InSAR atmospheric correction method and related device

Abstract

The invention provides an InSAR atmospheric correction method and a related device, and the method comprises the steps: determining an SAR image and orbit data in a preset space-time range, and carrying out the interference based on the SAR image and the orbit data, and obtaining a differential interferogram; environmental parameters in a preset space-time range are determined, the difference interferogram is corrected through a pre-trained atmospheric correction model and the environmental parameters, an atmospheric correction interference image is obtained, the pre-trained atmospheric correction model comprises at least two sub atmospheric correction models, and the atmospheric correction interference image is configured to perform inversion on earth surface deformation. According to the invention, the accuracy, generalization and interpretability of atmospheric correction can be improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of electric digital data processing, and in particular, to an InSAR atmospheric correction method and related device. BACKGROUND

[0002] This section is intended to provide background or context to the embodiments of the disclosure recited in the claims. The description herein does not constitute admission that the prior art is prior art nor does it constitute an admission of any description in this section as prior art to a currently claimed embodiment.

[0003] InSAR (Synthetic Aperture Radar Interferometric) is the full name of Synthetic Aperture Radar Interferometric technology, which mainly realizes the monitoring of ground deformation and changes by being carried on a satellite. InSAR performs remote sensing measurement mainly by obtaining the differential phase of SAR images to infer the deformation variable, but since the radar wave is irradiated to the ground through the atmosphere to form an image, the atmospheric delay becomes the main error source of InSAR, and the phase of the interferogram needs to be corrected.

[0004] However, in the related art, there are problems such as inability to correct end flow signals, low accuracy, and weak interpretability, which further leads to poor correction effect. SUMMARY

[0005] Therefore, the purpose of the present disclosure is to provide an InSAR atmospheric correction method and related device, which at least solves one of the technical problems in the related art to some extent.

[0006] To achieve the above purpose, in the first aspect, an InSAR atmospheric correction method is provided, and the method comprises: determining SAR images and orbit data of a preset space-time range, performing interference based on the SAR images and the orbit data to obtain a differential interferogram; determining environmental parameters of the preset space-time range, correcting the differential interferogram by using a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric correction interferogram, wherein the pre-trained atmospheric correction model comprises at least two sub-atmospheric correction models, and the atmospheric correction interferogram is configured to be used for inversion of ground surface deformation.

[0007] In some example embodiments, the interference based on the SAR images and the orbit data to obtain the differential interferogram comprises: registering the SAR images based on the orbit data to obtain a SAR single-view complex image stack; generating a connection graph of the SAR single-view complex image stack, and performing interference on the SAR images based on the connection graph to obtain the differential interferogram.

[0008] In some example embodiments, the environmental parameters include: meteorological data, vegetation data, and elevation data.

[0009] In some example embodiments, the method further comprises training the adaptive weight prediction model by: constructing a sample set comprising a plurality of samples; wherein the samples comprise: sample data and label data; the sample data comprises training differential interferograms and training environmental parameters; the label data comprises training atmospheric correction interferograms; inputting the sample data into a to-be-trained atmospheric correction model to obtain prediction data output by the model, wherein the prediction data comprises atmospheric correction prediction interferograms output by the model; determining the difference between the prediction data and the label data; based on the difference, updating the parameters of the to-be-trained atmospheric correction model through back propagation until the difference between the prediction data and the label data is minimized, obtaining the pre-trained atmospheric correction model.

[0010] In some example embodiments, the inputting the sample data into a to-be-trained atmospheric correction model to obtain prediction data output by the model comprises: the to-be-trained atmospheric correction model performs feature fusion on the training environmental parameters and the training differential interferograms to obtain training multi-modal fusion features; weight calculation is performed on the training multi-modal fusion features to obtain at least two correction weight coefficients; the training differential interferograms are corrected based on the at least two correction weight coefficients through at least two sub-atmospheric correction models to obtain the prediction data.

[0011] In some example embodiments, the to-be-trained atmospheric correction model performs feature fusion on the training environmental parameters and the training differential interferograms to obtain training multi-modal fusion features, comprising: the to-be-trained atmospheric correction model performs feature extraction on the meteorological data based on a meteorological data encoder to obtain meteorological data features; feature extraction is performed on the vegetation data based on a vegetation data encoder to obtain vegetation data features; feature extraction is performed on the elevation data based on a digital elevation data encoder to obtain elevation digital features; feature extraction is performed on the differential interferograms based on an interferogram data encoder to obtain interferogram features; the meteorological data features, the vegetation data features, the elevation digital features, and the interferogram features are spliced to obtain the training multi-modal fusion features.

[0012] In some example embodiments, the correcting the training differential interferogram based on the at least two correction weight coefficients through the at least two sub-atmosphere correction models to obtain the prediction data comprises: correcting the differential interferogram through the at least two sub-atmosphere correction models to obtain at least two corrected interferograms; weighting and fusing the at least two corrected interferograms based on the at least two correction weight coefficients to obtain the prediction data.

[0013] Based on the same inventive concept, the second aspect of the example embodiments of the present disclosure provides an InSAR atmosphere correction device, comprising: an interferogram determination module configured to determine SAR images and orbit data of a preset spatiotemporal range, perform interference based on the SAR images and the orbit data to obtain a differential interferogram; an interferometric image determination module configured to determine environmental parameters of the preset spatiotemporal range, correct the differential interferogram through a pre-trained atmosphere correction model and the environmental parameters to obtain an atmosphere-corrected interferometric image, wherein the pre-trained atmosphere correction model comprises at least two sub-atmosphere correction models, and the atmosphere-corrected interferometric image is configured to be used for inversion of ground surface deformation.

[0014] In some example embodiments, the interferogram determination module is specifically configured to: determine SAR images and orbit data of a preset spatiotemporal range, perform registration on the SAR images based on the orbit data to obtain a SAR single-view complex image stack; generate a connection graph of the SAR single-view complex image stack, and perform interference on the SAR images based on the connection graph to obtain a differential interferogram.

