A multi-baseline time-series InSAR fine-scale DEM generation method and system

Abstract

The application provides a multi-baseline time-series InSAR fine-scale DEM generation method and system, signal separation is carried out by using space ICA technology, and atmospheric and noise are directly removed, so that the interference of atmospheric and noise is weakened; baseline error in phase observation value is removed by using polynomial fitting, accurate differential interference phase is obtained; the influence of linear deformation, periodic deformation and environmental factors is comprehensively considered, a time-series InSAR physical model between differential interference phase and terrain and deformation is constructed, accurate terrain and deformation parameters are obtained, and the generation of fine-scale DEM is realized. The time-series InSAR environmental modeling scheme based on ICA separation combined with multi-baseline error correction provided by the application can provide an important reference for the generation of large-area fine-scale DEM.

Description

Technical Field

[0001] This invention belongs to the field of photogrammetry and remote sensing technology, and specifically relates to a method and system for generating fine-scale DEMs using multi-baseline temporal InSAR. Background Technology

[0002] Digital Elevation Models (DEMs) are of great value in national economic development, disaster prevention and mitigation, urban operation and management, and military fields. They are an important component of spatial information service technology and remote sensing technology, playing a crucial role in geographic information science and remote sensing science, including topographic analysis and land use change. Therefore, the generation of high-scale DEMs is of significant research importance.

[0003] Traditional effective methods for DEM acquisition mainly include topographic map digitization, ground surveying, aerial photogrammetry, and 3D laser scanning. While these techniques can meet the accuracy requirements of DEM generation, their application in large-scale, fine-scale DEM generation is limited by factors such as low point density, limited monitoring range, and poor spatiotemporal resolution. Multi-Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) technology has been proven to be an effective method for high-precision DEM acquisition. It has advantages such as all-weather, all-day operation and immunity to cloud, rain, and fog obstruction. This technology is currently an important remote sensing tool for generating large-scale DEM products and has broad application prospects in the field of high-resolution, high-precision DEM mapping worldwide.

[0004] With the continuous development of multi-platform, multi-orbit high-resolution satellites, the accuracy of spaceborne InSAR-generated DEMs has been greatly improved. However, due to limitations in its observation modes and side-looking imaging, the following limitations still exist in the generation of fine-scale DEMs using multi-baseline InSAR technology: First, atmospheric and noise factors in InSAR signal processing during repeated orbit modes inevitably introduce errors into the acquired phase observations, which are then propagated to subsequent DEM generation. Second, there is an inherent contradiction between the vertical baseline length and the reliability of phase unwrapping in the InSAR altimetry principle; how to reduce interference fringes caused by baseline errors in the interferogram is a current research hotspot and challenge. Finally, the phase signals acquired by InSAR contain both deformation and elevation information. Most multi-baseline InSAR methods for generating fine-scale DEMs use a single purely mathematical empirical model for fitting or directly convert the phase to elevation, ignoring the influence of deformation factors. When the deformed phase cannot be accurately separated from the interferogram, errors are introduced into the DEM, limiting the accuracy of multi-baseline DEM generation. In summary, a multi-baseline time-series InSAR method for generating fine-scale DEMs is urgently needed to address the problems existing in current technologies.

[0005] Yang et al. analyzed the quality of generated DEMs based on geostationary orbit InSAR data, comprehensively analyzing the potential error impacts of InSAR technology on DEM generation from factors such as slant range error, spatial baseline error, and baseline inclination error. Du et al. evaluated the accuracy of commonly used DEM error calculation models using simulation data. Dong et al. studied the DEM of Songshan Mountain in Henan Province using a cascaded multi-baseline interferometry method, obtaining accurate mountain DEM products. Scholars have made improvements to the shortcomings of InSAR-generated DEMs, successfully obtaining relatively accurate DEM products. However, DEM generation methods that comprehensively consider the impact of multiple error factors still require further research. In this invention, firstly, ICA technology is used to separate signals based on the spatiotemporal distribution characteristics of the atmosphere and noise, and these signals are directly removed from the phase signal, effectively reducing atmospheric and noise interference. This method does not rely on external atmospheric data or complex atmospheric models, making it widely applicable. Secondly, based on the analysis of the impact of baseline errors on InSAR-generated DEMs, a polynomial fitting method is used to remove baseline errors from phase observations, improving the accuracy of multi-baseline unwrapping phase. Finally, in the deformation modeling process, the influence of linear deformation, seasonal deformation, and environmental factors is comprehensively considered. Using multi-baseline data, a time-series InSAR model that takes environmental parameters into account is established based on the functional relationship between deformation, terrain, and phase. The terrain and deformation parameters are then solved, improving the accuracy of InSAR modeling and the precision of DEM acquisition. Summary of the Invention

[0006] This invention improves the generation of DEM products from InSAR by utilizing ICA phase separation, multi-baseline error correction, and temporal InSAR environment modeling. The invention aims to enhance the accuracy of DEMs generated by InSAR technology and proposes a multi-baseline temporal InSAR fine-scale DEM generation scheme to further improve the accuracy of InSAR-generated fine-scale DEMs.

