Error correction method and system for radar differential interferometric image pairs based on reference surface

By selecting high coherence points and mountain apex as reference points in radar differential interference image pairing, interpolation is done to the reference surface, the problem of difficulty in removing atmospheric influence is solved, the accuracy and consistency of deformation monitoring is improved, and error accumulation is reduced.

CN118584435BActive Publication Date: 2024-11-19NANJING UNIV OF INFORMATION SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411060843.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-19
Estimated Expiration
2044-08-05

AI Technical Summary

Technical Problem

The prior art is difficult to effectively weaken the atmospheric influence in radar differential interference image pairs, resulting in the accumulation of errors and increased uncertainty of deformation monitoring results, especially when multiple processings in a large range, the deformation results at the same position are inconsistent.

Method used

By meshing in the target area, cells whose coherence coefficient, sky width and terrain position index are all higher than the threshold are selected as reference points, interpolate to the reference surface, and remove the reference surface in the interference image pair to correct the data.

Benefits of technology

It realizes effective removal of atmospheric delay phase, reduces error accumulation, improves the accuracy and consistency of deformation monitoring, and reduces the impact of other noise errors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118584435B_ABST
    Figure CN118584435B_ABST
Patent Text Reader

Abstract

The present invention discloses an error correction method and system for radar differential interferometric image pairs based on reference surfaces. First, the interferometric image pairs and the interference coefficient time series diagram of the target area are obtained by using the differential radar interferometric method; then the target area is gridded, and a number of pixels whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds are selected in each grid as reference points; the reference points are interpolated to obtain the reference surface; finally, the reference surface is removed from the interferometric image pairs to obtain the corrected interferometric image pairs. The present invention obtains differential interferometric image pairs with atmospheric delay and other error corrections with the same spatiotemporal resolution as the original interferometric image pairs. While quickly and effectively removing the atmospheric delay phase over a large range, other errors such as the snow depth phase caused by snow accumulation can also be effectively suppressed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of remote sensing image data processing, in particular to an error correction method and system for a radar differential interference image pair based on a reference surface. Background Art

[0002] When microwave signals emitted by SAR satellites pass through the atmosphere, the propagation speed and propagation path of electromagnetic waves will be affected due to the non-uniformity of the propagation medium in the atmosphere, thereby generating additional atmospheric signals. Since the state of the atmosphere is different at different times, the SAR satellite is affected differently by the atmosphere at different imaging times. The difference between the atmospheric signals at two imaging times is the so-called "atmospheric delay". The influence of atmospheric delay is one of the main error sources in the current InSAR surface deformation monitoring and is a technical problem that needs to be solved urgently.

[0003] In the prior art, the methods for removing the atmosphere from the radar differential interferogram include: (1) using a GPS machine to receive signals from multiple GPS satellites to calculate the tropospheric atmospheric delay phase and then subtract the delay phase from the differential interferogram; (2) using an optical sensor to obtain water vapor data, using an atmospheric delay model to calculate the atmospheric delay, and then subtracting the delay phase from the differential interferogram; (3) using the interferogram on the time series to weaken the influence of the atmosphere by filtering or adjustment in the time domain. However, the spatial resolution of GPS data and raw meteorological data is much lower than that of SAR images, optical sensors only work during the day, and the spatial domain tropospheric function modeling method can only simulate the atmospheric delay of the troposphere. The time series filtering method is often based on the assumption that the tropospheric delay in the time domain conforms to the Gaussian distribution, which is often inconsistent with the actual situation.

