Interseismic large-range interferogram block unwrapping method and related equipment

Abstract

The invention provides an inter-seismic wide-range interferogram partitioning and unwrapping method and related equipment. The inter-seismic wide-range interferogram partitioning and unwrapping method comprises the following steps: partitioning an inter-seismic interferogram of a research area into a plurality of sub-blocks with overlapping areas; inputting each sub-block into the trained neural network for integer ambiguity prediction to obtain integer ambiguity and confidence of each pixel point in each sub-block, averaging the confidence of all the pixel points in each sub-block to obtain the confidence of each sub-block, and respectively aiming at the confidence of each sub-block, calculating the integer ambiguity of each pixel point in each sub-block to obtain the integer ambiguity of each pixel point in each sub-block; unwrapping the corresponding sub-block again or correcting the integer ambiguity of each pixel point in the corresponding sub-block according to the confidence degree to obtain the updated integer ambiguity of each pixel point in each sub-block; and splicing the updated integer ambiguity of each pixel point in each sub-block to obtain an integer ambiguity map of the inter-seismic interference pattern, and obtaining an unwrapping interference pattern based on the integer ambiguity map and the inter-seismic interference pattern, thereby greatly improving the unwrapping precision of the island region.

Description

Technical Field

[0001] The present invention relates to the technical field of geodetic surveying, and particularly relates to a method for unwrapping large-scale interferograms in earthquake intervals and related equipment. Background Art

[0002] Spaceborne Interferometric Synthetic Aperture Radar (InSAR) has the advantages of wide coverage, high spatial resolution, and high precision in surface deformation monitoring. Currently, it has been widely used for large-scale surface deformation monitoring in the inter-seismic stage of large fault zones. InSAR can characterize the details of the strain rate field over a large area, reveal blind faults, and evaluate the regional earthquake risk. It can analyze the kinematic parameters along the fault zone and explore the dynamic driving factors of the fault zone, playing an important role in revealing the mechanism of crustal deformation. The wide spatial coverage and short revisit period of satellites provide rich data sources for obtaining large-scale inter-seismic deformations, but also bring new challenges to InSAR data processing.

[0003] Inter-seismic interferograms are composed of complete image frames (Frames) collected by 1-2 Sentinel satellites along the orbit. Moreover, since the covered areas are mostly in the wild, the surface coverage includes complex landforms such as deserts, vegetation, and snow, and its time span is relatively long, resulting in being affected by various decoherence factors and being prone to island areas. Local islands surrounded by decoherence noise are extremely likely to generate local unwrapping errors; and the decoherence noise will also cause the unwrapping errors to propagate in the interferogram, forming large-scale triangular closed-loop errors. Therefore, the low-coherence areas of large-scale inter-seismic interferograms are likely to cause unwrapping errors and their propagation, affecting the accuracy of subsequent time series solutions and the inversion of inter-seismic motion parameters of faults.

[0004] In existing related technical research, the unwrapping of large-scale inter-seismic interferograms can be divided into two categories: spatial dimension unwrapping and spatio-temporal dimension unwrapping. For spatial dimension unwrapping, the Minimum Cost Flow (MCF) is currently the most commonly used robust unwrapping method. When using this method, in order to reduce the influence of decoherence noise, a coherence threshold mask is required, but this threshold is difficult to determine and has limited effects. For spatio-temporal dimension unwrapping, related research usually introduces time dimension information as a constraint to guide or cooperate with the spatial dimension to complete the unwrapping. Due to the characteristics of "large scale - small gradient - multi-decoherence - severe atmospheric changes" of inter-seismic deformation interferograms, relevant scholars considered the triangular closure characteristics in the time dimension of multi-temporal interferograms and proposed an iterative weighted unwrapping method on the basis of traditional three-dimensional unwrapping methods to gradually reduce large-area unwrapping errors.

[0005] In addition to improving the unwrapping method itself, the correction of unwrapping errors is also a research focus for improving the accuracy of interseismic data processing. The correction of unwrapping errors is also divided into two categories: spatial dimension correction and temporal dimension correction. Spatial correction relies on connectivity and is not applicable to island regions; temporal dimension correction depends on redundant interferometric observations in the time dimension, which is difficult to achieve for low coherence points, and the correction effect is limited by the solution method itself.

[0006] Therefore, for large-scale multi-temporal interseismic deformation interferograms, introducing temporal dimension information for unwrapping and error correction may face problems such as inefficiency and invalidity; and the dependence of the spatial dimension on the traditional phase continuity assumption is the root cause of unwrapping errors in the decoherence area. Therefore, the above two existing unwrapping methods have low unwrapping accuracy for interseismic interferograms. Summary of the Invention

[0007] The present invention provides a method for block-wise unwrapping of large-scale interseismic interferograms and related equipment, aiming to improve the unwrapping accuracy of large-scale low coherence interseismic interferograms.

