptive extraction.
[0020] Furthermore, the Kaise window function is superimposed on the two-dimensional low-pass filter to obtain the final phase filter:
[0021] S{rect(|Z(u,v)|)}·K(n)
[0022] Where rect() represents the initial low-pass filter, S{·} represents Gaussian smoothing, Z(u,v) represents the frequency domain signal, u,v represent the horizontal and vertical frequency components during the two-dimensional Fourier transform, and K(n) represents the value of the two-dimensional Kaiser window function at time n.
[0023] Furthermore, in the Kaiser window function, the control parameter β is set to 10; the size of the two-dimensional Kaiser window function is consistent with the size of the interferogram patches.
[0024] Furthermore, the filtered frequency domain signal is represented as:
[0025] H(u,v)=S{rect(|Z(u,v)|)}·K(n)·Z(u,v)
[0026] Where H(u,v) represents the filtered frequency domain signal.
[0027] Furthermore, the filtered frequency domain signal is converted to the spatial domain using the inverse fast Fourier transform.
[0028] The edges of each interferogram patch obtained by transformation in the spatial domain are discarded with a preset overlap, and only the remaining parts are retained. The patches are then stitched together along the azimuth and range directions to obtain the filtered InSAR interferogram.
[0029] A terminal device includes a processor, a memory, and a computer program stored in the memory; when the processor executes the computer program, it implements the InSAR phase filtering method based on complex interferograms.
[0030] A computer-readable storage medium storing a computer program; when executed by a processor, the computer program implements the InSAR phase filtering method based on complex interferograms.
[0031] Compared with the prior art, the present invention has the following technical features:
[0032] 1. This invention, based on an InSAR phase filtering method using complex interferograms, does not rely on any external data. Instead, it fully utilizes the differences in noise and signal power, as well as frequency, within the spectrum of interferogram patches after Fourier transform to design a novel filter. The key difference between this new filter and the original Goldstein filter method and its improved versions lies in the design of the filter shape. The new filter changes the original frequency domain filter's limitation of amplifying or attenuating signals and noise to preserving the signal or completely filtering out noise, resulting in better filtering performance. The K-means clustering method is used to extract the filter shape, allowing for an adaptive filter shape and optimal filtering effect for each patch. The Kaiser window function is used to address the Gibbs phenomenon in the Fourier transform.
[0033] 2. This invention not only effectively solves one of the main difficulties currently limiting the further improvement of the accuracy of time-series InSAR technology, but also designs a filtering method from the perspective of the frequency domain, which is more in line with the nature of noise generation in synthetic aperture radar imaging. Furthermore, the algorithm is simple, computationally efficient, requires only the complex interferogram after interferometry, and is easy to implement and apply, thus contributing to the promotion and advancement of current time-series InSAR interferometric measurement technology and its engineering applications. Attached Figure Description
[0034] Figure 1 A schematic flowchart illustrating the InSAR phase filtering method for complex interferograms provided by this invention;
[0035] Figure 2 (a) shows the spectral signal after Fourier transform of the simulated interferogram patch, including signal and noise components; (b) shows the shape of the designed low-pass filter, as shown by the blue solid line, which can retain only the signal s. x part;
[0036] Figure 3 A flowchart illustrating the process of K-means clustering into two classes;
[0037] Figure 4 A schematic diagram for extracting the shape of a low-pass filter from a two-dimensional spectral image;
[0038] Figure 4 (a) is the two-dimensional spectrum image after Fourier transform; (b) is the clustering result obtained by K-means clustering based on the two-dimensional spectrum image, and the image is divided into two categories; (c) is the shape of the low-pass filter extracted according to geometric principles after classification, with red representing the signal passing through;
[0039] Figure 5 The Kaiser window function used in the filter of this invention;
[0040] Figure 5 (a) is a two-dimensional Kaiser window function; (b) is a one-dimensional Kaiser window function, and is the cross-section at the center frequency of (a);
[0041] Figure 6 (a) is the simulated noise-free interferogram phase; (b) is the simulated noisy interferogram phase.