[0015] In some example embodiments, the interferometric image determination module is specifically configured to: determine environmental parameters of the preset spatiotemporal range, correct the differential interferogram through a pre-trained atmosphere correction model and the environmental parameters to obtain an atmosphere-corrected interferometric image; wherein the environmental parameters comprise meteorological data, vegetation data and elevation data; the pre-trained atmosphere correction model comprises at least two sub-atmosphere correction models, and the atmosphere-corrected interferometric image is configured to be used for inversion of ground surface deformation; constructing an atmosphere correction model by: constructing a sample set comprising a plurality of samples; wherein the samples comprise sample data and label data; the sample data comprises training differential interferograms and training environmental parameters; and the label data comprises training atmosphere-corrected interferometric images; The sample data is input into the atmospheric correction model to be trained. The atmospheric correction model to be trained extracts features from the meteorological data based on the meteorological data encoder to obtain meteorological data features. Based on the vegetation data encoder, feature extraction is performed on the vegetation data to obtain vegetation data features; Based on the digital elevation data encoder, feature extraction is performed on the elevation data to obtain digital elevation features; Based on the interferogram data encoder, feature extraction is performed on the differential interferogram to obtain the interferogram features; The meteorological data features, vegetation data features, elevation digital features, and interferogram features are stitched together to obtain training multimodal fusion features; The training multimodal fusion features are weighted to obtain at least two corrected weight coefficients; At least two corrected interferograms are obtained by correcting the differential interferogram using at least two of the aforementioned sub-atmosphere correction models. Based on at least two of the aforementioned correction weight coefficients, at least two of the corrected interferograms are weighted and fused to obtain prediction data, wherein the prediction data includes atmospheric correction prediction images output by the model; Determine the difference between the predicted data and the label data; Based on the difference, the parameters of the atmospheric correction model to be trained are updated through backpropagation until the difference between the predicted data and the label data is minimized, thus obtaining the pre-trained atmospheric correction model.

[0016] Based on the same inventive concept, a third aspect of the exemplary embodiments of this disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method as described in the first aspect.

[0017] Based on the same inventive concept, a fourth aspect of the exemplary embodiments of this disclosure provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method as described in the first aspect.

[0018] Based on the same inventive concept, a fifth aspect of the exemplary embodiments of this disclosure provides a computer program product including computer program instructions that, when run on a computer, cause the computer to perform the method as described in the first aspect.

[0019] As can be seen from the above description, the InSAR atmospheric correction method and related apparatus provided in this disclosure include: A SAR image and orbital data within a predetermined spatiotemporal range are determined. Interferometry is performed based on the SAR image and orbital data to obtain a differential interferogram. Environmental parameters within the predetermined spatiotemporal range are determined, and the differential interferogram is corrected using a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospherically corrected interferometric image. The pre-trained atmospheric correction model includes at least two sub-atmospheric correction models, and the atmospherically corrected interferometric image is configured to invert surface deformation. This disclosure improves the accuracy, generalization, and interpretability of atmospheric correction. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this disclosure or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A schematic diagram illustrating an application scenario of the InSAR atmospheric correction method provided as an exemplary embodiment of this disclosure; Figure 2 A schematic flowchart of an InSAR atmospheric correction method provided for an exemplary embodiment of this disclosure; Figure 3 A schematic diagram of an atmospheric correction model of an InSAR atmospheric correction device provided as an exemplary embodiment of this disclosure; Figure 4 A schematic diagram of an InSAR atmospheric correction device provided as an exemplary embodiment of the present disclosure; Figure 5 A schematic diagram of the structure of an electronic device hardware provided for an exemplary embodiment of this disclosure. Detailed Implementation

[0022] It is understood that before using the technical solutions disclosed in the various embodiments of this application, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this application in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0023] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this application's technical solution, based on the prompt message.

[0024] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0025] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this application. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this application.

[0026] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0027] To make the objectives, technical solutions, and advantages of this disclosure clearer, the principles and spirit of this disclosure will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement this disclosure, and are not intended to limit the scope of this disclosure in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0028] In this article, it is important to understand that any number of elements in the accompanying figures is for illustrative purposes and not for limitation, and any naming is for distinction only and has no limiting meaning.

[0029] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this disclosure should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "first," "second," and similar words used in the embodiments of this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly. The article "a" or "an" preceding an element does not exclude the existence of multiple such elements.

[0030] The principles and spirit of this disclosure will be explained in detail below with reference to several representative embodiments.

[0031] As described in the background section, related technologies suffer from problems such as the inability to correct end-flow signals, low accuracy, and weak interpretability, leading to poor correction results. Specifically, InSAR (Synthetic Aperture Radar Interferometric) is a technique that primarily monitors ground deformation and changes by mounting it on satellites. Due to its advantages of all-weather, all-time, continuous spatial coverage, high accuracy, low cost, and high automation, this technology has greatly expanded the scope and accuracy of Earth science research, and therefore has been widely used in Earth observation and mapping.

[0032] InSAR was first proposed for topographic surveying in 1974, primarily using Differential InSAR (D-InSAR) to measure surface deformation. However, D-InSAR typically calculates surface deformation by the phase difference between two SAR images, making it susceptible to atmospheric phase delay, topographic phase, and spatiotemporal incoherence. Therefore, time series InSAR (TS-InSAR) was subsequently developed. Currently, the most prevalent TS-InSAR techniques include Persistent Scatterers InSAR (PS-InSAR) and Small Baseline Subset InSAR (SBAS-InSAR). Both techniques utilize multiple images (typically more than 20) for interferometry, improving accuracy from centimeter-level to millimeter-level compared to D-InSAR, and have become the mainstream techniques. While PS-InSAR uses permanent scatterers as a reference for deformation measurement, it is mainly suitable for urban and infrastructure monitoring scenarios. SBAS-InSAR, on the other hand, is suitable for large-scale natural surface deformation or vegetation-covered areas, making it applicable to a wider range of scenarios. This invention primarily focuses on SBAS-InSAR technology.

[0033] SBAS-InSAR primarily constructs a redundant observation network by combining subsets of interferometric pairs from multiple short-spatial baselines. It then obtains the optimal solution through singular matrices and inverts it into a time-series deformation. The main steps include: selecting interferometric pairs based on spatial baselines (typically <300 meters) and temporal baselines (typically less than 180 days) and creating a connectivity map; generating an interferogram using Digital Elevation Model (DEM) elevation information; performing phase unwrapping; obtaining phase information after various corrections; and finally calculating the deformation information.