[0008] To achieve the above objective, the present invention provides a method for block-wise unwrapping of large-scale interseismic interferograms, including: Step 1, obtaining an interseismic interferogram of the research area and dividing the interseismic interferogram into a plurality of sub-blocks with overlapping areas; Step 2, respectively for each sub-block with overlapping areas, inputting the sub-block into a trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block; Step 3, taking the average value of the confidence of all pixel points in each sub-block to obtain the confidence of each sub-block. Respectively for the confidence of each sub-block, re-unwrap the corresponding sub-block or correct the integer ambiguity of each pixel point in the corresponding sub-block according to the size of the confidence to obtain the updated integer ambiguity of each pixel point in each sub-block; Step 4, splicing the updated integer ambiguity of each pixel point in each sub-block to obtain an integer ambiguity map of the interseismic interferogram, and obtaining an unwrapped interferogram based on the integer ambiguity map and the interseismic interferogram.

[0009] Furthermore, training the neural network includes: Obtaining a plurality of interferograms processed by the Lics platform as real samples; Generating simulated samples and sample labels based on a plurality of real coherence maps and digital elevation models of the research area; Training the neural network using the real samples and the simulated samples to obtain a trained neural network.

[0010] Furthermore, generating simulated samples and sample labels based on a number of real coherence maps and digital elevation data in the research area, including: Simulating decorrelation noise based on a number of real coherence maps in the research area; Simulating vertically stratified atmosphere, turbulent atmosphere, orbital errors, and interseismic deformation based on the regional digital elevation model and mathematical models, and adding the turbulent atmosphere, orbital errors, interseismic deformation, and vertically stratified atmosphere to obtain the unwrapped true value of the simulated interseismic interferogram; Adding the decorrelation noise and the unwrapped true value, then performing wrapping and filtering to obtain the simulated interferogram and the coherence corresponding to the simulated interferogram, and masking the integer ambiguity of low-coherence pixels in the simulated interferogram to 0 using the coherence corresponding to the simulated interferogram; Subtracting the unwrapped true value from the simulated interferogram to obtain the true value of the integer ambiguity of each pixel in the simulated interferogram; Dividing the simulated interferogram and the true value of the integer ambiguity of each pixel in the simulated interferogram into blocks to obtain simulated samples.

[0011] Furthermore, the confidence of each pixel point within each sub-block is the difference between the maximum probability predicted by the trained neural network for each pixel point within each sub-block and the second-largest probability; The confidence of each sub-block includes the average credibility of each pixel and the average credibility of non-decorrelated pixels.

[0012] Furthermore, taking the average of the confidence of all pixel points within each sub-block to obtain the confidence of each sub-block, including: For each sub-block, taking the average of the confidence of all pixel points within the sub-block to obtain the average credibility of each pixel within each sub-block; For each sub-block, taking the average of the confidence of all non-decorrelated pixel points within the sub-block to obtain the average credibility of non-decorrelated pixels within each sub-block; The average credibility of each pixel within each sub-block and the average credibility of non-decorrelated pixels within each sub-block constitute the confidence of each sub-block.

[0013] Furthermore, respectively for the confidence of each sub-block, re-unwrapping the corresponding sub-block or correcting the integer ambiguity of the corresponding sub-block according to the magnitude of the confidence, including: Respectively for each sub-block, when the average credibility of pixels in the confidence of the sub-block is less than the first preset threshold or the average credibility of non-decorrelated pixels is less than the second preset threshold, re-unwrap the sub-block using the MCF algorithm; When the average pixel credibility in the confidence of the sub-block is greater than or equal to the first preset threshold and less than the third preset threshold, or the average non-decoherence pixel credibility is greater than or equal to the second preset threshold and less than the third preset threshold, the trained DeeplabV3+ model is used to correct the integer ambiguity of the sub-block.

[0014] Furthermore, step 4 includes: For each sub-block, the integer ambiguity of each pixel point in the overlapping area on the sub-block is subtracted from the integer ambiguity of each pixel point in the overlapping area on the adjacent sub-block to obtain the relative difference value of each pixel point in the overlapping area between two adjacent sub-blocks, and the mode of the relative difference values of each pixel point in the overlapping area between two adjacent sub-blocks is used as the relative difference observation value of the overlapping area; Based on the proportion of decoherence pixel points in the overlapping area and the pixel proportion corresponding to the mode of the relative difference values, the overlapping area is screened, and the relative difference observation value of the corresponding overlapping area is retained or discarded according to the screening result; An observation equation is established according to the retained relative difference observation value, and the weighted least squares method is used to solve the observation equation to obtain the relative difference value of the integer ambiguity of the whole of each sub-block relative to the whole of the first sub-block; The integer ambiguity of each pixel point in each sub-block is corrected according to the relative difference value of the integer ambiguity of the whole of each sub-block relative to the whole of the first sub-block to obtain the corrected integer ambiguity; The mean value of the corrected integer ambiguities of all pixel points in the overlapping area on each sub-block is taken to obtain the average integer ambiguity, and the integer ambiguity of the interseismic interferogram is spliced based on the corrected integer ambiguity and the average integer ambiguity to obtain the integer ambiguity map of the interseismic interferogram; Through the formula , the integer ambiguity map of the interseismic interferogram is added to the interseismic interferogram to obtain the unwrapped interferogram, where represents the unwrapped phase of the pixel point, represents the interference phase of the pixel point, represents the integer ambiguity of the pixel point.

