A multi-angle fusion high-precision positioning method based on dataset optimization

Abstract

The present invention discloses a multi-angle fusion high-precision positioning method based on data set optimization in the field of microwave remote sensing technology. The multi-angle fusion high-precision positioning method based on data set optimization includes the following steps: determining the conversion relationship between the pixel position and the geographical location in the multi-angle SAR image; establishing an equivalent equation between the multi-angle SAR images, and estimating the positioning error of the multi-angle SAR images in pairs; multi-angle positioning error consistency identification and image optimization; obtaining the estimation of the multi-angle SAR image positioning error by the average sum method; compensating the estimation result of the SAR image positioning error in the SAR image positioning, and completing the SAR image high-precision positioning. The multi-angle fusion high-precision positioning method based on data set optimization improves the positioning accuracy and the robustness of the positioning result.

Description

Technical Field

[0001] The present invention relates to the technical field of high-power microwave source devices, and in particular, to a multi-angle fusion high-precision positioning method based on dataset optimization. Background Art

[0002] Synthetic Aperture Radar (SAR) is a high-resolution imaging system. It can not only image the target area all-weather and with high resolution, but also achieve high-precision positioning of targets in the scene, playing an important role in the effective monitoring of key targets and areas. Airborne SAR is an important branch of the SAR system and has high military and industrial value due to the flexibility of the carrier. In addition, with the maturity of high-resolution airborne SAR technology, higher requirements for target positioning accuracy have been put forward. Therefore, researching high-precision positioning technology for airborne SAR has important significance and practical value. Effectively utilizing airborne synthetic aperture SAR images often requires precise positioning of each image pixel. Historically, the positioning of airborne SAR images either relies on using reliable reference points to determine the relative position of the image or requires accurate motion information of the aircraft and the characteristics of the SAR data acquisition system as inputs to determine the absolute position of the image. However, in many applications, due to the difficulty in obtaining the accurate geographical location of reliable reference points and the inaccuracy of the recorded aircraft motion information, the accuracy of traditional positioning methods is not high.

[0003] In order to achieve high-precision planar positioning of airborne SAR images in the case of inaccurate geographical location information of ground reference points and inaccurate aircraft trajectory data, some scholars have proposed a two-angle geometric positioning method for airborne SAR images. Starting from two airborne SAR images with different flight angles, on the basis of completing rough positioning of the four corner points in the airborne SAR image using the RD model, a system-level geometric positioning model of multi-view SAR images is jointly utilized. By estimating the two-dimensional geometric positioning errors of each SAR image in the range direction and azimuth direction, high-precision planar positioning of the airborne SAR image is achieved. For the specific method, please refer to the reference document "An Image Planar Position Method Base on Fusion of Dual-View Airborne SAR Data. Remote Sens. 15, 2023, 15, 2499". However, the accuracy of the positioning result of this algorithm mainly depends on the consistency of the multi-angle system positioning errors and the geometric relationship between the angles, and the algorithm has low accuracy and insufficient robustness. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-angle fusion high-precision positioning method based on data set optimization, to solve the technical problems of low accuracy and insufficient robustness of airborne SAR image algorithms in the prior art, to provide an effective multi-angle fusion positioning method, which can determine whether the positioning errors of the multi-angle SAR image system are relatively consistent, and to effectively fuse multiple estimated values ​​through numerical calculation methods to form an effective estimate of the system positioning, thereby completing high-precision SAR geometric correction.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] The present invention provides a multi-angle fusion high-precision positioning method based on data set optimization, comprising the following steps:

[0007] S1. Input N multi-angle SAR images and determine the conversion relationship between the pixel position and the geographical location in the multi-angle SAR images;

[0008] S2. Establish the equivalent equation between multi-angle SAR images, and estimate the positioning error by using two-by-two multi-angle SAR images;

[0009] S3. Multi-angle positioning error consistency identification and image optimization;

[0010] S4. Obtain an estimate of the positioning error of the multi-angle SAR image using the average sum method;

[0011] S5. Compensate the estimated result of the SAR image positioning error in the SAR image positioning to complete the SAR image high-precision positioning.

[0012] Furthermore, the S1 comprises the following steps:

[0013] S11. Input N multi-angle SAR images to determine the positioning auxiliary parameters of the initial airborne SAR image;

[0014] S12. Determine the conversion relationship between the pixel position and the geographic location in the multi-view SAR image through the positioning auxiliary parameters of the initial airborne SAR image.