[0007] The technical solution adopted in this invention is a multi-baseline time-series InSAR fine-scale DEM generation method. It utilizes space ICA technology to separate the signal and directly remove the atmosphere and noise, reducing the interference of the atmosphere and noise. It uses polynomial fitting to remove the baseline error in the phase observation values ​​and obtain accurate differential interferometric phase. It integrates the influence of linear deformation, periodic deformation and environmental factors to construct a time-series InSAR physical model between differential interferometric phase and terrain and deformation, obtains accurate terrain and deformation parameters, and realizes the generation of fine-scale DEM.

[0008] Moreover, the implementation process includes the following steps:

[0009] Step 1, generation of differential interferometric phase, includes generating coherence coefficient maps and differential interferograms based on InSAR using preprocessing of previous data, and generating a time-series differential interferometric phase matrix based on high coherence points;

[0010] Step 2, ICA-based removal of atmospheric and noise phases, including using FastICA to separate the generated differential interference phases and directly remove atmospheric and noise;

[0011] Step 3, polynomial fitting to remove baseline error, includes fitting the baseline error of the differential interferometric phase after removing atmosphere and noise using a polynomial fitting method, and subtracting the fitted baseline error from the differential interferometric phase to obtain the accurate differential interferometric phase.

[0012] Step 4: Construct a temporal InSAR physical model between differential interferometric phase and terrain and deformation, including comprehensively considering the effects of linear deformation, periodic deformation and environmental factors in the process of multi-baseline InSAR modeling, and constructing a temporal InSAR physical model that takes into account environmental parameters.

[0013] Step 5, optimal estimation of terrain and deformation parameters, including solving the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters;

[0014] Step 6, generation of fine-scale DEM, including generating fine-scale DEM using the relationship between the solved terrain parameters and the actual surface elevation;

[0015] Step 7, accuracy evaluation of fine-scale DEM, including calculating the error between the DEM generated in steps 1-6 and the actual surface elevation data, and evaluating the accuracy of the generated fine-scale DEM;

[0016] Step 8, mapping of large-scale fine-scale DEM, including generating DEM products that match the actual geographical location by geocoding and other steps from the acquired fine-scale DEM.

[0017] Furthermore, the generation of the differential interferometric phase involves data format conversion, baseline estimation and connection diagram generation, differential interferometric phase generation, selection of high coherence points, and the resulting unwrapped phase is represented as follows:

[0018]

[0019] In the formula, Let M represent the unwrapped phase at the i-th pixel (x,y) on the m-th differential interferogram, where 1≤m≤M; λ represents the deformation phase; λ represents the radar wavelength. This represents the deformation component along the line of sight upwards in the m-th differential interferogram; Indicates terrain phase θ and B m denoted by and respectively, representing the radar incident angle and the length of the vertical baseline of the m-th interferometric pair; R represents the distance between the coherent target and the location of the radar satellite; ΔH(x,y) is the elevation error; Indicates orbital error; This represents the residual phase, primarily including the atmospheric delay phase. noise phase (and Qualcomm) deformation components

[0020] Moreover, the method for removing atmospheric and noise phase based on ICA is to use FastICA to separate the time series phase signal, directly remove the atmospheric and noise-related components, and extract the important components.

[0021] The InSAR time series phase signal is represented as:

[0022]

[0023] In the formula, X represents the time series mixed signal, which is regarded as the InSAR time series phase matrix. A represents the differential interference phase; A represents the spatial response, which is the combination of the spatial responses of all independent components. SR represents the spatial response of each independent component, where n is the number of independent components; S n×N = [IC1 IC2…IC n ]T The time series represents all independent components, including deformation phase, terrain phase, atmosphere, noise, and other residual phase signals; IC represents the time series of each independent component. and These represent the phase signals related to deformation and terrain, respectively.

[0024] Phase after ICA separation Represented as:

[0025]

[0026] In the formula, and These represent the spatial responses corresponding to the deformation component and the topography component, respectively; and These represent the time series corresponding to the deformation component and the topographic component, respectively.

[0027] Moreover, the method of removing baseline error by polynomial fitting is as follows: peak elimination; approximating the baseline using polynomial fitting to obtain the baseline error; and subtracting the baseline error from the phase signal obtained by ICA separation to obtain the corrected phase signal.

[0028] The baseline-corrected phase signal is represented as follows:

[0029]

[0030] In the formula, Indicates the phase after baseline correction. This represents the baseline error of the polynomial fit.