[0015] Furthermore, the expression of the observation equation is: ; where represents the relative difference observation value, , represents the th sub-block and the th sub-block on the relative difference observation value of the overlapping area, represents the coefficient matrix, , It represents the relative difference value of the integer ambiguity of each sub-block as a whole with respect to the first sub-block as a whole. , It represents that the interseismic interferogram includes sub-blocks, It represents the relative difference value of the integer ambiguity of the th sub-block as a whole with respect to the first sub-block as a whole.

[0016] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method for unwrapping the interseismic large-scale interferogram by block.

[0017] The present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, it implements the method for unwrapping the interseismic large-scale interferogram by block.

[0018] The above solution of the present invention has the following beneficial effects: The present invention divides the interseismic interferogram of the research area obtained into several sub-blocks with overlapping areas; for each sub-block with overlapping areas, the sub-block is input into the trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block. The average value of the confidence of all pixel points in each sub-block is taken to obtain the confidence of each sub-block. For the confidence of each sub-block respectively, according to the size of the confidence, the corresponding sub-block is re-unwrapped or the integer ambiguity of each pixel point in the corresponding sub-block is corrected to obtain the updated integer ambiguity of each pixel point in each sub-block; the updated integer ambiguities of each pixel point in each sub-block are stitched together to obtain the integer ambiguity map of the interseismic interferogram, and the unwrapped interferogram is obtained based on the integer ambiguity map and the interseismic interferogram; compared with the prior art, the present invention predicts the integer ambiguity through the trained neural network, re-unwraps the corresponding sub-block according to the size of the confidence or corrects the integer ambiguity of each pixel point in the corresponding sub-block, and obtains the updated integer ambiguity of each pixel point in each sub-block, which can consider the global information of each sub-block, can more accurately predict the integer ambiguity of the isolated island area, and can also automatically identify the de-coherent and low-coherent areas through the trained neural network, reducing the negative impact of the de-coherence noise on the traditional spatial dimension unwrapping while improving the processing efficiency, and greatly improving the unwrapping accuracy of the isolated island area.

[0019] Other beneficial effects of the present invention will be described in detail in the subsequent specific implementation part. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic flowchart of an embodiment of the present invention; Figure 2 It is the structural diagram of the neural network in the embodiment of the present invention; Figure 3 It is the flow chart of simulated sample generation in the embodiment of the present invention; Figure 4 It is the comparison chart of the unwrapping accuracy between the traditional method and this method on 25 simulated interferograms; Figure 5 It is the statistical comparison chart of the unwrapping accuracy between the traditional method and this method on real interferograms; Figure 5 (a) is the schematic diagram of the normalized frequency distribution of the closed-loop UEP under two unwrapping methods, Figure 5 (b) is the schematic diagram of the specific performance of this method on each closed loop compared with the traditional method; Figure 6 It is the structural schematic diagram of the terminal device in the embodiment of the present invention. Specific embodiments

[0021] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] In the description of the present invention, it should be noted that the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0023] In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, the term "connection" should be understood in a broad sense. For example, it can be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meaning of the above terms in the present invention can be understood according to specific situations.

[0024] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] The present invention provides an inter-seismic large-scale interferogram block unwrapping method and related equipment for existing problems.

[0026] As Figure 1 shown, the embodiment of the present invention provides an inter-seismic large-scale interferogram block unwrapping method, including: Step 1, obtain the inter-seismic interferogram of the research area and divide the inter-seismic interferogram into several sub-blocks with overlapping areas; Step 2: For each sub-block with an overlapping area, input the sub-block into the trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block. Step 3: Take the average of the confidences of all pixel points in each sub-block to obtain the confidence of each sub-block. For the confidence of each sub-block, re-unwrap the corresponding sub-block according to the magnitude of the confidence or correct the integer ambiguity of each pixel point in the corresponding sub-block to obtain the updated integer ambiguity of each pixel point in each sub-block. Step 4: Stitch the updated integer ambiguities of each pixel point in each sub-block to obtain the integer ambiguity map of the inter-seismic interferogram, and obtain the unwrapped interferogram based on the integer ambiguity map and the inter-seismic interferogram.

[0027] In the embodiment of the present invention, the west end of a certain fault zone in the northern part of a certain plateau is used as the research area. The terrain at the west end of this fault zone has large undulations, and the height difference can reach more than three thousand meters. The phase of the interferogram is mainly dominated by inter-seismic deformation and atmospheric noise, and it contains complex landforms such as deserts, plateau permafrost, and snow cover. The decorrelation noise of the interferogram is complex and serious, and there are many isolated areas caused by desert decorrelation. Therefore, taking this fault zone as the research area, the method provided in the embodiment of the present invention is specifically described as follows: Step 1: Obtain the inter-seismic interferogram of the west end of a certain fault zone in the northern part of a certain plateau. The time span of data collection is from January 2019 to June 2022. Use Gamma software for interferometric processing, and only select the interferograms with a time baseline greater than 300 days and a spatial baseline less than 20m to obtain 763 long-time baseline inter-seismic deformation interferograms. Segment the inter-seismic interferogram to obtain several sub-blocks with overlapping areas. The size of the sub-block is set to 2048×2048 pixels, and the corresponding moving step is 1600 pixels, that is, the size of the overlapping area in the inter-seismic interferogram is 448×2048 pixels.