[0015] Further, the S2 comprises the following steps:

[0016] S21. Use homologous points to establish equivalent equations between multi-view SAR images;

[0017] S22. All multi-angle SAR images are combined in pairs to obtain multiple estimated positioning error values.

[0018] Further, the S3 comprises the following steps:

[0019] S31. Use the dispersion coefficient to determine the degree of dispersion of the data in each feature dimension;

[0020] S32. Eliminate one by one the images with the largest difference from other - angle SAR images in all images until a set of images with a qualified degree of dispersion is obtained.

[0021] Further, the S31 includes the following steps:

[0022] S311. Take N multi - angle SAR images. By pairwise combination, obtain N(N - 1) / 2 estimation results;

[0023] S312. Calculate the dispersion coefficient V of each feature dimension according to the estimation results, and use the dispersion coefficient to judge the two - dimensional dispersion degree of the data in each feature dimension.

[0024] Further, the calculation formula of the dispersion coefficient V of each feature dimension in the S312 is as follows:

[0025]

[0026] Where V r is the dispersion degree of the set of estimation results in the range - direction positioning error, V a is the dispersion degree of the set of estimation results in the azimuth - direction positioning error, s r is the standard deviation of the set of estimation results in the range - direction positioning error, s a is the standard deviation of the set of estimation results in the azimuth - direction positioning error, u r is the mean value of the set of estimation results in the range - direction positioning error, u a is the mean value of the set of estimation results in the azimuth - direction positioning error.

[0027] Further, the S32 includes the following steps:

[0028] S321. Identify the consistency of the positioning error by comparing the dispersion coefficient with the threshold, and determine whether the dispersion coefficients are all less than the threshold;

[0029] S322. If the dispersion coefficient is greater than the threshold, eliminate one image from the N images one by one, obtain N sets of SAR images with different flight - over angles, calculate multiple estimated values of each set of images. For the set of images with the smallest dispersion coefficient, judge whether the dispersion coefficients of its each feature dimension are all less than the threshold;

[0030] S323. If until only three images are left after elimination, and the dispersion coefficient of any feature dimension of the set of estimated values is greater than the threshold, then the multi - angle positioning error does not have consistency, does not meet the algorithm usage conditions, and the system positioning error cannot be estimated.

[0031] Further, the S322 includes the following steps:

[0032] S3221. If the discrete coefficient of any dimension is greater than the threshold, the real positioning error between images is too large and an accurate estimation value cannot be obtained;

[0033] S3222. Eliminate one image from the N images one by one to obtain N groups of SAR images at different flying angles, each group of images containing N-1 images;

[0034] S3223. Calculate multiple estimated values ​​of each group of images and calculate the discrete coefficient of each feature dimension. For a group of images with the smallest discrete coefficient, determine whether the discrete coefficient of each feature dimension is less than a threshold.

[0035] S3224. If the discrete coefficients of each feature dimension are all less than the threshold, the images in this group are taken as a group of images with relatively consistent true positioning errors;

[0036] S3225. If there is a situation where the discrete coefficients of each feature dimension are all smaller than the threshold, continue to eliminate them until a set of SAR images with relatively consistent real system positioning errors at different flying angles is obtained.

[0037] Further, the S4 comprises the following steps:

[0038] S41. averaging and summing multiple estimated values ​​obtained by combining two or more multi-angle SAR images that meet the consistency condition;

[0039] S42. Obtain the positioning error estimate of each angle SAR image, and take the multi-angle SAR that meets the constraints as M images.

[0040]

[0041] in, and is the estimated value of the positioning error in the range and the estimated value of the positioning error in the azimuth obtained when the system positioning errors between images i and j are equal, r i * represents the initial solution of the distance positioning error of the i-th image, a i * Represents the initial solution of the azimuth positioning error of the i-th image.

[0042] By adopting the above technical solution, the present invention has the following advantages:

[0043] The present invention provides a multi - angle fusion high - precision positioning method based on dataset optimization. The coefficient of variation of the estimation results is used to identify the consistency of the multi - angle positioning errors, and a group of SAR images with the most consistent positioning errors is selected. At the same time, the average sum of the two - view system positioning error estimates between different images is obtained, and a more reliable positioning error estimation result is acquired. This multi - angle fusion high - precision positioning method based on dataset optimization can determine whether the multi - angle SAR image system positioning errors are relatively consistent, and effectively fuse multiple estimated values through numerical calculation methods to form an effective estimation of the system positioning, completing high - precision SAR geometric correction and improving the positioning accuracy and the robustness of the positioning results. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flowchart of the multi - angle fusion high - precision positioning method based on dataset optimization of the present invention;

[0045] Figure 2 is a flowchart of the consistency identification of angle positioning errors and image optimization of the present invention;

[0046] Figure 3 are SAR images of measured multi - view different flight - over angles;

[0047] Figure 4 is a comparison of the effects of multi - angle SAR images and optical mosaic images before and after correction using the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The technical solutions of the present invention will be specifically described below with reference to the accompanying drawings of the specification. It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non - exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device.