[0004] When processing multiple InSAR images over a large area, the deformation splicing results often have obvious block phenomena due to the large differences in the positions of the reference points when processing differential interferometric images of different amplitudes. The current methods are mostly to extract the probability density curve of the deformation results of two adjacent images at the overlap, and use the results of one of the images as the benchmark to correct the results of the other image. However, this approach increases uncertainty. When the deformation monitoring area is large and there are many images, when the results of one image are used as the benchmark to correct other images in sequence, the error is transmitted and accumulated, which increases the error of the deformation results far away from the reference image and increases the uncertainty. The radar differential interferometric results obtain the deformation relative to the reference point. The pixels close to the reference point are similarly affected by the phase residuals such as the atmosphere and terrain because they are close to the reference point. The phase residuals at the reference point can be subtracted for correction to reduce the influence of the interference term and reduce the final deformation error; as the distance from the reference point increases, the residual influence is different, and the error gradually increases. In large-scale differential interferometry processing, especially when multiple differential interferometry results of the same track need to be spliced, the deformation results at the same position in adjacent images are often inconsistent. This is because the reference points of the deformation results of different images are inconsistent, resulting in large differences in the deformation results at the same position in different images. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a reference surface-based radar differential interferometry image pair error correction method and system, which can reduce the influence of the atmosphere and facilitate the stitching processing of large-scale wide-band interference results.

[0006] Technical solution: The error correction method of the radar differential interferometer image pair based on the reference surface of the present invention comprises the following steps:

[0007] The differential radar interferometry method is used to obtain the interference image time series and coherence coefficient time series diagram of the target area;

[0008] The target area is divided into grids, and pixels whose coherence coefficient, sky openness, and terrain position index are all higher than their corresponding thresholds are selected as reference points in each grid;

[0009] Interpolate the reference points to obtain the reference surface;

[0010] The reference surface is removed from the interference image pair to obtain a corrected interference image pair time series.

[0011] Further, the step of selecting pixels whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds in each grid as reference points includes:

[0012] (2.1) Filter pixels in the target area whose coherence coefficients in the time series are all greater than the first threshold as high coherence points to obtain a set of high coherence points;

[0013] (2.2) Selecting a mountain top point as a reference point from the set of highly coherent points; the mountain top point is a pixel with a large sky openness and a terrain position index greater than its corresponding threshold;

[0014] (2.3) If the number of mountain top points in the high coherence point set is 0, then high coherence points are selected as reference points in the high coherence point set according to the coherence coefficient from high to low;

[0015] (2.4) If the reference point cannot be obtained according to steps (2.1)-(2.3), reduce the size of the first threshold and repeat steps (2.1)-(2.3) until the reference point is obtained or the first threshold is reduced to no less than the second threshold.

[0016] Furthermore, the first threshold of the coherence coefficient is 0.98, and the second threshold of the coherence coefficient is 0.9.

[0017] Furthermore, the mountain top point is a pixel with a sky openness greater than 0.98 and a terrain position index greater than 10.

[0018] Further, interpolating the reference points to obtain the reference surface includes:

[0019] ;

[0020] in, RF idw is the reference surface, p is the power of the distance weight, p The larger the value, the greater the influence of distance on the weight. z j It is j The value of the reference point, d j is the point to be interpolated and the j The distance from the reference point, m is the number of reference points involved in the interpolation.

[0021] Furthermore, the sky openness calculation method is:

[0022] ;

[0023] in, SVF For the openness of the sky, β i For the i The angle between a pixel and the sight lines of all its surrounding pixels, n The number of angles to consider in the calculation.

[0024] Furthermore, the terrain position index calculation method is:

[0025] ;

[0026] in, z is the elevation of a pixel, It is the average elevation of all pixels within a radius of several pixels around the pixel.

[0027] The reference surface-based radar differential interferometric image pair error correction system of the present invention comprises:

[0028] An interference image pair acquisition unit, used for acquiring the interference image pair time series and interference coefficient time series diagram of the target area by using a differential radar interference method;

[0029] A reference point calculation unit is used to divide the target area into grids, and select a number of pixels in each grid whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds as reference points;

[0030] A reference surface calculation unit, used for interpolating reference points to obtain a reference surface;

[0031] The interference image pair correction unit is used to remove the reference surface in the interference image pair to obtain a corrected interference image pair time series.