[0042] Figure 7 (a) is the shape of the low-pass filter extracted from the spectrum of 16 interferogram patches uniformly selected along the azimuth and range directions from the simulated interferogram data; (b) is the InSAR phase map after filtering the 16 interferogram patches uniformly selected along the azimuth and range directions from the simulated interferogram data; (c) is the InSAR phase map after filtering the simulated interferogram data; and (d) is the scatter density map between the InSAR phase after filtering the simulated interferogram data and the noise-free InSAR phase. Detailed Implementation
[0043] This invention provides an InSAR phase filtering method based on complex interferograms, which can be applied to InSAR elevation measurement, time-series deformation monitoring and other fields.
[0044] See Figure 1 This invention provides a flowchart of an InSAR phase filtering method based on complex interferograms; the method includes the following steps:
[0045] Step 1: Preprocess the acquired primary and secondary SAR images, including registration, resampling, interferometry, removal of flat-ground phase, and phase correction of orbital errors caused by inaccurate orbital state data, so as to obtain complex InSAR interferograms of primary and secondary SAR images.
[0046] The complex InSAR interferogram is divided into square interferogram patches along the azimuth and range directions with a step size of 8 pixels. Each interferogram patch contains N×N pixels, where N is 32. It is also ensured that there is at least 75% overlap between each adjacent interferogram patch.
[0047] Step 2: Use a two-dimensional fast Fourier transform to convert each interferogram patch from the spatial domain to the frequency domain signal, and construct the corresponding spectrum amplitude map by taking the absolute value of the frequency domain signal.
[0048] like Figure 2 The orange solid line in (a) shows the signal and noise components on the spectrum; the signal and noise components can be represented by s respectively. x and n represent:
[0049] s = s x +n (1)
[0050] Where s is the frequency domain representation of the interferogram patch signal.
[0051] Based on two main differences between signal components and noise components on the spectrum amplitude diagram: 1) The amplitude of the signal component is generally greater than that of the noise component; 2) On the spectrum diagram, the signal component is always continuously located in the low-frequency region, that is, near the center frequency, while the noise component is located in the high-frequency region.
[0052] Based on these two characteristics, this invention designs an ideal low-pass filter shape rect(), such as... Figure 2 As shown in (b), it can be achieved solely through signal component s x Furthermore, since it is a two-dimensional image, the filter is also two-dimensional. Extracting the filter from the noise based on the spectrum can be transformed into a classification problem for two-dimensional images.
[0053] Step 3: Use the K-means clustering method to extract the initial low-pass filter from the spectral amplitude diagram, and combine it with the Kaiser window function to form the phase filter proposed in this invention; specifically including:
[0054] Step 31: Based on the spectral amplitude map obtained by fast Fourier transform of each interferogram patch, in order to extract the boundary between the signal component and the noise component in the spectral amplitude map and thus obtain the low-pass filter shape rect(), K-means clustering method is used to cluster in the spectral amplitude image, dividing the image into two classes and extracting the low-pass filter.
[0055] The K-means algorithm identifies each cluster by its cluster center, formalizing the clustering problem as minimizing the distance between sample points and cluster centers. Through alternating iterations of sample point allocation and cluster center update, it achieves geometric partitioning of the samples. In the sample point allocation phase, the cluster to which a data point belongs is determined based on the distance calculation results from all data points to the cluster center. In the cluster center update phase, the average value of all data points in that cluster is set as the new cluster center.
[0056] The flowchart of the K-means algorithm is as follows: Figure 3 As shown:
[0057] (1) First, determine the number of clusters K=2, and then randomly select 2 data points in the spectrum amplitude diagram as cluster centers.
[0058] (2) Associate all data points in the spectrum with the cluster center closest to them, and use this as the basis for clustering. The classification formula is:
[0059] C i =arg min‖x i - j || 2 (2)
[0060] In the formula, C i Represents the i-th data point x i The class closest to the cluster; arg is a marker indicating the class to which the sample parameter belongs; μ j Let j be the cluster center of the j-th cluster.