[0049] Figure 1 shows a flowchart of the multi - angle fusion high - precision positioning method based on dataset optimization of the present invention. A multi - angle fusion high - precision positioning method based on dataset optimization includes the following steps specifically as Figure 1 shown:

[0050] S1. Input N multi - angle SAR images and determine the conversion relationship between the pixel positions and geographical positions in the multi - view SAR images;

[0051] Among them, S1 includes the following specific steps:

[0052] S11. Input N multi-angle SAR images to determine the positioning auxiliary parameters of the initial airborne SAR image;

[0053] S12. Determine the conversion relationship between the pixel position and the geographic location in the multi-view SAR image through the positioning auxiliary parameters of the initial airborne SAR image.

[0054] S2. Establish the equivalent equation between multi-angle SAR images, and estimate the positioning error by using two-by-two multi-angle SAR images;

[0055] Among them, S2 includes the following specific steps:

[0056] S21. Use homologous points to establish equivalent equations between multi-view SAR images;

[0057] S22. Combine all multi-angle SAR images in pairs to obtain multiple estimated positioning error values. For specific methods, please refer to the paper "An Image Planar Position Method Base on Fusion of Dual-View Airborne SAR Data. Remote Sens. 15, 2023, 15, 2499).

[0058] Figure 2 A flow chart of the angle positioning error consistency identification and image optimization of the present invention is shown.

[0059] S3. Multi-angle positioning error consistency identification and image optimization;

[0060] S3 includes the following steps: Figure 2 As shown:

[0061] S31. Use the dispersion coefficient to determine the degree of dispersion of the data in each feature dimension;

[0062] Wherein, S31 includes the following specific steps:

[0063] S311. Take N multi-angle SAR images, and obtain N(N-1) / 2 estimation results by combining them two by two. Use the discrete coefficient to determine the two-dimensional discrete degree of the estimation result. The discrete coefficient also reflects the discrete degree of the real system positioning error of the multi-angle SAR image.

[0064] S312. Calculate the discrete coefficient V of each feature dimension based on the estimation results, and use the discrete coefficient to determine the two-dimensional discrete degree of the data in each feature dimension.

[0065] The calculation formula for the coefficient of variation V of each feature dimension in S312 is as follows. We calculate the coefficient of variation of the range-direction positioning error and the coefficient of variation of the azimuth-direction positioning error respectively:

[0066]

[0067] Among them, V r is the degree of dispersion of the range-direction positioning error of this group of estimation results, and V a is the degree of dispersion of the azimuth-direction positioning error of this group of estimation results. s r is the standard deviation of the range-direction positioning error of this group of estimation results, and s a is the standard deviation of the azimuth-direction positioning error of this group of estimation results. u r is the mean of the range-direction positioning error of this group of estimation results, and u a is the mean of the azimuth-direction positioning error of this group of estimation results.

[0068] S32. One by one, eliminate the image with the largest difference from other-angle SAR images in all images until a group of images with a dispersion degree meeting the conditions is obtained.

[0069] S32 includes the following steps:

[0070] S321. After obtaining the coefficient of variation of multi-angle SAR images, identify the consistency of the positioning error by comparing the coefficient of variation with the threshold, and determine whether the coefficient of variation is less than the threshold. It is known from experience that the threshold is generally selected as 1. Therefore, if the coefficient of variation of any dimension is greater than 1, the true positioning error between images is too large to obtain an accurate estimated value.

[0071] S322. If the coefficient of variation is greater than 1, eliminate one image from the N images one by one to obtain N groups of SAR images with different flight angles, calculate multiple estimated values of each group of images, and for the group of images with the smallest coefficient of variation, determine whether the coefficient of variation of each of its feature dimensions is less than 1;

[0072] S322 includes the following specific steps:

[0073] S3221. If the coefficient of variation of any dimension is greater than the threshold, the true positioning error between images is too large to obtain an accurate estimated value;

[0074] S3222. Eliminate one image from the N images one by one to obtain N groups of SAR images with different flight angles, and each group of images contains N - 1 images;

[0075] S3223. Calculate multiple estimated values of each group of images and calculate the coefficient of variation of each of its feature dimensions. For the group of images with the smallest coefficient of variation, determine whether the coefficient of variation of each of its feature dimensions is less than the threshold;

[0076] S3224. If the discrete coefficients of each feature dimension are all less than the threshold, the images in this group are taken as a group of images with relatively consistent true positioning errors;

[0077] S3225. If there is a situation where the discrete coefficients of each feature dimension are all smaller than the threshold, continue to eliminate them until a set of SAR images with relatively consistent real system positioning errors at different flying angles is obtained.