[0032] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the error correction method of the radar differential interferometer image pair based on the reference surface is implemented.

[0033] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the error correction method of the reference surface-based radar differential interferometer image pair is implemented.

[0034] Beneficial effects: Compared with the prior art, the advantages of the present invention are: (1) The present invention selects reference points through the original image and obtains the reference surface with the same spatial and temporal resolution as the original data by interpolation. Compared with the conventional correction method of atmospheric delay estimated by meteorological data, the atmospheric delay phase residue is less after correction by the reference surface with the same spatial and temporal resolution as the original data. At the same time, the atmospheric delay error is obtained by spatial interpolation of the residual value at the stable reference point of the interference pair result. The error obtained should be mainly the atmospheric delay phase, including errors of other noise phases. Some difficult to remove, such as the snow depth phase in winter, can also be removed together; (2) The reference points selected by the present invention are evenly distributed, and the reference points include the overlapping area between the two sub-frames, thereby eliminating the sub-frame error caused by the different reference point positions between different sub-frames to a certain extent. Since each image is corrected by selecting points separately, the cumulative error caused by conventional image-by-image correction will not be generated, and finally a wide range of error correction of radar differential interferometer images is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the working principle of InSAR in an embodiment of the present invention.

[0036] Figure 2 FIG. 4 is a flow chart of the error correction method of the present invention.

[0037] Figure 3 The following is a flow chart of reference point selection according to an embodiment of the present invention.

[0038] Figure 4 Graphs showing original interference image pairs at four times and correction results using two methods according to an embodiment of the present invention.

[0039] Figure 5 1 is a stitching result diagram of the original and corrected interference image pairs according to an embodiment of the present invention.

[0040] Figure 6 4 is a probability density curve diagram of the overlapping parts of the original and corrected interference image pairs according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0042] Synthetic Aperture Radar (SAR) is a radar imaging technology that uses SAR sensors that can be mounted on a mobile platform, such as an airplane, satellite, or aeroplane, to "expand" the antenna aperture by "synthesizing" radar images acquired along the flight path, thereby obtaining a spatial resolution that exceeds that of traditional scanning radar systems for regional or global monitoring. Interferometric Synthetic Aperture Radar (InSAR) aims to obtain a digital elevation model by calculating the phase difference between SAR images during two passes. It has quickly become one of the commonly used geodetic surveying technologies due to its high accuracy, high resolution, and all-weather capabilities. The subsequent differential radar interferometry technology (Differential Interferometric Synthetic Aperture Radar, D-InSAR) realizes surface deformation monitoring by introducing external DEM or three-track / four-track differential. This method is to perform "subtraction calculation" on the phase and amplitude signals contained in the same pixel collected from two SAR images with the same or slightly different positions observed at different times. By applying the SAR phase signal contained in each pixel, high-precision measurement of a large area can be achieved. The interference result between two SAR images (master and slave SAR images) is called an interferogram. Figure 1 Shown is a schematic diagram of the working principle of InSAR.

[0043] S 1 and S 2 are the locations of the satellite observing the same area of ​​the surface twice, B To correspond to the baseline, is a vertical baseline, and the interference phase at this time is the phase difference between the two observations:

[0044] ;

[0045] In the formula, λ is the wavelength of the radar signal, is the differential phase, and are the phases obtained during the first and second satellite passes, is the distance from the Earth's surface to the satellite obtained when the satellite first passes by. is the difference in distance between the Earth's surface and the satellite between the two passes.

[0046] The interference phase contains multiple components:

[0047] ;

[0048] In the formula, is the flat ground phase, is the terrain phase, is the surface deformation phase, is the atmospheric delay phase, is the orbital offset phase, The radar interferometry technology that removes the components other than the surface deformation phase and retains only the surface deformation phase to obtain the surface deformation information is called D-InSAR technology.

