A registration method for UAV-borne long-baseline InSAR images combined with multiple short-baseline InSAR images

Abstract

This paper provides a method for unmanned aerial vehicle (UAV) long-baseline InSAR image registration using multiple short-baseline InSAR images. Baselines are designed according to baseline design criteria, and corresponding short-baseline SAR images are obtained. The offsets of adjacent short-baseline SAR images are estimated and accumulated to obtain the offset of the long-baseline SAR image. Furthermore, based on a gradient-based offset estimation method, both pixel-level and subpixel-level offset registration are performed simultaneously, avoiding the repetitive operations of pixel-level and subpixel-level offset estimation and improving the real-time performance of the registration.

Description

Technical Field

[0001] The present invention relates to the technical field of SAR image registration, and in particular to a multi-short-baseline InSAR combined unmanned aerial vehicle (UAV) long-baseline InSAR image registration method. Background Art

[0002] When using InSAR technology onboard small unmanned aerial vehicles (UAVs) to invert ground feature elevation, the length of the vertical baseline determines the sensitivity of the interferometric phase to elevation. The longer the baseline, the higher the inversion accuracy. However, long baselines can cause problems such as decoherence between UAV-based SAR image pairs, leading to interferometric measurement failure.

[0003] In order to achieve accurate registration of long-baseline SAR images carried by small unmanned aerial vehicles, the baselines are designed according to the baseline design criteria, and the corresponding short-baseline SAR images are obtained respectively. The offset estimation of the long-baseline SAR image is achieved by estimating and accumulating the offsets of adjacent short-baseline SAR images.

[0004] However, in existing short-baseline SAR image registration methods, the registration of a single control point consists of pixel-level and sub-pixel-level registration, both of which use correlation coefficients as criteria. The offset is estimated by sliding the main image window pixel by pixel and sub-pixel by sub-pixel over the auxiliary image window. Once the offsets of all control points are estimated, the offset of the entire SAR image is fitted using the least squares method, completing the resampling and precise registration of the auxiliary image. In this process, the offset estimation of adjacent short-baseline SAR images consists of the offset estimation of all control points, and the long-baseline SAR image registration step consists of the registration of multiple adjacent short-baseline SAR images, which is very time-consuming. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a UAV-mounted long-baseline InSAR image registration method combined with multiple short-baseline InSAR images to solve the technical problem that in the process of long-baseline SAR image registration in the existing technology, the registration of short-baseline SAR images needs to be repeated multiple times; and each short-baseline SAR image registration involves pixel-level registration and sub-pixel-level registration, which is very time-consuming.

[0006] The present invention provides a method for unmanned aerial vehicle (UAV)-borne long-baseline InSAR image registration combined with multiple short-baseline InSAR images, comprising:

[0007] S1. Set N small UAV flight paths, where the X-axis and Y-axis coordinates of the starting and ending points of each small UAV remain unchanged, and the Z-axis value increases step by step along the vertical direction. Obtain N tracks of echo data through the airborne SAR of the small UAV according to the established path, and use the BP algorithm to process the data to obtain N SAR images. , , 2, 3 ;

[0008] S2, SAR images acquired by a small UAV with adjacent short baselines As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio as control points, set the matching window with the coordinates of the control points as the center, and A search window is set with the coordinates of the control point as the center, and the search window is larger than the matching window;

[0009] S3. Calculating the coherence coefficient of the matching window and the search window pixel by pixel, and using the coherence coefficient to derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point, and obtaining the coordinate offset of the control point in the matching window and the search window by using a gradient algorithm;

[0010] S4, repeat step S3 until the coordinate offset estimation of M control points is completed, and the main image is obtained by fitting using the least squares method according to the coordinate offset and auxiliary images The offset of

[0011] S5. SAR images of adjacent short baselines The offset is resampled and the resampled SAR image is The offset between them is directly accumulated until the offset between the first image and the Nth image in the control point SAR image is obtained, and the SAR image is fitted using the least squares method. The offset of SAR image is obtained by bilinear interpolation. Perform resampling.

[0012] Optionally, the pixel-by-pixel calculation of the coherence coefficients of the matching window and the search window includes:

[0013] The calculation method of the coherence coefficient is expressed as:

[0014] ;

[0015] in, Represent the distance coordinate and azimuth coordinate of the control point respectively, Represents control points The offset in range and azimuth, represents the complex conjugate, represents the coherence coefficient between the matching window and the search window.

[0016] Optionally, the step of using the coherence coefficient to respectively derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point includes:

[0017] Coherence coefficient respectively and The calculation method is expressed as:

[0018] ;

[0019] in, express right Derivative, and in the above derivation process, the coordinates of the control points are calculated by bilinear interpolation Value .