[0028] Step 2: For each sub-block with an overlapping area, input the sub-block into the trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block.

[0029] Specifically, the neural network is the SegFormer (Simple and Efficient Design for Semantic Segmentation with Transformers) neural network based on the Transformer architecture. This neural network combines the advantages of the Transformer and the convolutional network and has good robustness performance in tasks such as semantic segmentation. Its network structure is as Figure 2 shown, including: An encoder consisting of an embedding module, a first Transformer module, a second Transformer module, a third Transformer module, and a fourth Transformer module connected in sequence; A decoder consisting of a first multi-layer perceptron module and a second multi-layer perceptron module connected in sequence; The output ends of the first Transformer module, the second Transformer module, the third Transformer module, and the fourth Transformer module are all connected to the input end of the first multi-layer perceptron module; The first Transformer module, the second Transformer module, the third Transformer module, and the fourth Transformer module each include an efficient attention layer, a hybrid feed-forward network layer, and an overlapping block fusion layer connected in sequence. The input end of the efficient attention layer is the input end of the corresponding Transformer module, and the output end of the overlapping block fusion layer is the output end of the corresponding Transformer module; The first multi-layer perceptron module and the second multi-layer perceptron module each include a multi-layer perceptron and an upsampling layer connected in sequence. Among them, the input end of the multi-layer perceptron is the input end of the corresponding multi-layer perceptron module, and the output end of the upsampling layer is the output end of the corresponding multi-layer perceptron module.

[0030] In the neural network in the embodiments of the present invention, in the encoder, it adopts a new overlapping Patch without position encoding, thus avoiding the interpolation of position encoding, increasing the flexibility of the model and the adaptability to inputs of various resolutions; a hybrid feed-forward network layer (Mix-FFN) is introduced, which can capture local and global features simultaneously; in the decoder, a lightweight multi-layer perceptron (MLP) is used to complete decoding, aggregating information from different layers, so as to combine local attention and global attention to present a powerful representation.

[0031] Specifically, training the neural network includes: Obtaining a number of interferograms processed by the Lics platform as real samples; Generating simulated samples and sample labels based on a number of real coherence maps and digital elevation models of the research area; Training the neural network using the real samples and the simulated samples to obtain the trained neural network.

[0032] In order to fully train the neural network, the embodiments of the present invention adopt simulation generation and the Lics (Looking into the Continents from Space) platform for processing, obtaining nearly 40,000 data samples with decoherence noise of different degrees, covering rich pattern features such as deformation, atmospheric delay, and orbital error; considering the limited video memory, the spatial dimensions of all training samples are downsampled to 512×512 pixels.

[0033] Specifically, based on a number of real coherence maps and digital elevation data, simulation samples and sample labels are generated, including: Simulating decoherence noise based on a number of real coherence maps of the research area; Simulating vertically stratified atmosphere, turbulent atmosphere, orbital error, and interseismic deformation based on the regional digital elevation model and mathematical models, and adding the turbulent atmosphere, orbital error, interseismic deformation, and vertically stratified atmosphere to obtain the unwrapped true value of the simulated interseismic interferogram; After adding the decoherence noise and the unwrapped true value, performing wrapping and filtering to obtain the simulated interferogram and the coherence corresponding to the simulated interferogram, and using the coherence corresponding to the simulated interferogram to mask the integer ambiguity of the low-coherence pixels in the simulated interferogram as 0; Subtracting the unwrapped true value from the simulated interferogram to obtain the integer ambiguity true value of each pixel in the simulated interferogram; Partitioning the simulated interferogram and the integer ambiguity true value of each pixel in the simulated interferogram to obtain simulation samples.

[0034] First, in the embodiments of the present invention, on the Lics platform, for the ascending-track data at the west end of the fault zone in the northern part of a certain plateau, interferograms without unwrapping errors and with complex patterns are visually selected, and manual corrections are made to some samples to obtain 13,500 real samples; Subsequently, as Figure 3 shown, in the embodiments of the present invention, decoherence noise is simulated based on a number of real coherence maps at the west end of the fault zone in the northern part of a certain plateau, vertically stratified atmosphere is simulated based on the regional digital elevation model, turbulent atmosphere, orbital error, and interseismic deformation are respectively simulated based on mathematical models such as fractal function, quadratic polynomial, and screw dislocation, and the turbulent atmosphere, orbital error, interseismic deformation, and vertically stratified atmosphere are added to obtain the unwrapped true value of the simulated interseismic interferogram; Then, after adding the decoherence noise and the unwrapped true value, performing wrapping and filtering to obtain the simulated interferogram and the coherence corresponding to the simulated interferogram, and using the coherence corresponding to the simulated interferogram to mask the integer ambiguity of the low-coherence pixels in the simulated interferogram as 0; Furthermore, subtracting the unwrapped true value from the simulated interferogram to obtain the integer ambiguity true value (1-6) of each pixel in the simulated interferogram; Finally, the simulated interference pattern and the true values of the integer ambiguities of each pixel in the simulated interference pattern are segmented to obtain multiple sub-block interference pattern samples and their corresponding integer ambiguity labels (0 - 6). After manually screening multiple sub-block interference pattern samples, more than 26,000 groups of simulated samples are obtained.