[0049] like Figure 2 As shown, the error correction method of the radar differential interferometer image pair described in the present invention is mainly divided into three steps. The first step is the selection of reference points, the second step is the reference point interpolation reference surface, and the last step is the removal of the reference surface.

[0050] Step 1: Select reference points.

[0051] Step 1-1: Divide the large-scale image into sub-grids.

[0052] Based on the reference point correction, the correction result of the area close to the reference point is better, and the error of the area far away from the reference point is larger. It is beneficial to improve the correction effect if the reference points cover the entire image as evenly as possible. In order to ensure that the reference points cover the entire image range as evenly as possible and achieve the best correction effect, the interference image pair coverage range is divided into a × b sub-grids and select 1 reference point in each grid partition.

[0053] Step 1-2: Reference point selection rules.

[0054] The original SAR data is subjected to differential interference to obtain the original interference image pair, and the coherence coefficient map of the target area is obtained at the same time. The coherence coefficient of each pixel of the interference image pair in all the interference image pairs is counted, and the reference point is determined in combination with the position of the mountain top point.

[0055] The selection of reference points in the sub-grid follows the principle of combining the highest coherence with the mountain top point. The combination of the two can effectively select stable points on the surface as reference points for InSAR deformation, such as Figure 3 As shown, the reference point selection rule in each sub-grid includes the following steps.

[0056] (1) Before the number of reference points reaches 1, start selecting from those with a coherence coefficient greater than 0.98;

[0057] (2) Pixels with a coherence coefficient greater than 0.98 at each time point in the time series are selected as candidate points. If the number of candidate point pixels is greater than 1, the mountain top point is selected as the reference point. If the number of mountain top points is greater than 1, the candidate points are sorted according to the size of the coherence coefficient, and the mountain top point with the highest coherence coefficient is selected as the reference point. If the number of mountain top points is 0, the candidate points are sorted according to the size of the coherence coefficient, and the point with the highest coherence is selected as the reference point.

[0058] (3) If the number of pixels with a coherence coefficient greater than 0.98 at each time point in the time series is zero, then the coherence threshold is reduced by 0.001, and then gradually reduced by 0.001 until a reference point is selected or the coherence threshold is as low as 0.9.

[0059] The mountain top point is determined jointly by the sky view factor (SVF) and the topographic position index (TPI). The pixels with a sky view factor greater than 0.98 and a topographic position index greater than 10 are set as the mountain top point.

[0060] ;

[0061] in, β is the angle between each pixel and all surrounding pixels. n is the number of angles considered in the calculation (such as 16, 32, etc.).

[0062] ;

[0063] in z is the elevation of a pixel, It is the average elevation of all pixels within a radius of several pixels around the pixel.

[0064] Step 2: Interpolate the reference surface from the reference points.

[0065] The Inverse Distance Weighted (IDW) interpolation method is used to interpolate the differential phase or deformation (change in surface elevation, the amount of surface uplift or subsidence within a certain period of time) of the reference point to obtain the reference surface.

[0066] ;

[0067] in, z j It is j The value of the reference point, d j is the point to be interpolated and the j The distance from the reference point,m is the number of reference points involved in the interpolation, p is the power of the distance weight, p The larger the value, the greater the influence of distance on the weight.

[0068] Step 3: Remove the reference surface.

[0069] ;

[0070] in, IP c is the result after reference surface correction, IP is the original pair, RF idw It is the reference surface.

[0071] In this embodiment, the C-band Sentinel-1 SAR differential interferometric image pairs of four time periods are respectively corrected for atmospheric effects using GACOS (a method that uses the iterative tropospheric decomposition (ITD) model, combined with global meteorological data and a high-resolution meteorological model to generate a high-spatial-resolution zenith total delay product) and the results are compared with those obtained by the method of the present invention.

[0072] This experiment operates on multiple interference patterns, so a batch processing is required to perform a unified preprocessing on all interference patterns to obtain the maximum range of the same scene image with different framing rules, as well as the coherence coefficient values ​​of all images for reference point selection.