[0078] S323. If the discrete coefficient of any characteristic dimension of the group of estimated values ​​is greater than 1 until only three images are left, the multi-angle positioning error is not consistent, the algorithm usage conditions are not met, and the system positioning error cannot be estimated.

[0079] S4. Obtain an estimate of the positioning error of the multi-angle SAR image using the average sum method;

[0080] Among them, S4 includes the following specific steps:

[0081] S41. averaging and summing multiple estimated values ​​obtained by combining two or more multi-angle SAR images that meet the consistency condition;

[0082] S42. Obtain the positioning error estimate of each angle SAR image, and take the multi-angle SAR that meets the constraints as M images.

[0083]

[0084] in, and is the estimated value of the positioning error in the range and the estimated value of the positioning error in the azimuth obtained when the system positioning errors between images i and j are equal, r i * represents the initial solution of the distance positioning error of the i-th image, a i * Represents the initial solution of the azimuth positioning error of the i-th image.

[0085] Through the above calculation, we can obtain the system positioning error of a group of images with relatively consistent real positioning errors. Using the geographical location equivalence relationship between two images, we can obtain multiple estimated values ​​of the SAR images with removed angles, and use the estimated mean as the initial solution of the SAR images with removed angles. At this point, we have obtained a rough estimate of the positioning error of the multi-angle SAR system.

[0086] S5. Compensate the estimated result of the SAR image positioning error in the SAR image positioning to complete the SAR image high-precision positioning.

[0087] Figure 3Shows the measured SAR images with different flight angles at multiple perspectives. In a specific embodiment, the measured data used in the present invention is specifically as Figure 3 shown. This set of multi-angle SAR images comes from a certain airborne SAR platform. Among them, Figure 3 (1) is a SAR image with a track angle of 148.97°, Figure 3 (2) is a SAR image with a track angle of 160.44°, Figure 3 (3) is a SAR image with a track angle of 173.22°, Figure 3 (4) is a SAR image with a track angle of -173.20°.

[0088] Figure 4 Shows the effect comparison of multi-angle SAR images and optical mosaicked images before and after correction using the method of the present invention.

[0089] Figure 4 (1) is the mosaic of the locally enlarged image in Figure 3 (1) and the high-precision optical image without correction using the method of the present invention; Figure 4 (2) is the effect comparison of the mosaic of the locally enlarged image in Figure 3 (1) and the high-precision optical image after correction using the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention. It can be seen from Figure 4 (1), Figure 4 (2) that without correction using the method of the present invention, compared with the mosaic of the high-precision optical image, the quality is poor, there are obvious breaks between the images, and at this time, the positioning accuracy is poor. After correction using the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention, the mosaic quality has been effectively improved, which shows that the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention effectively improves the positioning accuracy.

[0090] The comparison of the planar positioning errors of the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention and the two-view method for airborne SAR targets is specifically shown in Table 1 below:

[0091] Table 1: Comparison of the planar positioning errors of the method of the present invention and the two-view method for airborne SAR targets

[0092]

[0093]

[0094] As can be seen from Table 1 above, the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention has effectively improved the positioning accuracy of multi-angle images compared to the average positioning accuracy of the two-view method. In addition, it can also be seen from Table 1 that the estimation accuracy of the two-view method is not stable, and moreover, when the angle difference between images is small and the system positioning errors are inconsistent, it may even cause deterioration of the estimation accuracy. The multi-angle fusion high-precision positioning method based on dataset optimization of the present invention not only greatly improves the estimation accuracy, but also ensures the robustness of the estimation. This proves that the multi-angle fusion high-precision positioning method based on dataset optimization of the present invention is effective.

[0095] Finally, it should be noted that although the present invention has been described with reference to the current specific embodiments, those of ordinary skill in the art should recognize that the above embodiments are only used to illustrate the present invention and are not intended to limit the present invention. Various equivalent changes or substitutions can be made without departing from the spirit of the present invention. Therefore, as long as the changes and modifications of the above embodiments are within the scope of the spirit of the present invention, they will fall within the scope of the claims of the present invention.