[0031] Moreover, the implementation of the temporal InSAR physical model between differential interferometric phase and terrain and deformation is as follows: constructing a surface deformation model; constructing a temporal InSAR physical model between differential interferometric phase and linear, periodic, environmental and terrain parameters;

[0032] The temporal InSAR physical model is represented as follows:

[0033]

[0034] In the formula, t A and t B v represents the main image of the m-th differential interferogram and the time of acquisition from the image, respectively. k t represents the linear deformation rate between any two images; the linear deformation rate varies at different times. k and t k-1α1 and α2 represent the time corresponding to the linear deformation rate at different times and the time of the previous time; α1 and α2 represent the periodic deformation coefficients, which are considered unknown parameters; T represents the time of one year; α3 and α4 represent the temperature deformation coefficient and the rainfall deformation coefficient, which are considered unknown parameters; Tem and Pre represent the monthly average temperature and the monthly average rainfall, respectively.

[0035] Furthermore, the fine-scale DEM is generated by utilizing the relationship between the solved terrain parameters and the actual surface elevation. The elevation at any point (x, y) on the ground is represented as follows:

[0036] Z(x,y)=H ref (x,y)+ΔH(x,y)

[0037] In the formula: Z(x,y) represents the actual elevation at any point (x,y), and H... ref (x,y) represents the elevation of the reference external DEM, and ΔH(x,y) represents the residual elevation error obtained after ICA phase separation, multi-baseline error correction, and time-series InSAR environment modeling.

[0038] On the other hand, the present invention also provides a multi-baseline temporal InSAR fine-scale DEM generation system for implementing the multi-baseline temporal InSAR fine-scale DEM generation method described above.

[0039] Moreover, it includes the following modules,

[0040] The first module is used for the generation of differential interferometric phase, including generating coherence coefficient maps and differential interferograms based on InSAR using preprocessing of previous data, and generating a time-series differential interferometric phase matrix based on high coherence points;

[0041] The second module is used for ICA-based removal of atmospheric and noise phases, including using FastICA to separate the generated differential interferometric phases and directly remove atmospheric and noise.

[0042] The third module is used for polynomial fitting to remove baseline errors. This includes fitting the baseline error of the differential interferometric phase after removing atmospheric and noise using a polynomial fitting method, and subtracting the fitted baseline error from the differential interferometric phase to obtain an accurate differential interferometric phase.

[0043] The fourth module is used to construct a temporal InSAR physical model between differential interferometric phase and terrain and deformation. This includes comprehensively considering the effects of linear deformation, periodic deformation and environmental factors during the multi-baseline InSAR modeling process, and constructing a temporal InSAR physical model that takes environmental parameters into account.

[0044] The fifth module is used for optimal estimation of terrain and deformation parameters, including solving the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters.

[0045] The sixth module is used for generating fine-scale DEMs, including generating fine-scale DEMs by utilizing the relationship between the solved terrain parameters and the actual surface elevation.

[0046] The seventh module is used for the accuracy evaluation of fine-scale DEMs, including calculating the error between the generated DEM and the actual surface elevation data, and evaluating the accuracy of the generated fine-scale DEMs.

[0047] The eighth module is used for mapping large-scale, fine-scale DEMs, including generating DEM products that match the actual geographical location by taking steps such as geocoding from the acquired fine-scale DEMs.

[0048] Alternatively, it may include a processor and a memory, with the memory used to store program instructions and the processor used to call the stored instructions in the memory to execute a multi-baseline time-series InSAR fine-scale DEM generation method as described above.

[0049] The present invention provides a multi-baseline time-series InSAR fine-scale DEM generation method that improves upon the shortcomings of traditional multi-baseline InSAR DEM generation processes, thereby further enhancing the accuracy of DEM products generated by InSAR technology.

[0050] The application of the technical solution of the present invention has the following beneficial effects:

[0051] (1) In this invention, the InSAR phase signal is separated by using ICA technology, and the atmospheric and noise phase is directly removed. ICA technology can generate a clear signal source for the differential interferometric phase of InSAR. The various signals are statistically independent of each other. The independent components are separated from the time series, and the important components are extracted, which effectively reduces the interference of the atmosphere and noise. Moreover, it does not rely on external atmospheric data and complex atmospheric models, making it widely applicable.

[0052] (2) In this invention, a polynomial fitting method is used to remove baseline errors in the InSAR phase signal. An adaptive judgment condition for the polynomial fitting residual is established to approximate and correct the baseline error in the original phase signal, thereby obtaining accurate InSAR phase observations and improving the accuracy of multi-baseline unwrapped phase.

[0053] (3) In this invention, a time-series InSAR physical model is constructed that relates differential interferometric phase to terrain and deformation. During the modeling process, the effects of linear deformation, periodic deformation, and environmental factors are comprehensively considered. Using multi-baseline data, a time-series InSAR physical model that takes into account environmental parameters is established based on the functional relationship between deformation, terrain, and phase. Then, the terrain and deformation parameters are solved to improve the accuracy of InSAR modeling and the precision of obtaining residual elevation errors.