[0073] Calculate the SVF and TPI of the target area, set the thresholds of 0.98 and 10 respectively to extract the peak points of the area, and calculate the TPI. Take the average elevation of all pixels within a radius of 5 pixels around the pixel. Divide the maximum range covered by all interferometric image pairs (about 400km×320km) into 25×20 subgrids (each subgrid is about 16km×16km), and select reference points in each subgrid. Use the IDW (inverse distance weighted interpolation) interpolation method to interpolate the phase / deformation results at the reference point to obtain the reference surface, and then remove the reference surface from the interferometric image pair.

[0074] like Figure 4 As shown, the original differential interferograms of four time periods, the results after GACOS atmospheric phase correction, and the results after correction by the method of the present invention are shown. Figure 4 Middle (a) ~ Figure 4The time periods corresponding to (d) are 20161221_20170102, 20180214_20180226, 20190104_20190116, and 20200204_20200216. It can be clearly seen that the conventional GACOS atmospheric correction data can remove the influence of the atmospheric delay phase to a certain extent, but there are many residual phases and the image quality cannot meet the requirements. After the reference surface correction of the present invention, the atmospheric delay phase is removed more completely, with only a small amount of residual phase.

[0075] To verify the effect of error correction between two images of different frames, two C-band Sentinel-1 SAR differential interferometric image pairs, Path3, Frame420 and Path3, Frame425, which are adjacent to each other in the north and south and have a time period of 20170116_20170128, were selected for the experiment. Generally, if the interferometric image pair is not corrected and the reference point position is not considered, there will be a large error between the two frames, such as Figure 5 As shown in (a), after the reference surface correction, the error of the stitching gap between the two scenes is reduced, as shown in Figure 5 As shown in (b). The probability density distribution of all pixel values ​​at the overlapping positions between the original result and the corrected result is calculated. Figure 6 As can be seen from (a) and (b), the difference in the overlapping part is reduced from the original 5mm to less than 1mm.

[0076] The reference surface-based radar differential interferometric image pair error correction system of the present invention comprises:

[0077] An interference image pair acquisition unit, used to acquire an interference image pair and an interference coefficient map of a target area using a differential radar interference method;

[0078] A reference point calculation unit is used to divide the target area into grids, and select pixels in each grid whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds as reference points;

[0079] A reference surface calculation unit, used for interpolating reference points to obtain a reference surface;

[0080] The interference image pair correction unit is used to remove the reference surface in the interference image pair to obtain a corrected interference image pair.

[0081] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the error correction method of the radar differential interferometer image pair based on the reference surface is implemented.

[0082] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the error correction method of the reference surface-based radar differential interferometer image pair is implemented.

[0083] The computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store the desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0084] The processor is used to execute the computer program stored in the memory to implement each step of the method involved in the above embodiment.

[0085] In summary, the present invention is based on the data of the coherence coefficient of the radar interferometer image itself, combined with external high-resolution DEM data to calculate the position of the mountain top point, and by dividing the grid area, a high-density reliable reference point can be obtained in a controllable manner. The reference surface containing the atmospheric delay phase and other error phases with the same spatiotemporal resolution as the original interferometer image is obtained by interpolation. On the one hand, the uniformity of spatiotemporal resolution is guaranteed, and on the other hand, while removing the atmospheric delay phase, other errors can be effectively suppressed, such as the snow depth phase caused by snow accumulation in high-latitude areas. At the same time, the reference points of the overlapping parts between the two images participate in the interpolation reference surface and the subsequent correction, which can reduce the error caused by the reference point problem between the two images to a certain extent.