[0034] InSAR remote sensing measurements primarily rely on differential phase inference of deformation from SAR images. However, since radar waves must pass through the atmosphere to reach the ground for imaging, atmospheric delay becomes a major source of error in InSAR, necessitating phase correction of the interferogram. Early correction methods depended on physical models and statistical experience, including using GNSS (Global Navigation Satellite System) to obtain zenith atmospheric delay or using meteorological models to estimate atmospheric delay, and establishing linear or exponential regression models based on phase-topographic correlation. Of these three methods, the first two are limited by spatiotemporal resolution and struggle to capture local turbulence, while regression methods ignore atmospheric spatiotemporal heterogeneity, resulting in significant errors in complex terrain areas. The subsequently proposed GACOS (Generic Atmospheric Correction Online Service) system, combining GNSS water vapor tomography and spectrophotometer (MODIS) inversion data, improves the accuracy of wet delay correction. After 2020, a research team proposed a deep learning-based method to perform atmospheric correction by building ARU-Net (Attention-Based Deep Residual U-net).

[0035] In existing technologies, the following methods are mainly used for atmospheric correction: (1) Pairwise comparison This method requires at least two interferometric images of shared SAR imagery. If atmospheric anomalies exist in the shared SAR imagery, they will contaminate the deformation signals in all relevant interferometric images. Therefore, subtracting two interferometric images can significantly reduce the influence of the atmosphere. Its disadvantages are that interferometric images of shared SAR imagery may not be available, and this method is relatively coarse and lacks high accuracy.

[0036] (2) Layering method This method considers atmospheric delay as a form of random error and aims to mitigate its impact by averaging N time-independent interferometric images. However, this approach assumes a constant deformation rate and is therefore unsuitable for regions with nonlinear deformation rates.

[0037] (3) Atmospheric delay conversion model method based on ground-based meteorological data This method establishes a model of the zenith-direction moist atmospheric delay based on surface meteorological observation data. The most common models are based on temperature, total atmospheric pressure, and water vapor pressure. Some methods also incorporate seasonal, latitude, climatic, or statistical methods. However, this method has limited modeling accuracy and cannot mitigate the effects of topographically independent water vapor disturbances.

[0038] (4) Numerical atmospheric model method This method uses high-precision surface and atmospheric reanalysis data for atmospheric delay correction, which can effectively improve the accuracy of the correction model. However, since the spatiotemporal resolution of reanalysis data is often low, its ability to correct for rapidly changing turbulence is limited.

[0039] (5) Random filtering method This method identifies a stable region adjacent to the target area using prior information, extracts atmospheric signals from the stable region using an autocovariance function, estimates the atmospheric impact of the deformed region based on correlation characteristics, and finally subtracts the estimated atmospheric impact from the target phase. However, this method is only applicable to small-scale deformations and depends on the presence or absence of an adjacent stable region, thus limiting its versatility.

[0040] (6) Time series analysis techniques Time series analysis techniques mainly include PS-InSAR and SBAS-InSAR. By selecting highly coherent points and performing joint solutions, the influence of atmospheric delay can be weakened. Its accuracy depends on the number and distribution of highly coherent points.

[0041] (7) Linear regression method The paper "Evaluation of Time Series InSAR Tropospheric Delay Correction Methods on the Northwest Margin of the Tibetan Plateau" uses phase regression analysis to perform InSAR atmospheric delay correction.

[0042] This method assumes that the interferogram is mainly affected by static tropospheric delay, and its delay signal can be considered as the interpolation of the delay signal corresponding to the master-slave influence, thus obtaining a simple linear relationship:

[0043] in This represents the overall phase shift of a single interferogram. For elevation With regard to process delays Transfer function. Since there is a certain linear relationship between phase delay and elevation, a tropospheric delay phase map can be obtained by fitting the linear relationship between the delayed phase and ground elevation.

[0044] Since the atmospheric delay phase itself is unknown, the main assumption is that the atmospheric delay phase is 0 for phase unwrapping. Deformation points smaller than a specified threshold are extracted as stable points, and linear fitting is performed using the phase and elevation data of the stable points.

[0045] (8) Combination method The paper, "Research on GACOS Temporal InSAR Atmospheric Secondary Correction Method for Inland Canyon Reservoirs," employs linear regression, exponential regression, GACOS, and regression + GACOS models for atmospheric correction.

[0046] The GACOS model primarily uses ERA5 (5th generation ECMWF atmospheric reanalysis of the global climate dataset) and GNSS data (GNSS data itself can provide atmospheric delay values) to generate a Zenith Total Delay (ZTD) map, and then combines it with an exponential correction regression algorithm for comprehensive atmospheric correction.

[0047] The results show that, among the various methods, the regression model + GACOS approach is superior in most cases, but in a few instances, it can lead to overcorrection and decreased accuracy. The influencing factors are related to vegetation cover and elevation variations.

[0048] (9) Deep learning methods The paper "ARU-Net Reduction of Atmospheric Phase Screen in SAR Interferometry Using Attention-Based Deep Residual U-Net" and the patent "An InSAR Atmospheric Correction Method Based on Sparse Convolutional Neural Networks" propose a method based on deep learning networks, which has been proven to effectively improve correction accuracy in coastline monitoring. This method establishes a U-Net-based deep learning network. It selects atmospheric samples (assuming no deformation signals in the original interferometric images within a short time) and deformation samples (data with an average coherence value greater than 0.3 and no atmospheric delay signal interference within a 1km range) from existing InSAR multi-scene images. The atmospheric samples are then randomly added to the deformation samples to form the model input. The model is labeled with the deformation samples. Through model training, the model can decompose the deformation phase from interferograms containing deformation and atmospheric delay phase, thereby performing atmospheric correction.

[0049] Currently, the mainstream atmospheric correction methods include linear / exponential regression model methods, methods using GNSS, methods using meteorological reanalysis numerical methods, and combined methods.