[0061] (3) Move the cluster center to the center (means) of the current partition cluster that contains all data points. The formula for calculating the cluster center of a data class is:
[0062]
[0063] Where l{·} represents the set of distances, and m represents the number of data points.
[0064] Repeat steps (2) and (3) N = 10 times; there are two termination conditions for the iteration: the number of iterations is reached or the cluster center no longer changes significantly. Finally, the spectrum amplitude diagram can be divided into two categories; the category with larger amplitude value is assigned a value of 1, and the category with smaller amplitude value is assigned a value of 0.
[0065] Step 32: Based on the classification results of Step 31, and combining geometric graph theory and statistical principles, extract the shape with the largest area and closed geometric shape in the class that is assigned a value of 1; the geometric shape is used as the shape of the low-pass filter, the pixel value inside the shape is 1, indicating that the signal passes through; the other pixels are assigned a value of 0, indicating that the signal is filtered out; the obtained geometric shape is used as the initial low-pass filter rect() for adaptive extraction.
[0066] Figure 4 This is a schematic diagram of the initial low-pass filter rect() extraction; four 32×32 pixel interferogram patches are randomly selected from the real interferogram data, and their two-dimensional Fourier transforms are performed to obtain two-dimensional spectra, as shown below. Figure 4As shown in (a); (b) is the clustering result of the K-means method on the spectrogram, which is divided into two categories; only two values are assigned to the two-dimensional image: 1 for signal regions and 0 for non-signal regions; in addition, it can be seen that the signal is discontinuous in the clustering result; based on the above, the geometry of the largest continuous region in the shape of the signal component in (b) is extracted again as the initial low-pass filter shape rect(), and the pixels outside the geometry are assigned 0, as shown in (a). Figure 4 As shown in (c).
[0067] Step 33: Generate a two-dimensional Kaiser window function.
[0068]
[0069] Where K(n) represents the value of the two-dimensional Kaiser window function at time n; N is the length of the window function, and in this embodiment, n = 32 represents the size of the window function; I0 represents the first-type modified Bessel function, β is the parameter controlling the window function, and in this embodiment, β = 10; I0[β] represents the value of the first-type modified Bessel function at β, and the size of the two-dimensional window function is consistent with the size of the divided interferogram patches, which is 32×32 pixels in this embodiment, and K(n) is as follows: Figure 5 As shown.
[0070] Step 33: The Kaiser window function can suppress the Gibbs phenomenon caused by signal transitions. Superimposing the Kaiser window function onto the two-dimensional low-pass filter yields the final phase filter.
[0071] S{rect(|Z(u,v)|)}·K(n) (5)
[0072] Where rect() represents the initial low-pass filter, S{·} represents Gaussian smoothing, Z(u,v) represents the frequency domain signal, and u and v represent the horizontal and vertical frequency components during the two-dimensional Fourier transform, respectively.
[0073] Step 4: The frequency domain signal of the interferogram patch is passed through a phase filter composed of an initial low-pass filter and a Kaiser window function. The signal power in the frequency band will be preserved, noise at high frequencies will be filtered out, and noise at low frequencies will be filtered out by a smoothing operation.
[0074] H(u,v)=S{rect(|Z(u,v)|)}·K(n)·Z(u,v) (6)
[0075] Where H(u,v) represents the filtered frequency domain signal.
[0076] Step 5: Perform an inverse fast Fourier transform on the filtered frequency domain signal H(u,v) to obtain the filtered interferogram patches in the spatial domain.
[0077] Step 6: Stitch together the interferogram patches to obtain the filtered InSAR interferogram.
[0078] Step 61: Based on the fact that the interferogram patches have 75% overlapping pixels, after the inverse Fourier transform, 75% of the pixels at the edge of the patch are discarded, and only the pixels at the center of the patch are retained, that is, the small patch of 8×8 pixels at the center of the patch.
[0079] Step 62: Using the retained center pixels of the patches, stitch the patches together along the azimuth and range directions to obtain the filtered InSAR interferogram.