Claims

1. A method for error correction of radar differential interferometric image pairs based on a reference surface, characterized in that: The steps include: The differential radar interferometry method is used to obtain the interference image time series and interference coefficient time series diagram of the target area; The target area is divided into grids, and pixels whose coherence coefficient, sky openness, and terrain position index are all higher than their corresponding thresholds are selected as reference points in each grid; Interpolate the reference points to obtain the reference surface; Removing a reference surface from the interference image pair to obtain a corrected interference image pair time series; The step of selecting pixels whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds in each grid as reference points comprises: (1.1) Filtering pixels in the target area whose coherence coefficients in the time series are all greater than the first threshold as high coherence points to obtain a set of high coherence points; (1.2) Selecting a mountain top point as a reference point from the set of highly coherent points; the mountain top point is a pixel with a large sky openness and a terrain position index greater than its corresponding threshold value; (1.3) If the number of mountain top points in the high coherence point set is 0, then high coherence points are selected as reference points in the high coherence point set according to the coherence coefficient from high to low; (1.4) If the reference point cannot be obtained according to steps (1.1)-(1.3), reduce the size of the first threshold and repeat steps (1.1)-(1.3) until the reference point is obtained or the first threshold is reduced to no less than the second threshold.

2. The error correction method of radar differential interferometry image pair based on reference surface according to claim 1, characterized in that: The first threshold value of the coherence coefficient is 0.98, and the second threshold value of the coherence coefficient is 0.

9.

3. The error correction method of radar differential interferometry image pair based on reference surface according to claim 1, characterized in that: The mountain top point is a pixel with a sky openness greater than 0.98 and a terrain position index greater than 10.

4. The error correction method of radar differential interferometry image pair based on reference surface according to claim 1, characterized in that: The interpolating the reference points to obtain the reference surface comprises: Among them, RF idw is the reference surface, p is the power of the distance weight, the larger the p value, the greater the influence of distance on the weight, z j is the value of the jth reference point, d j is the distance between the point to be interpolated and the jth reference point, and m is the number of reference points involved in the interpolation.

5. The error correction method of radar differential interferometry image pair based on reference surface according to claim 1, characterized in that: The sky openness calculation method is: Among them, SVF is the sky openness, β i is the angle between the i-th pixel and all the surrounding pixels, and n is the number of angles considered in the calculation.

6. The error correction method of radar differential interferometry image pair based on reference surface according to claim 1, characterized in that: The terrain position index calculation method is: Among them, z is the elevation value of a pixel, It is the average elevation of all pixels within a radius of several pixels around the pixel.

7. A radar differential interferometry image pair error correction system based on a reference surface, characterized in that: include: An interference image pair acquisition unit, which uses differential radar interference technology to acquire the interference image pair time series and interference coefficient time series diagram of the target area; A reference point calculation unit is used to divide the target area into grids, and select a number of pixels in each grid whose coherence coefficient, sky openness and terrain position index are all higher than their corresponding thresholds as reference points; A reference surface calculation unit, used for interpolating reference points to obtain a reference surface; An interference image pair correction unit, used for removing a reference surface from the interference image pair to obtain a corrected radar differential interference image pair time series; In the reference point calculation unit, the method for calculating the reference point is: (1.1) selecting pixels in the target area whose coherence coefficients in the time series are all greater than a first threshold as high coherence points to obtain a high coherence point set; (1.2) Selecting a mountain top point as a reference point from the set of highly coherent points; the mountain top point is a pixel with a large sky openness and a terrain position index greater than its corresponding threshold value; (1.3) If the number of mountain top points in the high coherence point set is 0, then high coherence points are selected as reference points in the high coherence point set according to the coherence coefficient from high to low; (1.4) If the reference point cannot be obtained according to steps (1.1)-(1.3), reduce the size of the first threshold and repeat steps (1.1)-(1.3) until the reference point is obtained or the first threshold is reduced to no less than the second threshold.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into a processor, the error correction method for a radar differential interferometer image pair based on a reference surface according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the error correction method for a radar differential interferometer image pair based on a reference surface is implemented according to any one of claims 1 to 6.