[0054] (4) In this invention, the elevation information of any point on the ground can be reconstructed by adding the elevation error obtained by InSAR to the elevation of an external reference DEM. InSAR has the advantages of wide monitoring range, high spatial resolution, and non-contact measurement, which greatly makes up for the shortcomings of traditional methods in the process of generating DEM and provides a more effective means for obtaining fine-scale surface elevation. This application uses multi-baseline time-series InSAR technology to generate fine-scale DEM products, which broadens the application field of InSAR technology and can be used for the generation of large-scale fine-scale DEMs, providing important theoretical support for the field of global high-resolution and high-precision DEM mapping.

[0055] The present invention is simple and convenient to implement, highly practical, and solves the problems of low practicality and inconvenience in actual application of related technologies. It can improve user experience and has significant market value. Attached Figure Description

[0056] Figure 1 This is a flowchart of a multi-baseline temporal InSAR fine-scale DEM generation method in an embodiment of the present invention.

[0057] Figure 2 This is a flowchart illustrating the removal of atmospheric and noise phase using ICA technology in step 2 of an embodiment of the present invention.

[0058] Figure 3 This is a flowchart of the polynomial fitting process for removing baseline error in step 3 of an embodiment of the present invention. Detailed Implementation

[0059] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0060] This invention addresses the unavoidable errors caused by atmospheric and noise factors in the acquisition of phase observations during traditional multi-baseline time-series InSAR DEM generation; the inherent contradiction between the vertical baseline length and the reliability of phase unwrapping in InSAR altimetry principles; and the fact that InSAR-acquired phase signals simultaneously contain deformation and elevation information, while most multi-baseline InSAR DEM generation processes employ a single, purely mathematical empirical model, neglecting the influence of deformation factors and limiting the accuracy of the generated DEM. To resolve these issues, this invention utilizes spatial ICA technology to separate the signal and directly remove atmospheric and noise interference, effectively reducing their impact. It employs polynomial fitting to remove baseline errors in phase observations, improving the accuracy of multi-baseline unwrapping phase analysis. Furthermore, the deformation modeling process comprehensively considers the effects of linear deformation, seasonal deformation, and environmental factors, enhancing the accuracy of InSAR DEM generation and obtaining large-scale, high-precision, fine-scale DEM products.

[0061] See Figure 1 The present invention provides a method for generating a fine-scale DEM using multi-baseline temporal InSAR, comprising the following steps:

[0062] Step 1: Generation of differential interference phase.

[0063] Step 1.1: Data Format Conversion. The data format of multiple SAR (Synthetic Aperture Radar) satellite data (i.e., N+1 SAR images) is batch converted to generate N+1 single-look complex images (SLC) that can be read by processing software. Pre-filtering of the initial images is performed to reduce noise and phase distortion. During the format conversion process, a corresponding parameter file is generated for each image.

[0064] By convention, N refers to the number of images acquired minus 1. When N=1, it is called D-InSAR, which uses differential interferometry between two images to acquire deformation and DEM, resulting in a larger error. When N>1, it is called multi-baseline MT-InSAR, which acquires temporal deformation and DEM, increases redundant observations, and is beneficial for the estimation and removal of atmospheric phase. In practice, the value of N can be set according to the specific situation. Generally, the larger the value, the more data is acquired and the better the results are.

[0065] Step 1.2: Baseline estimation and connectivity map generation. Baseline estimation is performed on all images, and then one image is selected as the super master image for all images. Spatiotemporal thresholds are set, and all interferometric pairs that meet the set parameters are generated. Based on this, all images are registered and resampled to the super master image. During the registration process, a polynomial fitting method is used for registration. Interferograms are generated using the generated interferometric pairs, and coherence coefficient maps and intensity maps are generated concurrently during this process.

[0066] Step 1.3: Generate differential interferometric phase. Perform the following operations on the interferogram: remove the flat phase and track error using precise orbit data; remove the terrain phase using external 12-meter resolution DEM (Digital Elevation Model) data. ref (x,y) represents the elevation at any point in the external DEM data (x,y); then residual phase filtering is performed to remove noise from the image; a differential interferogram is generated, and the differential interferogram is unwrapped in phase (some interferometric pairs are unwrapped in 3D, and those that cannot be unwrapped in 3D are unwrapped using the minimum cost flow method), generating the unwrapped temporal differential interferometric phase;