[0050] Linear / exponential regression models can effectively correct atmospheric vertical stratification delays in areas with complex topographic relief, but they cannot correct turbulence signals. GNSS-based algorithms suffer from insufficient spatial resolution due to the sparsity of GNSS stations, reducing accuracy. Meteorological reanalysis-based algorithms also suffer from low spatiotemporal resolution, with the best reanalysis data having a resolution of only 31 km and 1 hour, resulting in poor correction performance. Existing combined methods simply involve proportional superposition of several methods, indicating varying superposition effects in different scenarios, but without providing quantitative considerations. The ARU-Net deep learning method relies solely on its own imagery data, lacking support from digital elevation and meteorological data in terms of physical mechanisms, thus requiring further improvement and exhibiting weak interpretability.

[0051] To address the aforementioned problems, this disclosure provides an InSAR atmospheric correction method and related apparatus, the method comprising: This invention addresses the issue of low accuracy in meteorological reanalysis data by introducing an atmospheric correction model, effectively improving spatial resolution. It employs four images to assist the meteorological data encoder in inferring atmospheric conditions at the time of SAR image capture, thereby obtaining better results. Simultaneously, by introducing multi-source data and utilizing the data encoder to extract features, model weights are generated through feature concatenation, overcoming the limitations of simple combination algorithms in traditional atmospheric correction methods. This allows the model to adapt to different environments and autonomously select the optimal algorithm, significantly improving overall accuracy and generalization performance. Furthermore, the model weight calculation module applies SHAP technology to directly output the weight coefficients of each atmospheric correction model, enhancing the interpretability of the method.

[0052] After introducing the basic principles of this disclosure, various non-limiting embodiments of this disclosure will be described in detail below.

[0053] refer to Figure 1 This is a schematic diagram illustrating an application scenario of the InSAR atmospheric correction method provided in an exemplary embodiment of this disclosure.

[0054] This application scenario includes a terminal device 101 and a server 102. The terminal device 101 and the server 102 can be connected via a wired or wireless communication network to achieve data interaction.

[0055] Terminal device 101 may be an electronic device located close to the user side, possessing data transmission and multimedia input / output functions, including but not limited to desktop computers, mobile phones, portable computers, tablet computers, media players, smart wearable devices, personal digital assistants (PDAs), or other electronic devices capable of performing the aforementioned functions. This electronic device may include a processor and a display screen with touch input functionality. The display screen is used to present a graphical user interface (GUI), which can display an application interface. The processor is used to process application data, generate the GUI, and control the display of the GUI on the screen.

[0056] Server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0057] In some exemplary embodiments, the InSAR atmospheric correction method can be run on terminal device 101 or server 102.

[0058] When the InSAR atmospheric correction method is running on server 102, server 102 is used to provide InSAR atmospheric correction services to users of terminal device 101.

[0059] Server 102 determines SAR images and orbital data within a preset spatiotemporal range. Server 102 performs interferometry based on the SAR images and orbital data to obtain a differential interferogram. Server 102 determines the environmental parameters of the preset spatiotemporal range. Server 102 corrects the differential interferogram using a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric correction interferometric image. Server 102 then transmits the atmospheric correction interferometric image to terminal device 101. The pre-trained atmospheric correction model includes at least two sub-atmospheric correction models, and the atmospheric correction interferometric image is configured to invert surface deformation.

[0060] It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of this disclosure, and the implementation of this disclosure is not limited in any way. On the contrary, the implementation of this disclosure can be applied to any applicable scenario.

[0061] refer to Figure 2 InSAR atmospheric correction method, the method includes the following steps: Step S210: Determine the SAR image and orbit data within a preset spatiotemporal range, and perform interferometry based on the SAR image and orbit data to obtain a differential interferogram.

[0062] In practice, the method for determining SAR imagery and orbital data within a preset spatiotemporal range is as follows: By accessing open data platforms such as the European Space Agency's Copernicus Data Space or Alaska Satellite Facilities, queries are performed according to the preset spatiotemporal range (i.e., the geographical boundaries and time windows of the target monitoring area), and the corresponding SAR image sequences (such as Sentinel-1 SLC data) and their precise orbital ephemeris files are automatically downloaded to ensure that all data spatially cover the target area, are temporally continuous, and meet the baseline requirements for interferometric processing.

[0063] In some embodiments, interferometry is performed based on the SAR image and the orbital data to obtain a differential interferogram, including: Based on the orbital data, the SAR images are registered to obtain a SAR single-look complex image stack; A connection diagram of the SAR single-view complex image stack is generated, and the SAR images are interferometrically processed based on the connection diagram to obtain the differential interferogram.

[0064] In practical implementation, SAR single-look complex image stack refers to: A SAR single-view complex image stack refers to a three-dimensional data set formed by arranging multiple SAR single-view complex images acquired at different times in the same monitoring area in chronological order; each SLC image contains not only the intensity information of backscattering from ground objects.

[0065] In specific implementation, the SAR images are registered based on the orbital data to obtain a SAR single-look complex image stack; a connection diagram of the SAR single-look complex image stack is generated; and the SAR images are interferometrically processed based on the connection diagram to obtain the differential interferogram. Register SAR images to obtain SAR single-look complex (SLC) images, generate a connectivity map, perform interferometry on the image pairs based on the connectivity map to generate a differential interferogram, and generate a time-series InSAR interferometric phase through filtering and phase unwrapping.

[0066] Step S220: Determine the environmental parameters of the preset spatiotemporal range, and correct the differential interferogram using a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric correction interferometric image. The pre-trained atmospheric correction model includes at least two sub-atmospheric correction models, and the atmospheric correction interferometric image is configured to invert surface deformation.

[0067] In practice, meteorological data refers to atmospheric state information (such as temperature, air pressure, and humidity), which is used to directly estimate the delay caused by the atmosphere to radar signals.

[0068] In practice, vegetation data refers to information on the status of surface vegetation cover, which is used to quantify the scattering and delay effects of the vegetation canopy on radar signals.

[0069] In practice, elevation data refers to the altitude information of the ground, which is used to correct for the atmospheric vertical stratification delay effect caused by topographic relief.