Claims

1. A multi-angle fusion high-precision positioning method based on data set optimization, characterized in that: The following steps are involved: S1. Input N multi-angle SAR images and determine the conversion relationship between the pixel position and the geographical location in the multi-angle SAR images; S2. Establish the equivalent equation between multi-angle SAR images, and estimate the positioning error by using two-by-two multi-angle SAR images; S3. Multi-angle positioning error consistency identification and image optimization; The S3 comprises the following steps: S31. Use the dispersion coefficient to determine the degree of dispersion of the data in each feature dimension; S32. Eliminate the images with the largest difference from other angle SAR images one by one among all images until a set of images with a discrete degree meeting the conditions is obtained; S4. Obtain an estimate of the positioning error of the multi-angle SAR image using the average sum method; The S4 comprises the following steps: S41. averaging and summing multiple estimated values ​​obtained by combining two or more multi-angle SAR images that meet the consistency condition; S42. Obtain the positioning error estimate of each angle SAR image, and take the multi-angle SAR that meets the constraints as M images. in, and is the estimated value of the positioning error in the range and the estimated value of the positioning error in the azimuth obtained when the system positioning errors between images i and j are equal, r i * represents the initial solution of the distance positioning error of the i-th image, a i * represents the initial solution of the azimuth positioning error of the i-th image; S5. Compensate the estimated result of the SAR image positioning error in the SAR image positioning to complete the SAR image high-precision positioning.

2. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 1 is characterized in that: The S1 comprises the following steps: S11. Input N multi-angle SAR images to determine the positioning auxiliary parameters of the initial airborne SAR image; S12. Determine the conversion relationship between the pixel position and the geographic location in the multi-view SAR image through the positioning auxiliary parameters of the initial airborne SAR image.

3. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 1 is characterized in that: The S2 comprises the following steps: S21. Use homologous points to establish equivalent equations between multi-view SAR images; S22. All multi-angle SAR images are combined in pairs to obtain multiple estimated positioning error values.

4. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 1 is characterized in that: The S31 comprises the following steps: S311. Take N multi-angle SAR images and obtain N(N-1) / 2 estimation results by combining them two by two; S312. Calculate the discrete coefficient V of each feature dimension based on the estimation results, and use the discrete coefficient to determine the two-dimensional discrete degree of the data in each feature dimension.

5. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 4 is characterized in that: The calculation formula of the discrete coefficient V of each feature dimension in S312 is as follows: Among them, V r V is the discrete degree of the positioning error of the estimation result in the distance direction, a is the discrete degree of the positioning error of the estimation result in the azimuth direction, s r is the standard deviation of the positioning error of the estimation result in the range direction, s a is the standard deviation of the positioning error of the estimation result in azimuth, u r is the mean value of the positioning error of the estimation result in the range direction, u a is the mean value of the positioning error of the estimation result in the azimuth direction.

6. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 5 is characterized in that: The S32 comprises the following steps: S321. Identify the consistency of positioning error by comparing the discrete coefficient with the threshold, and determine whether the discrete coefficients are all less than the threshold; S322. If the dispersion coefficient is greater than the threshold, one image is removed from the N images one by one to obtain N groups of SAR images with different flying angles, and multiple estimated values ​​of each group of images are calculated. For the group of images with the smallest dispersion coefficient, it is determined whether the dispersion coefficients of each feature dimension are all less than the threshold; S323. If the discrete coefficient of any characteristic dimension of the estimated value is greater than the threshold until only three images are left, the multi-angle positioning error is not consistent, the algorithm usage conditions are not met, and the system positioning error cannot be estimated.

7. The multi-angle fusion high-precision positioning method based on data set optimization according to claim 6 is characterized in that: The S322 includes the following steps: S3221. If the discrete coefficient of any dimension is greater than the threshold, the real positioning error between images is too large and an accurate estimation value cannot be obtained; S3222. Eliminate one image from the N images one by one to obtain N groups of SAR images at different flying angles, each group of images containing N-1 images; S3223. Calculate multiple estimated values ​​of each group of images and calculate the discrete coefficient of each feature dimension. For a group of images with the smallest discrete coefficient, determine whether the discrete coefficient of each feature dimension is less than a threshold. S3224. If the discrete coefficients of each feature dimension are all less than the threshold, the images in this group are taken as a group of images with relatively consistent true positioning errors; S3225. If there is a situation where the discrete coefficients of each feature dimension are all smaller than the threshold, continue to eliminate them until a set of SAR images with relatively consistent real system positioning errors at different flying angles is obtained.

Images