[0067] Step 1.4: Selection of High Coherence Points. High coherence points are extracted using the intensity, amplitude deviation, and coherence coefficient methods. The corresponding values ​​of these high coherence points are then extracted from the temporal differential interferometric phase, generating a temporal differential interferometric phase matrix based on these high coherence points. Assuming N+1 SAR images of the same area are acquired, generating a total of M multi-look differential interferograms, the unwrapped phase at the i-th pixel (x,y) in the m-th (1≤m≤M) differential interferogram is... It can be represented as:

[0068]

[0069] In the formula: λ represents the deformation phase; λ represents the radar wavelength. The deformation component that represents the upward line of sight (LOS) of the i-th pixel in the m-th differential interferogram is called the low-pass (LP) deformation component. Indicates terrain phase θ and B m denoted by and respectively, representing the radar incident angle and the length of the vertical baseline of the m-th interferometric pair; R represents the distance between the coherent target and the location of the radar satellite; ΔH(x,y) is the elevation error; Indicates orbital error; This represents the residual phase, primarily including the atmospheric delay phase. noise phase High-Pass (HP) deformation components

[0070] Step 2: Removal of atmospheric and noise phase using ICA (Independent Component Analysis) technology. See the detailed flowchart for a schematic diagram. Figure 2In this embodiment, after obtaining the time series phase matrix of the study area, FastICA (Fast Independent Component Analysis) is used to separate the time series phase signal. The InSAR time series phase signal can be expressed as:

[0071] X = A × S (2)

[0072] In the formula: X represents the time series mixed signal, which can be regarded as the InSAR time series phase matrix. This represents the differential interferometric phase, and N represents the number of acquired images minus 1, i.e. Let A represent the temporal phases at time 1, time 2, and time n, respectively; let A represent the spatial response, which is the combination of the spatial responses of all independent components. SR represents the spatial response of each independent component, where n is the number of independent components, i.e. Let S represent the spatial response of the first independent component, the second independent component, and so on up to the nth independent component; S represents the time series of all independent components, including deformation phase, terrain phase, atmospheric phase, noise, and the remaining residual phase signals, which can be expressed as:

[0073]

[0074] In the formula: IC represents the time series of each independent component, i.e., IC1 IC2…IC n Let represent the time series of the 1st independent component, the 2nd independent component, and so on up to the nth independent component. These represent the deformation phase, terrain phase, atmospheric phase, noise phase, and remaining residual phase at time 1, respectively. These represent the deformation phase, terrain phase, atmospheric phase, noise phase, and remaining residual phase at the second time point, respectively. Let N represent the deformation phase, terrain phase, atmospheric phase, noise phase, and remaining residual phase at time N; at this time, the InSAR time series phase signal can be expressed as:

[0075]

[0076] In the formula: and Let represent the phase signals related to deformation and terrain, respectively. For the phase signals acquired by InSAR, the various signals in equation (4) are statistically independent. Therefore, the FastICA algorithm can be used to separate the independent components from the time series, directly removing the atmospheric and noise-related components and extracting the important components. At this time, the phase signals after ICA separation are... It can be represented as:

[0077]

[0078] In the formula: and These represent the spatial responses corresponding to the deformation component and the topography component, respectively; and These represent the time series corresponding to the deformation component and the topographic component, respectively.

[0079] Step 3: Polynomial fitting to remove baseline error. See the flowchart for details. Figure 3 .

[0080] Step 3.1: Peak Removal. The least-squares smoothing filter function (Savitzky-Golay filtering, SGolayfilt) is used to remove peak values ​​from the differential interference phase matrix;

[0081] Step 3.2: Approximate the baseline using polynomial fitting to obtain the baseline error. The polynomial fitting residual can be expressed as:

[0082]

[0083] In the formula: This represents the baseline error of the i-th pixel in the m-th difference interferogram obtained by polynomial fitting; The average baseline error of all pixels after fitting is represented; the adaptive criterion for the polynomial fitting residual can be expressed as:

[0084]

[0085] Step 3.3: Subtract the baseline error from the phase signal obtained by ICA separation to obtain the corrected phase signal. It can be represented as:

[0086]

[0087] In the formula, Indicates the phase after baseline correction; This represents the baseline error of the polynomial fit.

[0088] Step 4: Construct a temporal InSAR physical model relating differential interferometric phase to terrain and deformation.