[0070] In some embodiments, the adaptive weight prediction model is further trained by the following method: A sample set is constructed, comprising several samples; wherein the samples include: sample data and label data; the sample data includes training differential interferograms and training environmental parameters; the label data includes training atmospheric-corrected interferometric images; The sample data is input into the atmospheric correction model to be trained to obtain the predicted data output by the model, wherein the predicted data includes the atmospheric correction predicted image output by the model. Determine the difference between the predicted data and the label data; Based on the difference, the parameters of the atmospheric correction model to be trained are updated through backpropagation until the difference between the predicted data and the label data is minimized, thus obtaining the pre-trained atmospheric correction model.

[0071] In specific implementation, a sample set is constructed, comprising several samples; wherein, the samples include: sample data and label data; the sample data includes training differential interferograms and training environmental parameters; the label data includes the method of training atmospheric corrected interferometric images. First, two key samples are extracted from the original interferograms: atmospheric samples, derived from interferograms with short time baselines and deformation signals below a threshold, whose phase primarily reflects atmospheric delay; and deformation samples, selected from interferograms with high coherence and no significant atmospheric interference, whose phase is considered a "clean" deformation signal. Subsequently, multi-source environmental data are acquired for each interferometric pair: including digital elevation information, multi-temporal vegetation data generated based on multispectral imagery and vegetation interpretation models (such as LuoJiaNET), and high-resolution, multi-temporal meteorological reanalysis data with improved accuracy through deep learning super-resolution models (such as FuXi-Regional). Finally, atmospheric and deformation samples are randomly combined and used together with the aforementioned environmental data as sample data, while the deformation samples themselves (i.e., atmospherically corrected interferograms) serve as label data, collectively forming a single sample. The sample set is constructed by performing sliding window slicing, data augmentation, and normalization on all samples, and dividing them into training and validation sets.

[0072] In practice, the sample data is input into the atmospheric correction model to be trained to obtain the predicted data output by the model, wherein the predicted data includes the atmospheric correction predicted image output by the model; the difference between the predicted data and the label data is determined as follows: During the model training phase, sample data from the sample set (i.e., noisy interferograms and environmental parameters) are input into the atmospheric correction model to be trained. This model extracts features from various data types through its internal multi-encoder network structure and adaptively fuses the outputs of multiple sub-correction models (such as linear regression, GACOS, and ARU-Net) via a weight calculation module, ultimately generating predicted data—i.e., atmospheric corrected predicted imagery. The difference (e.g., mean squared error) between this predicted data and the labeled data (i.e., "clean" deformed sample interferograms) is calculated, and this difference is used to update the model parameters via backpropagation.

[0073] In practice, based on the difference, the parameters of the atmospheric correction model to be trained are updated through backpropagation until the difference between the predicted data and the label data is minimized, thus obtaining the pre-trained atmospheric correction model. By continuously iterating the above process, referencing Figure 3 The parameters of the encoding data concatenation module and weight calculation module in the model are continuously optimized until the difference between the predicted data and the label data is minimized. Specifically, the validation set input data from the data sample set is input into the deep learning neural network model after the weights are updated. The validation set loss function value is calculated again and recorded in the computer memory. If the validation set loss function value is lower than the previous calculation, the weights of the deep learning neural network model at this time are saved and recorded as the optimal weight value. The above steps of obtaining weights are repeated. When the number of repetitions exceeds the specified maximum number of repetitions, or when the validation set loss function value increases instead of decreasing in multiple repetition rounds, training is stopped. The optimal weight value is used to adjust the network structure of the deep learning neural network model, and the above process is repeated until the validation set loss function value reaches the expected value and the optimal deep learning neural network weights are obtained. At this point, the model has learned how to comprehensively utilize multi-source environmental information to accurately estimate and subtract the atmospheric delay component in the interferometric phase, thereby obtaining a pre-trained atmospheric correction model that can be used for subsequent actual deformation monitoring tasks.

[0074] In some embodiments, the sample data is input into the atmospheric correction model to be trained to obtain the predicted data output by the model, including: The atmospheric correction model to be trained performs feature fusion on the training environmental parameters and the training differential interferogram to obtain training multimodal fusion features; The training multimodal fusion features are weighted to obtain at least two corrected weight coefficients; The predicted data is obtained by correcting the training differential interferogram using at least two of the said sub-atmospheric correction models based on at least two of the said correction weight coefficients.

[0075] In specific implementation, the atmospheric correction model to be trained fuses the training environmental parameters and the training differential interferogram to obtain training multimodal fusion features; weights are calculated on the training multimodal fusion features to obtain at least two correction weight coefficients; the training differential interferogram is corrected by at least two sub-atmospheric correction models based on at least two correction weight coefficients to obtain the predicted data. refer to Figure 3 The atmospheric correction model to be trained first extracts features from the training environmental parameters (including meteorological, vegetation, and elevation data) and the training differential interferogram, respectively. Then, the extracted multimodal features are stitched together by the coded data stitching module to form a multimodal fusion feature for training. Subsequently, the weight calculation module calculates the weights of the fusion feature to generate at least two correction weight coefficients. Finally, the differential interferogram is corrected by at least two sub-atmospheric correction models, and the multiple correction results are weighted and fused according to the weight coefficients to obtain the final atmospheric correction prediction image.

[0076] In specific implementation, refer to Figure 3 Sub-atmospheric correction models typically select n atmospheric correction sub-models with different conditions and principles. Each sub-model inputs the interferogram and additional data to generate a corrected interferogram. For example, three atmospheric correction models can be selected: the first is a linear regression model, the second is a deep learning network-based model (GNN) such as ARU-net, and the third is a GACOS correction model. If the sub-atmospheric correction model uses a GNN model, then the external data of the GNN data and the differential interferogram used for training need to be input into the GNN model for correction.

[0077] In some embodiments, the atmospheric correction model to be trained performs feature fusion on the training environmental parameters and the training differential interferogram to obtain training multimodal fusion features, including: The atmospheric correction model to be trained extracts features from the meteorological data based on the meteorological data encoder to obtain meteorological data features; Based on the vegetation data encoder, feature extraction is performed on the vegetation data to obtain vegetation data features; Based on the digital elevation data encoder, feature extraction is performed on the elevation data to obtain digital elevation features; Based on the interferogram data encoder, feature extraction is performed on the differential interferogram to obtain the interferogram features; The meteorological data features, vegetation data features, elevation digital features, and interferogram features are stitched together to obtain the training multimodal fusion features.