[0080] Example:
[0081] In one embodiment of the present invention, the specific implementation steps based on simulated data are as follows:
[0082] Step 1: Simulate and generate a 512×512 pixel mountain terrain using the `peak` function in MATLAB. Based on this terrain, reconstruct the InSAR phase stripes, such as... Figure 6 As shown in (a), the amplitude of the interferogram is uniformly assigned a value of 1; then, Gaussian noise of size 512×512 pixels with a standard deviation of 0.8 (rad) is generated and superimposed on the noise-free interferogram to simulate the noisy observation InSAR phase, as shown in (a). Figure 6 As shown in (b).
[0083] The simulated complex InSAR interferogram is divided into several 32×32 pixel interferogram patches with a step size of 8 pixels in both the azimuth and range directions.
[0084] Step 2: For each square interferogram patch, use a two-dimensional fast Fourier transform to convert the spatial domain interferogram patch signal into a frequency domain signal, and take the absolute value of the frequency domain signal to obtain a two-dimensional image with the spectral amplitude as the value.
[0085] Step 3: For each patch's spectral image, K-means clustering is used to divide the spectral amplitude image into two classes. Then, a closed classification shape with the largest continuous area at the center is extracted using geometry. Pixels inside the closed shape are assigned a value of 1, and pixels outside the shape are assigned a value of 0, serving as a low-pass filter. Figure 7 As shown in (a), the filter shape extracted after Fourier transform of interferogram patches uniformly selected along the azimuth and range directions in the simulation data is displayed.
[0086] The two-dimensional Kaiser window function is generated according to formula (4), and the two-dimensional low-pass filter and the Kaiser window function are combined to form the filter proposed in this invention according to formula (5).
[0087] Step 4: Based on formula (6), the Fourier transform spectrum signal is passed through the proposed filter to obtain the filtered spectrum signal; in the filtered spectrum signal, high-frequency noise is filtered out by the low-pass filter, low-frequency noise is filtered out by the smoothing operation, and the Gibbs phenomenon at the signal transition is suppressed by the Kaiser window function.
[0088] Step 5: Perform an inverse fast Fourier transform on the frequency domain signal to obtain filtered complex interferogram patches in the spatial domain; perform a phase take operation on the complex interferogram to obtain the filtered InSAR interferogram phase for each patch, as shown below. Figure 7 As shown in (b), the InSAR phase is displayed after filtering of interferogram patches uniformly selected along the azimuth and range directions in the simulated data.
[0089] Step 6: Based on the initial segmentation method of the interferogram patches, stitch the patches together to form a filtered interferogram. Since there is a 75% pixel overlap during segmentation, only an 8×8 pixel center patch is retained from each patch during stitching, and stitching is performed along the azimuth and range directions. Taking the phase of the stitched complex interferogram yields the filtered InSAR phase, as shown below. Figure 7 As shown in (c), the filtered phase is close to the noise-free InSAR phase, which also demonstrates that the filter method proposed in this invention effectively filters out the presence of noise.
[0090] To demonstrate the reliability of the filtered InSAR phase, it is compared with the noise-free true InSAR phase, generating a scatter plot, such as... Figure 7 As shown in (d), it can be seen that R 2 =0.997 indicates that the filtered InSAR phase is consistent with the true InSAR phase distribution; and the RMSE of 0.117 (rad) indicates that the filtering effect is good. These also demonstrate the reliability of the InSAR phase filtering method based on complex interferograms proposed in this invention.
[0091] In summary, this invention provides a novel frequency-domain InSAR phase filtering method for reducing phase noise in InSAR data. The invention analyzes the differences between noise in the interferogram signal and the spectrogram, demonstrating the presence and characteristics of various noise types, and proposes a new filter shape to effectively separate noise from the signal. Through the filtering process, the new filter shape aims to preserve both signal and noise, rather than amplifying or attenuating them. Importantly, our new filtering method eliminates the need for iterative filtering, thus achieving faster operation. Finally, we validate the performance of the filtering method using a simulated dataset. It can be seen that this method is simple and effective to implement, and has significant implications for large-scale, high-volume time-series InSAR deformation monitoring and topographic surveying.