[0089] Step 4.1: Construction of the Surface Deformation Model. After obtaining accurate InSAR phase observations, it is crucial to construct a functional model of the relationship between differential interferometric phase and topography and deformation. Surface deformation exhibits significant linear, periodic, and seasonal variations over time. Multi-rate model. It can be represented as:

[0090]

[0091] In the formula: t A and t B These represent the main image of the m-th differential interferogram and the time of acquisition from the image, respectively. k t represents the linear deformation rate between any two images; the linear deformation rate varies at different times. k and t k-1 These represent the time corresponding to the linear deformation rate at different times and the time of the previous time step. Periodic model. It can be represented as:

[0092]

[0093] In the formula: α1 and α2 represent the periodic deformation coefficients, which can be considered as unknown parameters; T represents one year, which can be considered as 365 days. The environmental model can be expressed as:

[0094]

[0095] In the formula: α3 and α4 represent the temperature deformation coefficient and the rainfall deformation coefficient, respectively, and can be considered as unknown parameters; Tem and Pre represent the monthly average temperature and the monthly average rainfall, respectively. Therefore, the surface deformation model can be expressed as:

[0096]

[0097] Step 4.2: Construct a time-series InSAR physical model relating the differential interferometric phase to linear, periodic, environmental, and terrain parameters, which can be expressed as:

[0098]

[0099] Step 5: Optimal estimation of terrain and deformation parameters. Solve the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters. The known parameters Tem and Pre in equation (13) can be obtained based on external environmental parameters. The physical model equation (13) can be written in the form V = f(X) - L, where f(X) = AX is the model parameter X = [v1v2 L v N The function is α1 α2 α3 α4 ΔH(x,y)], where A is an m×i coefficient matrix, L is an m×1 observation matrix, and V is an m×1 correction matrix. Based on the least squares adjustment theory, a dynamic optimization iterative algorithm model as shown in equation (14) can be constructed:

[0100]

[0101] in, It is the adjustment standard measurement of X, where V is an m×1 correction matrix.T Let V be the transpose of V, and P be the weight matrix in the least squares principle.

[0102] Step 6: Generating a fine-scale DEM. A fine-scale DEM is generated using the relationship between the solved terrain parameters and the actual surface elevation. The elevation of any point (x, y) on the ground can be represented as:

[0103] Z(x,y)=H ref (x,y)+ΔH(x,y) (15)

[0104] In the formula: Z(x,y) represents the actual elevation at any point (x,y), and H... ref (x,y) represents the elevation of the external DEM in step 1.3, and ΔH(x,y) represents the residual elevation error obtained after ICA phase separation, multi-baseline error correction, and time-series InSAR environment modeling.

[0105] Step 7: Accuracy evaluation of the fine-scale DEM. Calculate the R-value between the DEM generated in steps 1-6 and the actual surface elevation data. 2 Root mean square error (RMSE), mean absolute error (MAE), and maximum elevation error (E) max The accuracy of the generated fine-scale DEM is evaluated using methods such as [evaluation of these methods]. The relevant calculation formulas are shown below:

[0106]

[0107] In the formula: R 2 The coefficient of determination represents the model's predictive power. The elevation values ​​of the DEM generated by multi-baseline InSAR technology are represented by j; the number of DEM pixels is represented by y. i This represents the actual value of the ground surface elevation; R represents the average value of the observed data. 2 The numerical range of R is 0 to 1. 2 The larger the value, the better the generated DEM matches the actual surface elevation.

[0108]

[0109] The parameters in the formula are the same as above. The RMSE value represents the deviation between the fitted value and the true value to a certain extent. The smaller the value, the higher the fitting accuracy.

[0110]

[0111] The parameters in the formula are the same as above. MAE represents the average absolute error between the predicted and observed values. The smaller the MAE, the smaller the absolute error and the more accurate the DEM.

[0112]

[0113] The parameters in the formula are the same as above, E max E represents the maximum elevation error between the generated DEM product and the actual elevation. max The smaller the value, the closer the maximum elevation of the generated DEM will be to the actual elevation.

[0114] Step 8: Mapping of a large-scale, fine-scale DEM. The acquired fine-scale DEM is geocoded and other steps are used to generate a DEM product that matches the actual geographical location. The above process is performed in the SAR image coordinate system, meaning that the corresponding elevation values ​​obtained by phase-to-elevation conversion are still in the SAR image coordinate system. Usually, to facilitate the interpretation and analysis of the acquired geographic information, the acquired elevation map needs to be transformed from the SAR image coordinate system to the commonly used geodetic coordinate system. Using the geocoding lookup list generated by GAMMA software, the DEM is transformed from the range-Doppler coordinate system to the geographic coordinate system, and then map projection transformation is performed according to actual needs to generate a large-scale, fine-scale DEM product.

[0115] In specific implementation, the method proposed in the technical solution of this invention can be automatically executed by those skilled in the art using computer software technology. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of this invention and computer equipment including the computer program running the corresponding computer program, should also be within the protection scope of this invention.

[0116] In some possible embodiments, a multi-baseline temporal InSAR fine-scale DEM generation system is provided, including the following modules:

[0117] The first module is used for the generation of differential interferometric phase, including generating coherence coefficient maps and differential interferograms based on InSAR using preprocessing of previous data, and generating a time-series differential interferometric phase matrix based on high coherence points;

[0118] The second module is used for ICA-based removal of atmospheric and noise phases, including using FastICA to separate the generated differential interferometric phases and directly remove atmospheric and noise.