[0035] Specifically, the confidence of each pixel point within each sub-block is the difference between the maximum probability and the second-largest probability predicted by the trained SegFormer network for each pixel point within each sub-block. The confidence of each sub-block includes the average credibility of each pixel and the average credibility of non-decoherent pixels.

[0036] Considering that there are certain errors in the integer ambiguities predicted by the neural network, which will be directly reflected in the unwrapped phase and affect subsequent stitching, in the embodiments of the present invention, the confidence is calculated for the class prediction probability output by each pixel point. The confidence is the difference between the maximum probability and the second-largest probability predicted by the neural network at each pixel point; the greater the difference between the two, the more accurate the prediction of the neural network and the greater the credibility of the prediction result.

[0037] Specifically, the average value of the confidences of all pixel points within each sub-block is taken to obtain the confidence of each sub-block, including: For each sub-block, the average value of the confidences of all pixel points within the sub-block is taken to obtain the average credibility of each pixel within each sub-block ; For each sub-block, the average value of the confidences of all non-decoherent pixel points within the sub-block is taken to obtain the average credibility of non-decoherent pixels within each sub-block ; The average credibility of each pixel within each sub-block and the average credibility of non-decoherent pixels within each sub-block constitute the confidence of each sub-block .

[0038] Specifically, for the confidence of each sub-block respectively, according to the size of the confidence, the corresponding sub-block is re-unwrapped or the integer ambiguity of the corresponding sub-block is corrected, including: For each sub-block respectively, when the average credibility of pixels in the confidence of the sub-block is less than the first preset threshold or the average credibility of non-decoherent pixels is less than the second preset threshold, the MCF algorithm is used to re-unwrap the sub-block; When the average credibility of pixels in the confidence of the sub-block is greater than or equal to the first preset threshold and less than the third preset threshold or the average credibility of non-decoherent pixels is greater than or equal to the second preset threshold and less than the third preset threshold, the trained DeeplabV3+ model is used to correct the integer ambiguity of the sub-block.

[0039] In an embodiment of the present invention, the first preset threshold is set to 0.8, the second preset threshold is set to 0.65, and the third preset threshold is set to 0.95. When the average pixel credibility in the confidence of a sub-block is less than 0.8 or the average non-decoherence pixel credibility is less than 0.65, it indicates that a large area of prediction errors may have occurred in the sub-block, and the MCF algorithm needs to be used to re-unwrap the sub-block; when the average pixel credibility in the confidence of a sub-block is greater than or equal to 0.8 and less than 0.95 or the average non-decoherence pixel credibility is greater than or equal to 0.65 and less than 0.95, it indicates that prediction errors may only occur in some areas. At this time, the trained DeeplabV3+ model is used to correct the integer ambiguity of the sub-block.

[0040] It should be noted that the MCF algorithm is a conventional unwrapping method in the art, and the specific process of unwrapping through this algorithm is well-known technology. Therefore, the specific process of unwrapping through this algorithm will not be described in detail in the embodiments of the present invention.

[0041] Specifically, using the trained DeeplabV3+ model to correct the integer ambiguity of the sub-block includes: Adding the phase of each sub-block to the integer ambiguity output by the previous neural network according to the basic formula of InSAR phase unwrapping to obtain the initial unwrapping result of the sub-block, and using this initial unwrapping result as the input of the trained DeeplabV3+ model. Each pixel point in the sub-block is then classified into three categories: 0, 1, and 2 by the trained DeeplabV3+ model. Among them, categories 1 and 2 respectively indicate that there is an error in the integer ambiguity output by the neural network for this pixel, and 1 needs to be subtracted or added to it respectively to achieve correction, while category 0 indicates that there is no error in the integer ambiguity output by the neural network for this pixel point and no correction is required.

[0042] The DeeplabV3+ model in the embodiments of the present invention is also trained by simulating and generating nearly 40,000 data samples with different degrees of decoherence noise processed by the Lics (Looking into the Continents from Space) platform.