[0092] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.
Claims
1. A method of InSAR phase filtering based on complex interferograms, characterized in that, include: The acquired primary and secondary SAR images are preprocessed to obtain complex InSAR interferograms; the complex InSAR interferograms are then divided into square-shaped interferogram patches along the azimuth and range directions. Each interferogram patch is converted from the spatial domain to the frequency domain signal, and the corresponding spectral amplitude diagram is constructed. The K-means clustering method is used to extract the initial low-pass filter from the spectral amplitude plot, and the phase filter is constructed together with the Kaiser window function. The frequency domain signal of the interferogram patch is passed through a phase filter. The power of the signal frequency band is preserved, the noise at high frequency positions is filtered out, and the noise at low frequency positions is filtered out by a smoothing operation, resulting in a filtered frequency domain signal. The filtered frequency domain signal is converted to the spatial domain to obtain the filtered interferogram patches; By stitching together the interferogram patches, a filtered InSAR interferogram is obtained.
2. The complex-interferogram-based InSAR phase filtering method according to claim 1, characterized in that, Complex InSAR interferograms slide in the azimuth and range directions with preset step sizes to divide the interferogram patches; there is a preset overlap between adjacent interferogram patches.
3. The complex interferogram based InSAR phase filtering method of claim 1, wherein, The K-means clustering method divides the spectral amplitude map into two categories, specifically: (1) First, determine that the number of clusters is 2, and then randomly select 2 data points in the spectrum amplitude diagram as cluster centers; (2) Associate all data points in the spectrum with the cluster center closest to them, and use this as the basis for clustering; (3) Move the cluster center to the center of the current partition cluster that contains all data points; Repeat steps (2) and (3) until the number of iterations or the change in cluster center is less than the preset value. Finally, the spectrum amplitude map can be divided into two categories. The category with larger amplitude value is assigned a value of 1, and the category with smaller amplitude value is assigned a value of 0.
4. The InSAR phase filtering method based on complex interferograms according to claim 1, characterized in that, Based on the clustering results of the spectral amplitude diagram, and combining geometric graph theory and statistical principles, the shape with the largest area and closed geometric shape in the class with a value of 1 is extracted; the geometric shape is used as the shape of the low-pass filter, and the pixel value inside the shape is 1, indicating that the signal passes through; The remaining pixels are assigned a value of 0, indicating that the signal has been filtered out; The resulting geometry is used as the initial low-pass filter for adaptive extraction.
5. The InSAR phase filtering method based on complex interferograms according to claim 1, characterized in that, The final phase filter is obtained by superimposing the Kaiser window function onto the two-dimensional low-pass filter: in, This represents the initial low-pass filter. Indicates Gaussian smoothing; Represents a frequency domain signal; These represent the frequency components in the horizontal and vertical directions during the two-dimensional Fourier transform, respectively. Indicates the Kaiser window function in n The value at time.
6. The InSAR phase filtering method based on complex interferograms according to claim 1, characterized in that, Control parameters of the Kaiser window function The value is 10; the size of the Kaiser window function is consistent with the size of the interferogram patches.
7. The InSAR phase filtering method based on complex interferograms according to claim 5, characterized in that, The filtered frequency domain signal is represented as follows: in, This represents the filtered frequency domain signal.
8. The InSAR phase filtering method based on complex interferograms according to claim 1, characterized in that, The filtered frequency domain signal is converted to the spatial domain using the inverse fast Fourier transform. The edges of each interferogram patch obtained by transformation in the spatial domain are discarded with a predetermined degree of overlap, and only the remaining parts are retained. These parts are then stitched together along the azimuth and range directions to obtain the filtered InSAR interferogram.
9. A terminal device, comprising a processor, a memory, and a computer program stored in the memory; characterized in that, When the processor executes the computer program, it implements the InSAR phase filtering method based on complex interferograms as described in any one of claims 1-8.
10. A computer-readable storage medium storing a computer program; characterized in that, When the computer program is executed by the processor, it implements the InSAR phase filtering method based on complex interferograms as described in any one of claims 1-8.