[0119] The third module is used for polynomial fitting to remove baseline errors. This includes fitting the baseline error of the differential interferometric phase after removing atmospheric and noise using a polynomial fitting method, and subtracting the fitted baseline error from the differential interferometric phase to obtain an accurate differential interferometric phase.

[0120] The fourth module is used to construct a temporal InSAR physical model between differential interferometric phase and terrain and deformation. This includes comprehensively considering the effects of linear deformation, periodic deformation and environmental factors during the multi-baseline InSAR modeling process, and constructing a temporal InSAR physical model that takes environmental parameters into account.

[0121] The fifth module is used for optimal estimation of terrain and deformation parameters, including solving the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters.

[0122] The sixth module is used for generating fine-scale DEMs, including generating fine-scale DEMs by utilizing the relationship between the solved terrain parameters and the actual surface elevation.

[0123] The seventh module is used for the accuracy evaluation of fine-scale DEMs, including calculating the error between the generated DEM and the actual surface elevation data, and evaluating the accuracy of the generated fine-scale DEMs.

[0124] The eighth module is used for mapping large-scale, fine-scale DEMs, including generating DEM products that match the actual geographical location by taking steps such as geocoding from the acquired fine-scale DEMs.

[0125] In some possible embodiments, a multi-baseline time-series InSAR fine-scale DEM generation system is provided, including a processor and a memory. The memory is used to store program instructions, and the processor is used to call the stored instructions in the memory to execute a multi-baseline time-series InSAR fine-scale DEM generation method as described above.

[0126] In some possible embodiments, a multi-baseline temporal InSAR fine-scale DEM generation system is provided, including a readable storage medium on which a computer program is stored. When the computer program is executed, it implements a multi-baseline temporal InSAR fine-scale DEM generation method as described above.

[0127] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A method for generating a fine-scale DEM using multi-baseline temporal InSAR, characterized in that: The signal is separated using spatial ICA technology, and atmospheric and noise interference is directly removed, thus reducing the interference from atmospheric and noise. Polynomial fitting is used to remove baseline errors in phase observations and obtain accurate differential interferometric phase. By integrating the effects of linear deformation, periodic deformation, and environmental factors, a temporal InSAR physical model is constructed to connect differential interferometric phase with terrain and deformation, thereby obtaining accurate terrain and deformation parameters and enabling the generation of fine-scale DEMs. The implementation process includes the following steps: Step 1, generation of differential interferometric phase, including generating coherence coefficient map and differential interferogram based on InSAR using previous data preprocessing, and generating time-series differential interferometric phase matrix based on high coherence points; Step 2, ICA-based removal of atmospheric and noise phases, including using FastICA to separate the generated differential interference phases and directly remove atmospheric and noise; Step 3, polynomial fitting to remove baseline error, includes fitting the baseline error of the differential interferometric phase after removing atmosphere and noise using a polynomial fitting method, and subtracting the fitted baseline error from the differential interferometric phase to obtain the accurate differential interferometric phase. Step 4: Construct a temporal InSAR physical model between differential interferometric phase and terrain and deformation, including comprehensively considering the effects of linear deformation, periodic deformation and environmental factors in the process of multi-baseline InSAR modeling, and constructing a temporal InSAR physical model that takes into account environmental parameters. Step 5, optimal estimation of terrain and deformation parameters, including solving the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters; Step 6, generation of fine-scale DEM, including generating fine-scale DEM using the relationship between the solved terrain parameters and the actual surface elevation; Step 7, accuracy evaluation of fine-scale DEM, including calculating the error between the DEM generated in steps 1-6 and the actual surface elevation data, and evaluating the accuracy of the generated fine-scale DEM; Step 8, mapping of large-scale fine-scale DEM, including generating DEM products that match the actual geographical location by geocoding the acquired fine-scale DEM.

2. The multi-baseline temporal InSAR fine-scale DEM generation method according to claim 1, characterized in that: The generation of differential interferometric phase involves data format conversion, baseline estimation and connection diagram generation, differential interferometric phase generation, selection of high coherence points, and the resulting unwrapped phase representation: In the formula, Indicates the first On the first differential interferogram 1 pixel The untangling phase at the location, ; Indicates the deformation phase; Indicates the radar wavelength; Indicates the first The deformation component of the differential interferogram pointing upwards; Indicates terrain phase ; and Representing the radar incident angle and the first... The length of the vertical baseline of each interference pair; This indicates the distance between the coherent target and the location of the radar satellite; It is an elevation error; Indicates orbital error; Residual phase, mainly including atmospheric delay phase noise phase and high-pass deformation components .