[0043] Specifically, step 4 includes: For each sub-block respectively, taking the difference between the integer ambiguity of each pixel point in the overlapping area of the sub-block and the integer ambiguity of each pixel point in the overlapping area of the adjacent sub-block to obtain the relative difference value of each pixel point in the overlapping area of the adjacent two sub-blocks, and taking the mode of the relative difference values of each pixel point in the overlapping area of the adjacent two sub-blocks as the relative difference observation value of the overlapping area; Based on the proportion of decoherence pixel points in the overlapping area and the proportion of pixels corresponding to the mode of the relative difference values, the overlapping area is screened, and the relative difference observation value of the corresponding overlapping area is retained or discarded according to the screening result; An observation equation is established based on the reserved relative difference observations, and the weighted least squares method is used to solve the observation equation to obtain the relative difference value of the integer ambiguity of each sub-block as a whole relative to the first sub-block as a whole; The integer ambiguity of each pixel point in each sub-block is corrected according to the relative difference value of the integer ambiguity of each sub-block as a whole relative to the first sub-block as a whole, and the corrected integer ambiguity is obtained; The mean value of the corrected integer ambiguities of all pixel points in the overlapping area on each sub-block is taken to obtain the average integer ambiguity, and the integer ambiguity of the inter-seismic interferogram is mosaicked based on the corrected integer ambiguity and the average integer ambiguity to obtain the integer ambiguity map of the inter-seismic interferogram; Through the formula , the integer ambiguity map of the inter-seismic interferogram is added to the inter-seismic interferogram to obtain the unwrapped interferogram, where represents the unwrapped phase of the pixel point, represents the interference phase of the pixel point, represents the integer ambiguity of the pixel point.

[0044] In the embodiment of the present invention, considering that the difference in the overlapping area is theoretically a fixed value, the mode of the relative difference values of each pixel point in the overlapping area is taken as the relative difference observation value of the overlapping area; for the overlapping area where the proportion of decorrelated pixels exceeds 97% and the proportion of mode pixels is less than 80%, it is considered that there is a large unwrapping error in the overlapping area, and the relative difference observation value of the overlapping area is discarded.

[0045] Specifically, the expression of the observation equation is: ; where represents the relative difference observation value, , represents the relative difference observation value of the overlapping area between the th sub-block and the th sub-block, represents the coefficient matrix, , represents the relative difference value of the integer ambiguity of each sub-block as a whole relative to the first sub-block as a whole, , represents that the inter-seismic interferogram includes sub-blocks, represents the relative difference value of the integer ambiguity of the th sub-block as a whole relative to the first sub-block as a whole.

[0046] In the embodiment of the present invention, for an interferogram containing sub-blocks, it is necessary to solve parameters, there are observations of overlapping regions without discarding the overlapping regions.

[0047] It should be noted that considering the relative difference value of the integer ambiguity of each sub-block as a whole relative to the first sub-block as a whole is theoretically an integer, and the solution residual should theoretically be 0. For the interferogram with a solution residual greater than 0.00001, it will be discarded, otherwise large-area stitching errors will occur.

[0048] The unwrapping results obtained in the embodiments of the present invention and the unwrapping results obtained by the traditional MCF method are compared with the simulated true values. The accuracy evaluation index adopts a self-defined index - the proportion of unwrapping error (Unwrapping Error Proportion, UEP), that is, the ratio of the number of pixels with unwrapping errors to the number of pixels in the interferogram that are not masked during unwrapping. In this comparative experiment, the pixel points with the absolute value of the difference between the unwrapping result and the simulated true value reaching 5.5 radians are regarded as the pixel points with unwrapping errors. The UEP values of the two unwrapping methods on all simulated interferograms are respectively counted. The average UEP of the traditional MCF method is 2%, while the average UEP of the method provided in the embodiments of the present invention is only 0.8%. Therefore, it can be considered that the average accuracy of the method provided in the embodiments of the present invention on this simulated data set is improved by 60% compared with the traditional method. Figure 4 The UEP comparison of 25 simulated interferograms is shown, and it can be seen that the UEP of most interferograms based on the method provided in the embodiments of the present invention has been significantly reduced.

[0049] The embodiments of the present invention are applied to real data unwrapping. The unwrapping results obtained in the embodiments of the present invention and the unwrapping results obtained by the traditional MCF method are compared by means of triangular phase closure difference detection to judge the unwrapping error. Based on the remaining 739 interferograms, more than 1300 triangular closed loops are formed. The accuracy evaluation index also adopts the proportion of unwrapping error (Unwrapping Error Proportion, UEP). In the accuracy evaluation of this real data, the pixel points with the absolute value reaching 5.5 radians in the triangular closed loop are regarded as the pixel points with unwrapping errors. The UEP values of the two unwrapping methods on all triangular closed loops are respectively counted. The average UEP of the traditional MCF method is 1.26%, while the average UEP of the method provided in the embodiments of the present invention is only 0.77%. Therefore, it can be considered that the average accuracy of the method provided in the embodiments of the present invention on this real data set is improved by 38% compared with the traditional method. Figure 5 (a) shows the normalized frequency distribution of the closed-loop UEP under the two unwrapping methods. Obviously, the closed-loop UEP of the method provided in the embodiments of the present invention is concentrated in smaller values. Figure 5(b)Further shows the specific performance of the method provided by the embodiments of the present invention on each closed loop compared with the traditional method. It can be seen that about 56% of the closed-loop UEP has not changed significantly, and although about 11.7% of the closed-loop UEP has increased significantly, more than 30% of the closed-loop UEP has decreased significantly. Therefore, overall, the method provided by the embodiments of the present invention effectively reduces the unwrapping error of the large-scale inter-seismic interferogram.