[0078] In specific implementation, the atmospheric correction model to be trained extracts features from the meteorological data based on the meteorological data encoder, and obtains the meteorological data features in the following way: A meteorological data encoder is used to extract meteorological data features at four time points. Currently, various large-scale deep learning models have emerged in the field of global numerical weather prediction, and the downsampling components of these models can all serve as meteorological data encoders. For example, cube embedding can be used to extract 3D meteorological data features through 3D convolutional layers. After embedding, a normalization layer is applied to improve training stability. The data is then fed into a downsampling module, each of which contains a 2D convolutional layer, a normalization layer, and a sigmoid-weighted activation function layer.

[0079] In specific implementation, the vegetation data features are obtained by extracting features from the vegetation data based on the vegetation data encoder in the following way: A vegetation data encoder is used to extract features from vegetation interpretation data at four time points. The vegetation data encoder can use typical downsampling structures for data compression. For example, 2D CNN convolutional layers, pooling layers, and residual connections can be used to improve feature extraction capabilities during data compression.

[0080] In specific implementation, the elevation data is extracted based on the digital elevation data encoder to obtain the elevation digital features in the following way: A digital elevation data encoder is used to extract digital elevation features. The digital elevation data encoder can use typical downsampling structures for data compression. For example, 2D CNN convolutional layers, pooling layers, and residual connections can be used to improve feature extraction capabilities during data compression.

[0081] In specific implementation, the interferogram features are obtained by extracting features from the differential interferogram based on the interferogram data encoder. An interferogram data encoder is used to extract digital feature information from the interferogram before correction. Data compression can be performed using typical downsampling structures. For example, 2D CNN convolutional layers, pooling layers, and residual connections can be used to improve feature extraction capabilities during data compression.

[0082] In specific implementation, the meteorological data features, vegetation data features, elevation numerical features, and interferogram features are concatenated to obtain the training multimodal fusion features: The meteorological data, vegetation data, elevation data, and interferometric data are flattened and stitched together. Specifically, the meteorological, vegetation, elevation, and interferometric data are first flattened and stitched together. Figure Four The two-dimensional or three-dimensional feature maps output by each encoder are unfolded in the spatial dimension and transformed from a matrix structure into a one-dimensional vector form. Then, these one-dimensional feature vectors from different data sources are concatenated in sequence to form a longer and more comprehensive joint feature vector in the feature dimension, thereby completing the structured integration of multi-source heterogeneous features.

[0083] In some embodiments, the training differential interferogram is corrected using at least two of the said sub-atmospheric correction models based on at least two of the said correction weight coefficients to obtain the predicted data, including: At least two corrected interferograms are obtained by correcting the differential interferogram using at least two of the aforementioned sub-atmosphere correction models. The predicted data is obtained by weighted fusion of at least two corrected interferograms based on at least two of the corrected weight coefficients.

[0084] In specific implementation, at least two corrected interferograms are obtained by correcting the differential interferogram using at least two of the aforementioned sub-atmospheric correction models: refer to Figure 3 The input differential interferogram is independently corrected by at least two sub-atmospheric correction models (such as linear regression model, GACOS model or deep learning model ARU-Net, etc.). Each model processes the differential interferogram according to its specific algorithm principle and generates a corrected interferogram.

[0085] In specific implementation, the predicted data is obtained by weighted fusion of at least two corrected interferograms based on at least two of the aforementioned correction weight coefficients. refer to Figure 3 The system uses at least two correction weight coefficients calculated from multimodal fusion features to perform weighted fusion of these corrected interferograms, thereby obtaining the final atmospheric corrected prediction image.

[0086] In specific implementation, configuring the atmospheric corrected interferometric image for surface deformation inversion means that the atmospheric corrected interferometric image disclosed herein can be used for inversion to obtain deformation results and extract the overall deformation extraction effect. In SBAS-InSAR technology, the phases of the corrected interferograms generated from multiple small baseline pairs can be combined into a matrix, and the final deformation result can be obtained using the singular value decomposition (SVD) method. Specific deformation values ​​at a certain coordinate can be extracted from the deformation results and compared with higher-precision instrument measurements to obtain the actual error of the method. For example, in the field of dam deformation monitoring, the deformation monitoring values ​​of a level instrument can be obtained as a high-precision benchmark to measure the results of this method.

[0087] In some exemplary embodiments, the deformation results obtained based on the above exemplary embodiments were tested in dam deformation monitoring in Sichuan, Beijing, and Hebei. SAR data used Sentinel-1A satellite C-band imagery with a spatial resolution of 5m*20m and a revisit period of 12 days. For each dam, ascending VV (Vertical Vertical) polarimetric imagery from January 2022 to January 2024 was selected, totaling 186 scenes. Meteorological data used ECMWF ERA5 reanalysis data. Vegetation data used Gaofen-1 satellite multispectral data. Digital elevation data used ASTER GDEM v3 (Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model version 3) data from the Geospatial Cloud. The model was developed using Python based on the PyTorch library and trained on a single computing server (56-core CPU, 2 Nvidia A800 80G GPUs) for a total of 100 training epochs. In three test scenarios, based on SBAS-InSAR technology, the atmospheric delay correction of interferograms was performed using the method of this invention (which includes atmospheric correction models of linear regression, GACOS, and ARU-Net, respectively), the linear regression model method, the GACOS method, and the deep learning algorithm (ARU-Net), and finally the dam deformation results were generated. The root mean square error was calculated using the measurement results of the level instrument equipped on the dam as the benchmark. In the three test scenarios, the relative error of this method was the lowest, and the error was reduced by an average of 9.27% ​​compared with the suboptimal model.