3. The multi-baseline temporal InSAR fine-scale DEM generation method according to claim 2, characterized in that: The method for removing atmospheric and noise phase based on ICA is to use FastICA to separate the time series phase signal, directly remove the atmospheric and noise-related components, and extract the important components. The InSAR time series phase signal is represented as: In the formula, This represents a mixed time-series signal, considered as the InSAR time-series phase matrix. , Indicates the differential interference phase; The spatial response is the combination of the spatial responses of all independent components. , Represents the spatial response of each independent component. The number of independent components; This represents the time series of all independent components, including deformation phase, terrain phase, atmospheric phase, noise, and the remaining residual phase signal. Represents the time series of each independent component; and These represent the phase signals related to deformation and terrain, respectively. Phase after ICA separation Represented as: In the formula, and These represent the spatial responses corresponding to the deformation component and the topography component, respectively; and These represent the time series corresponding to the deformation component and the topographic component, respectively.

4. The multi-baseline temporal InSAR fine-scale DEM generation method according to claim 3, characterized in that: The method for removing baseline error using polynomial fitting is as follows: peak elimination; approximating the baseline using polynomial fitting to obtain the baseline error; and subtracting the baseline error from the phase signal obtained by ICA separation to obtain the corrected phase signal. The baseline-corrected phase signal is represented as: In the formula, Indicates the phase after baseline correction. This represents the baseline error of the polynomial fit.

5. The multi-baseline temporal InSAR fine-scale DEM generation method according to claim 4, characterized in that: The implementation of the temporal InSAR physical model between differential interferometric phase and terrain and deformation is as follows: constructing a surface deformation model; constructing a temporal InSAR physical model between differential interferometric phase and linear, periodic, environmental and terrain parameters; The temporal InSAR physical model is represented as follows: In the formula, and The first The time of acquisition of the main image and the time of acquisition from the differential interferogram. The linear deformation rate is the rate of deformation between any two images; the linear deformation rate varies at different times. and These represent the time corresponding to the linear deformation rate at different times and the time of the previous time; and These represent the periodic deformation coefficients, which are considered unknown parameters; This indicates a year; and These represent the temperature deformation coefficient and the precipitation deformation coefficient, respectively, and are considered as unknown parameters. and These represent the monthly average temperature and the monthly average rainfall, respectively.

6. The multi-baseline temporal InSAR fine-scale DEM generation method according to claim 5, characterized in that: The generation of a fine-scale DEM is achieved by utilizing the relationship between the solved terrain parameters and the actual surface elevation. This allows for the creation of a fine-scale DEM at any point on the ground. The elevation at a location is expressed as: In the formula: Represents any point The actual elevation of the location To reference the elevation of the external DEM, The residual elevation error is obtained after ICA phase separation, multi-baseline error correction, and time-series InSAR environment modeling.

7. A multi-baseline temporal InSAR fine-scale DEM generation system, characterized in that: This method is used to implement a multi-baseline temporal InSAR fine-scale DEM generation method as described in any one of claims 1-6.

8. The multi-baseline temporal InSAR fine-scale DEM generation system according to claim 7, characterized in that: Includes the following modules, The first module is used for the generation of differential interferometric phase, including generating coherence coefficient maps and differential interferograms based on InSAR using preprocessing of previous data, and generating a time-series differential interferometric phase matrix based on high coherence points; The second module is used for ICA-based removal of atmospheric and noise phases, including using FastICA to separate the generated differential interferometric phases and directly remove atmospheric and noise. The third module is used for polynomial fitting to remove baseline errors. This includes fitting the baseline error of the differential interferometric phase after removing atmospheric and noise using a polynomial fitting method, and subtracting the fitted baseline error from the differential interferometric phase to obtain an accurate differential interferometric phase. The fourth module is used to construct a temporal InSAR physical model between differential interferometric phase and terrain and deformation. This includes comprehensively considering the effects of linear deformation, periodic deformation and environmental factors during the multi-baseline InSAR modeling process, and constructing a temporal InSAR physical model that takes environmental parameters into account. The fifth module is used for optimal estimation of terrain and deformation parameters, including solving the constructed time-series InSAR physical model that takes into account environmental parameters to obtain accurate terrain and deformation parameters. The sixth module is used for generating fine-scale DEMs, including generating fine-scale DEMs by utilizing the relationship between the solved terrain parameters and the actual surface elevation. The seventh module is used for the accuracy evaluation of fine-scale DEMs, including calculating the error between the generated DEM and the actual surface elevation data, and evaluating the accuracy of the generated fine-scale DEM. The eighth module is used for mapping large-scale, fine-scale DEMs, including generating DEM products that match the actual geographical locations by geocoding the acquired fine-scale DEMs.

9. The multi-baseline temporal InSAR fine-scale DEM generation system according to claim 7, characterized in that: It includes a processor and a memory, the memory being used to store program instructions, and the processor being used to call the stored instructions in the memory to execute a multi-baseline time-series InSAR fine-scale DEM generation method as described in any one of claims 1-6.

Images