[0050] The embodiments of the present invention divide the obtained inter-seismic interferogram of the research area into several sub-blocks with overlapping areas; for each sub-block with overlapping areas, the sub-block is input into the trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block. Take the average value of the confidence of all pixel points in each sub-block to obtain the confidence of each sub-block. For the confidence of each sub-block, re-unwrap the corresponding sub-block according to the size of the confidence or correct the integer ambiguity of each pixel point in the corresponding sub-block to obtain the updated integer ambiguity of each pixel point in each sub-block; splice the updated integer ambiguity of each pixel point in each sub-block to obtain the integer ambiguity map of the inter-seismic interferogram, and obtain the unwrapped interferogram based on the integer ambiguity map and the inter-seismic interferogram; compared with the prior art, the embodiments of the present invention perform integer ambiguity prediction through the trained neural network, re-unwrap the corresponding sub-block according to the size of the confidence or correct the integer ambiguity of each pixel point in the corresponding sub-block to obtain the updated integer ambiguity of each pixel point in each sub-block, which can consider the global information of each sub-block, can more accurately predict the integer ambiguity of the island area, and can also automatically identify the decorrelation and low coherence areas through the trained neural network, reducing the negative impact of the decorrelation noise on the traditional spatial dimension unwrapping while improving the processing efficiency, and greatly improving the unwrapping accuracy of the island area.

[0051] The embodiments of the present invention also provide a terminal device, such as Figure 6 shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 6 only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100. When the processor D100 executes the computer program D102, the above-mentioned method for block-wise unwrapping of large-scale inter-seismic interferograms is implemented.

[0052] The terminal device D10 may be a computing device such as a desktop computer, a notebook, a palm computer, a server, a server cluster, and a cloud server. The terminal device may include, but is not limited to, a processor D100 and a memory D101. Those skilled in the art can understand, Figure 6This is only an example of the terminal device D10, which does not constitute a limitation on the terminal device D10. It may include more or fewer components than those shown in the figure, or combine certain components, or different components. For example, it may also include input / output devices, network access devices, etc.

[0053] The so-called processor D100 may be a central processing unit (CPU, Central Processing Unit), and this processor D100 may also be other general-purpose processors, digital signal processors (DSP, Digital Signal Processor), application-specific integrated circuits (ASIC, Application Specific Integrated Circuit), off-the-shelf programmable gate arrays (FPGA, Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0054] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as the hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart media card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card (Flash Card), etc. equipped on the terminal device D10. Further, the memory D101 may also include both the internal storage unit and the external storage device of the terminal device D10. The memory D101 is used to store an operating system, application programs, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory D101 may also be used to temporarily store data that has been output or will be output.

[0055] It should be noted that for the content such as information interaction and execution process between the above-mentioned devices / units, since it is based on the same concept as the method embodiment of the present application, for its specific functions and the technical effects brought, please refer to the method embodiment part for details, and will not be elaborated here.

[0056] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In practical applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0057] An embodiment of the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for unwrapping large-range interferograms in earthquake intervals.

[0058] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above method embodiments of the present application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above-mentioned method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the construction device / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disc, etc.

[0059] The above is the preferred implementation manner of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for unwrapping large-scale interferograms in the inter-seismic period, characterized in that Including: Step 1: Obtain the inter-seismic interferogram of the research area and divide the inter-seismic interferogram into several sub-blocks with overlapping areas; Step 2: For each sub-block with an overlapping area, input the sub-block into the trained neural network for integer ambiguity prediction to obtain the integer ambiguity of each pixel point in each sub-block and the confidence of each pixel point in each sub-block; Step 3: Take the average value of the confidence of all pixel points in each sub-block to obtain the confidence of each sub-block. For the confidence of each sub-block, re-unwrap the corresponding sub-block according to the size of the confidence or correct the integer ambiguity of each pixel point in the corresponding sub-block to obtain the updated integer ambiguity of each pixel point in each sub-block; Step 4: Stitch the updated integer ambiguities of each pixel point in each sub-block to obtain the integer ambiguity map of the inter-seismic interferogram, and obtain the unwrapped interferogram based on the integer ambiguity map and the inter-seismic interferogram.

2. The method for unwrapping large-scale interferograms in earthquake intervals according to claim 1, characterized in that, Training the neural network includes: Obtain several interferograms processed by the Lics platform as real samples; Generate simulated samples and sample labels based on several real coherence maps and digital elevation models of the research area; Use the real samples and the simulated samples to train the neural network to obtain the trained neural network.