[0088] Furthermore, the three test scenarios differ in terms of environment. Sichuan experiences significantly more precipitation than Hebei and Beijing, and the weight of the GACOS model in Sichuan is noticeably higher than in Hebei and Beijing in the proposed model. Additionally, analysis of the weight distribution in Beijing during different seasons reveals that in drier seasons like spring and winter, the atmospheric delay phase effect caused by meteorological factors is lower, resulting in a correspondingly lower weight for the GACOS model. Conversely, in summer, with more rainfall, the atmospheric moisture delay effect in the phase is greater, leading to a significantly higher weight for the GACOS model. Moreover, since dams are typically built in areas with significant topographic relief, the linear regression weight models all maintain a weight distribution of at least 0.2, consistent with physical laws. ARU-Net, as a recently proposed deep learning model, can learn underlying patterns through training, maintaining a weight distribution above 0.24 in all scenarios, further validating its effectiveness. From the above information, it can be seen that the method proposed in this invention significantly improves the generalization performance and interpretability of InSAR atmospheric correction.

[0089] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0090] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] Based on the same inventive concept, corresponding to any of the above embodiments, this disclosure also provides an InSAR atmospheric correction device.

[0092] refer to Figure 4 The InSAR atmospheric correction device includes: The interferogram determination module 410 is configured to determine SAR images and orbital data within a preset spatiotemporal range, and perform interferometry based on the SAR images and orbital data to obtain a differential interferogram; The interferometric image determination module 420 is configured to determine the environmental parameters of the preset spatiotemporal range, and to correct the differential interferogram using a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric correction interferometric image. The pre-trained atmospheric correction model includes at least two sub-atmospheric correction models, and the atmospheric correction interferometric image is configured to invert surface deformation.

[0093] In this exemplary embodiment, the interferogram determination module 410 is specifically configured as follows: SAR images and orbital data within a preset spatiotemporal range are determined, and the SAR images are registered based on the orbital data to obtain a SAR single-view complex image stack. A connection diagram of the SAR single-view complex image stack is generated, and the SAR images are interferometrically processed based on the connection diagram to obtain a differential interferogram.

[0094] In this exemplary embodiment, the interferometric image determination module 420 is specifically configured as follows: The environmental parameters of the preset spatiotemporal range are determined, and the differential interferogram is corrected by a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric corrected interferometric image; wherein, the environmental parameters include: meteorological data, vegetation data and elevation data; the pre-trained atmospheric correction model includes at least two sub-atmospheric correction models, and the atmospheric corrected interferometric image is configured to invert surface deformation; The atmospheric correction model is constructed using the following method: A sample set is constructed, comprising several samples; wherein the samples include: sample data and label data; the sample data includes training differential interferograms and training environmental parameters; the label data includes training atmospheric-corrected interferometric images; The sample data is input into the atmospheric correction model to be trained. The atmospheric correction model to be trained extracts features from the meteorological data based on the meteorological data encoder to obtain meteorological data features. Based on the vegetation data encoder, feature extraction is performed on the vegetation data to obtain vegetation data features; Based on the digital elevation data encoder, feature extraction is performed on the elevation data to obtain digital elevation features; Based on the interferogram data encoder, feature extraction is performed on the differential interferogram to obtain the interferogram features; The meteorological data features, vegetation data features, elevation digital features, and interferogram features are stitched together to obtain training multimodal fusion features; The training multimodal fusion features are weighted to obtain at least two corrected weight coefficients; At least two corrected interferograms are obtained by correcting the differential interferogram using at least two of the aforementioned sub-atmosphere correction models. Based on at least two of the aforementioned correction weight coefficients, at least two of the corrected interferograms are weighted and fused to obtain prediction data, wherein the prediction data includes atmospheric correction prediction images output by the model; Determine the difference between the predicted data and the label data; Based on the difference, the parameters of the atmospheric correction model to be trained are updated through backpropagation until the difference between the predicted data and the label data is minimized, thus obtaining the pre-trained atmospheric correction model.

[0095] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing this disclosure, the functions of each module can be implemented in one or more software and / or hardware.

[0096] The apparatus of the above embodiments is used to implement the corresponding InSAR atmospheric correction method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0097] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the InSAR atmospheric correction method described in any of the above embodiments.

[0098] Figure 5 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0099] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0100] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0101] The input / output interface 1030 is used to connect input / output modules to realize information input and output. The input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0102] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0103] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0104] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0105] The electronic devices described above are used to implement the corresponding InSAR atmospheric correction methods in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0106] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the InSAR atmospheric correction method as described in any of the above embodiments.

[0107] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0108] The aforementioned non-transitory computer-readable storage media can be any available medium or data storage device that a computer can access, including but not limited to magnetic storage (e.g., floppy disks, hard disks, magnetic tapes, magneto-optical disks (MOs), etc.), optical storage (e.g., CDs, DVDs, BDs, HVDs, etc.), and semiconductor storage (e.g., ROMs, EPROMs, EEPROMs, non-volatile memory (NAND flash), solid-state drives (SSDs)).

[0109] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the InSAR atmospheric correction method as described in any of the embodiments in the exemplary method section above, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0110] Based on the same inventive concept, corresponding to the InSAR atmospheric correction method described in any of the above embodiments, this disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the InSAR atmospheric correction method. Corresponding to the execution entity for each step in each embodiment of the InSAR atmospheric correction method, the processor executing the corresponding step can belong to the corresponding execution entity.

[0111] The computer program product of the above embodiments is used to cause the computer and / or the processor to execute the InSAR atmospheric correction method as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0112] Those skilled in the art will recognize that embodiments of this disclosure can be implemented as a system, method, or computer program product. Therefore, this disclosure can be implemented as entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this disclosure can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0113] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (not exhaustive) of a computer-readable storage medium may include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0114] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0115] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0116] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0117] It should be understood that each block of a flowchart and / or block diagram, as well as combinations of blocks in a flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine that, when executed by a computer or other programmable data processing device, creates means for implementing the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0118] These computer program instructions may also be stored in a computer-readable medium that enables a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable medium produce a product comprising an instruction apparatus that implements the functions / operations specified in the boxes of a flowchart and / or block diagram.

[0119] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, such that the instructions that execute on the computer or other programmable apparatus can provide a process for implementing the functions / operations specified in the boxes of a flowchart and / or block diagram.

[0120] Furthermore, although the operations of the methods of this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. Rather, the steps depicted in the flowcharts may be executed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0121] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0122] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0123] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application (including the claims) is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in the details for the sake of brevity.

[0124] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0125] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0126] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

[0127] While the spirit and principles of this disclosure have been described with reference to several specific embodiments, it should be understood that this disclosure is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for convenience of expression. This disclosure is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. The scope of the appended claims is to be interpreted in the broadest sense, thereby encompassing all such modifications and equivalent structures and functions.

Claims

1. An InSAR atmospheric correction method, characterized in that, The method comprises the following steps: determining SAR images and track data in a preset space-time range, and performing interference based on the SAR images and the track data to obtain a differential interferogram; determining environmental parameters in the preset space-time range, and correcting the differential interferogram based on a pre-trained atmospheric correction model and the environmental parameters to obtain an atmospheric correction interferogram, wherein the pre-trained atmospheric correction model comprises at least two sub atmospheric correction models, and the atmospheric correction interferogram is configured to be used for inversion of ground surface deformation.

2. The method of claim 1, wherein, The interference based on the SAR images and the track data to obtain the differential interferogram comprises: registering the SAR images based on the track data to obtain a SAR single-view complex image stack; generating a connection graph of the SAR single-view complex image stack, and performing interference on the SAR images based on the connection graph to obtain the differential interferogram.

3. The method of claim 1, wherein, The environmental parameters comprise meteorological data, vegetation data and elevation data.

4. The method of claim 1, wherein, The method further comprises training an adaptive weight prediction model by the following method: constructing a sample set comprising a plurality of samples; wherein the samples comprise sample data and label data; the sample data comprises training differential interferograms and training environmental parameters; and the label data comprises training atmospheric correction interferograms; inputting the sample data into a to-be-trained atmospheric correction model to obtain prediction data output by the model, wherein the prediction data comprises atmospheric correction prediction images output by the model; determining the difference between the prediction data and the label data; updating the parameters of the to-be-trained atmospheric correction model by back propagation based on the difference until the difference between the prediction data and the label data is minimized, thereby obtaining the pre-trained atmospheric correction model.

5. The method of claim 4, wherein, The inputting of the sample data into the to-be-trained atmospheric correction model to obtain the prediction data output by the model comprises: the to-be-trained atmospheric correction model performs feature fusion on the training environmental parameters and the training differential interferograms to obtain training multi-modal fusion features; weight calculation is performed on the training multi-modal fusion features to obtain at least two correction weight coefficients; the training differential interferograms are corrected based on the at least two correction weight coefficients by the at least two sub atmospheric correction models to obtain the prediction data.

6. The method of claim 5, wherein, The to-be-trained atmospheric correction model performs feature fusion on the training environmental parameters and the training differential interferograms to obtain training multi-modal fusion features, which comprises: the to-be-trained atmospheric correction model performs feature extraction on the meteorological data based on a meteorological data encoder to obtain meteorological data features; feature extraction is performed on the vegetation data based on a vegetation data encoder to obtain vegetation data features; feature extraction is performed on the elevation data based on a digital elevation data encoder to obtain elevation digital features; feature extraction is performed on the differential interferogram based on an interferogram data encoder to obtain interferogram features; the meteorological data features, the vegetation data features, the elevation digital features and the interferogram features are spliced to obtain the training multi-modal fusion features.

7. The method of claim 5, wherein, The correction of the at least two sub-atmosphere correction models on the differential interferogram obtains at least two corrected interferograms. The correction of the at least two sub-atmosphere correction models on the differential interferogram obtains at least two corrected interferograms. The at least two corrected interferograms are weighted and fused based on the at least two correction weight coefficients to obtain the prediction data.

8. An InSAR atmospheric correction apparatus, characterized by, The method comprises: The interferogram determination module is configured to determine SAR images and orbit data in a preset space-time range, and perform interference on the SAR images based on the orbit data to obtain a differential interferogram. The interferogram determination module is specifically configured to:

9. The apparatus of claim 8, wherein, Determine SAR images and orbit data in a preset space-time range, and perform registration on the SAR images based on the orbit data to obtain a SAR single-view complex image stack. Generate a connection graph of the SAR single-view complex image stack, and perform interference on the SAR images based on the connection graph to obtain a differential interferogram. The interferogram determination module is specifically configured to:

10. The apparatus of claim 9, wherein, Determine environmental parameters in the preset space-time range, and correct the differential interferogram by using a pre-trained atmosphere correction model and the environmental parameters to obtain an atmosphere-corrected interferogram; wherein the environmental parameters comprise meteorological data, vegetation data, and elevation data; the pre-trained atmosphere correction model comprises at least two sub-atmosphere correction models, and the atmosphere-corrected interferogram is configured to invert ground surface deformation. The atmosphere correction model is constructed in the following manner: A sample set comprising a plurality of samples is constructed; wherein the samples comprise sample data and label data; the sample data comprises a training differential interferogram and training environmental parameters; and the label data comprises a training atmosphere-corrected interferogram; The sample data is input into a to-be-trained atmosphere correction model, and the to-be-trained atmosphere correction model extracts features of meteorological data based on a meteorological data encoder to obtain meteorological data features; Features of vegetation data are extracted based on a vegetation data encoder to obtain vegetation data features; Features of digital elevation data are extracted based on a digital elevation data encoder to obtain digital elevation features; Features of the differential interferogram are extracted based on an interferogram data encoder to obtain interferogram features; The meteorological data features, the vegetation data features, the digital elevation features, and the interferogram features are spliced to obtain training multi-modal fusion features; At least two correction weight coefficients are calculated based on the training multi-modal fusion features; The correction of the at least two sub-atmosphere correction models on the differential interferogram obtains at least two corrected interferograms. ​ The at least two corrected interferograms are fused by weighting based on the at least two correction weight coefficients to obtain prediction data, wherein the prediction data comprises an atmospheric correction prediction image output by the model; determining a difference between the prediction data and the label data; updating parameters of the to-be-trained atmospheric correction model by back propagation based on the difference until the difference between the prediction data and the label data is minimized, to obtain the pre-trained atmospheric correction model.

11. An electronic device, comprising: A computer program product comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of claims 1 to 7 when executing the program.

12. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium stores computer instructions for causing a computer to execute the method of any one of claims 1 to 7.

13. A computer program product, characterised in that, The computer program instructions, when executed on a computer, cause the computer to execute the method of any one of claims 1 to 7.

Images