3. The method for unwrapping large-scale interferograms in the interseismic period according to claim 2, wherein Generating simulated samples and sample labels based on several real coherence maps and digital elevation data of the research area includes: Simulate the decorrelation noise based on several real coherence maps of the research area; Simulate the vertically stratified atmosphere, turbulent atmosphere, orbit error, and inter-seismic deformation based on the regional digital elevation model and mathematical model, and add the turbulent atmosphere, the orbit error, the inter-seismic deformation, and the vertically stratified atmosphere to obtain the unwrapped true value of the simulated inter-seismic interferogram; Add the decorrelation noise and the unwrapped true value, then perform wrapping and filtering to obtain the simulated interferogram and the coherence corresponding to the simulated interferogram, and use the coherence corresponding to the simulated interferogram to mask the integer ambiguity of the low-coherence pixels in the simulated interferogram as 0; Subtract the unwrapped true value from the simulated interferogram to obtain the integer ambiguity true value of each pixel in the simulated interferogram; Block the simulated interferogram and the integer ambiguity true value of each pixel in the simulated interferogram to obtain simulated samples.

4. The method for block-based unwrapping of large-scale inter-seismic interferograms according to claim 2, wherein The confidence of each pixel point in each sub-block is the difference between the maximum probability predicted by the trained neural network for each pixel point in each sub-block and the second largest probability; The confidence of each sub-block includes the average credibility of each pixel and the average credibility of non-decorrelated pixels.

5. The method for unwrapping large-scale interferograms in the inter-seismic period according to claim 4, wherein Taking the average value of the confidence of all pixel points in each sub-block to obtain the confidence of each sub-block includes: For each sub-block, take the average value of the confidence of all pixel points in the sub-block to obtain the average credibility of each pixel in each sub-block; For each sub-block, take the average value of the confidence of all non-decorrelated pixel points in the sub-block to obtain the average credibility of non-decorrelated pixels in each sub-block; The confidence of each sub-block is composed of the average pixel confidence of each sub-block and the average non-decoherence pixel confidence of each sub-block.

6. The method for unwrapping large-scale interferograms in the inter-seismic period according to claim 5, wherein For the confidence of each sub-block respectively, according to the magnitude of the confidence, re-unwrap the corresponding sub-block or correct the integer ambiguity of the corresponding sub-block, including: For each sub-block respectively, when the average pixel confidence in the confidence of the sub-block is less than the first preset threshold or the average non-decoherence pixel confidence is less than the second preset threshold, use the MCF algorithm to re-unwrap the sub-block; When the average pixel confidence in the confidence of the sub-block is greater than or equal to the first preset threshold and less than the third preset threshold, or the average non-decoherence pixel confidence is greater than or equal to the second preset threshold and less than the third preset threshold, use the trained DeeplabV3+ model to correct the integer ambiguity of the sub-block.

7. The method for unwrapping large-scale interferograms in earthquake intervals according to claim 6, characterized in that, Step 4 includes: For each sub-block respectively, subtract the integer ambiguity of each pixel point in the overlapping area of the sub-block from the integer ambiguity of each pixel point in the overlapping area of the adjacent sub-block to obtain the relative difference value of each pixel point in the overlapping area of two adjacent sub-blocks, and take the mode of the relative difference values of each pixel point in the overlapping area of two adjacent sub-blocks as the relative difference observation value of the overlapping area; Based on the proportion of decoherence pixel points in the overlapping area and the pixel proportion corresponding to the mode of the relative difference values, screen the overlapping area, and retain or discard the relative difference observation value of the corresponding overlapping area according to the screening result; Establish an observation equation according to the retained relative difference observation value, and use the weighted least squares method to solve the observation equation to obtain the relative difference value of the integer ambiguity of the whole of each sub-block relative to the whole of the first sub-block; Correct the integer ambiguity of each pixel point in each sub-block according to the relative difference value of the integer ambiguity of the whole of each sub-block relative to the whole of the first sub-block to obtain the corrected integer ambiguity; Take the average value of the corrected integer ambiguities of all pixel points in the overlapping area of each sub-block to obtain the average integer ambiguity, and splice the integer ambiguity of the inter-seismic interferogram based on the corrected integer ambiguity and the average integer ambiguity to obtain the integer ambiguity map of the inter-seismic interferogram; Through the formula , add the integer ambiguity map of the inter-seismic interferogram to the inter-seismic interferogram to obtain an unwrapped interferogram, where represents the unwrapped phase of the pixel represents the interference phase of the pixel represents the integer ambiguity of the pixel 8. The method for unwrapping large-scale interferograms in interseismic period according to claim 7, characterized in that, The expression of the observation equation is: Among them, represents the relative difference observation value, , represents the relative difference observation value of the overlapping area on the th sub-block and the th sub-block, represents the coefficient matrix, , represents the relative difference value of the integer ambiguity of each sub-block as a whole relative to the first sub-block as a whole, , represents that the interseismic interferogram includes sub-blocks, represents the relative difference value of the integer ambiguity of the th sub-block as a whole relative to the first sub-block as a whole.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the inter-seismic large-range interferogram block unwrapping method according to any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the inter-seismic large-range interferogram block unwrapping method according to any one of claims 1 to 8.

Images