[0020] Optionally, the SAR images of the adjacent short baselines Resampling is performed by the offset, including:

[0021] SAR images of adjacent short baselines The offset is resampled, and its calculation method is expressed as:

[0022] ;

[0023] in, and Represent control points In N SAR images, The offset between the first image and the first image; and Represents control points In N SAR images, images and The offset between images; and Represent control points In N SAR images, images and The resampling value of the offset between images.

[0024] Optionally, the SAR images of the adjacent short baselines Resampling is performed by the offset, including:

[0025] SAR images of adjacent short baselines The offset is resampled, and its calculation method is expressed as:

[0026] ;

[0027] in, and Represent control points In N SAR images, The offset between the first image and the first image; and Represent control points In N SAR images, images and The offset between images; and Represent control points In N SAR images, images and The resampling value of the offset between images.

[0028] Optionally, the resampled SAR image The offsets between them are directly accumulated, including:

[0029] The offset between the resampled SAR images is directly accumulated, and the calculation formula is expressed as:

[0030] .

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] This patent designs baselines according to baseline design criteria, obtaining corresponding short-baseline SAR images. By estimating the offsets of adjacent short-baseline SAR images and accumulating them, the offset of the long-baseline SAR image is obtained. Furthermore, based on the offset estimation method using a gradient algorithm, both pixel-level and sub-pixel offset registration are performed simultaneously, avoiding the repetitive operations of pixel-level and sub-pixel offset estimation and improving the real-time performance of the registration. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0035] Figure 1 It is a schematic diagram of the process of the present invention;

[0036] Figure 2 This is a flow chart of the long baseline small UAV SAR image registration process of the present invention;

[0037] Figure 3 1 is an unregistered interference phase image and coherence coefficient image in one embodiment;

[0038] Figure 4 An interference phase image and a coherence coefficient image registered using a conventional method in one embodiment;

[0039] Figure 5 The interference phase image and coherence coefficient image registered by the method of the present invention in one embodiment. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The functional units with the same labels in the examples of the present invention have the same and similar structures and functions.

[0041] See also Figure 1 The present invention provides a method for unmanned aerial vehicle (UAV) long baseline InSAR image registration combined with multiple short baseline InSAR images, comprising:

[0042] S1. Set N small UAV flight paths, where the X-axis and Y-axis coordinates of the starting and ending points of each small UAV remain unchanged, and the Z-axis value increases step by step along the vertical direction. Obtain N tracks of echo data through the airborne SAR of the small UAV according to the established path, and use the BP algorithm to process the data to obtain N SAR images. , , 2, 3 ;

[0043] S2, SAR images acquired by a small UAV with adjacent short baselines As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio as control points, set the matching window with the coordinates of the control points as the center, and A search window is set with the coordinates of the control point as the center, and the search window is larger than the matching window;

[0044] S3. Calculating the coherence coefficient of the matching window and the search window pixel by pixel, and using the coherence coefficient to derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point, and obtaining the coordinate offset of the control point in the matching window and the search window by using a gradient algorithm;

[0045] S4, repeat step S3 until the coordinate offset estimation of M control points is completed, and the main image is obtained by fitting using the least squares method according to the coordinate offset and auxiliary images The offset of

[0046] S5. SAR images of adjacent short baselines The offset is resampled and the resampled SAR image is The offset between them is directly accumulated until the offset between the first image and the Nth image in the control point SAR image is obtained, and the SAR image is fitted using the least squares method. The offset of SAR image is obtained by bilinear interpolation. Perform resampling.

[0047] In this embodiment, the baseline design is first performed. Based on the multi-track InSAR measurement, the spatial offset of the same target between two different tracks determines the coherence between the two SAR images. When designing the data acquisition, the radar start sampling time is strictly controlled, that is, there is no azimuth offset between InSAR images. Therefore, there is only a range offset caused by the baseline. According to the SAR imaging geometry, the target The range offset in the two-track InSAR image is:

[0048] ;

[0049] By transforming the above formula, it can be expressed as:

[0050] ;

[0051] Assume that the minimum sampling time is , the beam width is , the radar viewing angle is , then the highest target height that the radar can illuminate in the imaging scene is , which can be expressed as:

[0052] ;

[0053] The range of target heights that can be illuminated in the imaging scene is .

[0054] In the distance direction, with the center of the imaging scene as the reference, the slant distance value can be obtained by the number of sampling points in the distance direction. , distance sampling rate OK, that is .

[0055] Assume that the offset in the distance direction ,Will , and Substitute the same parameters into the baseline , it can be re-expressed as:

[0056] ;

[0057] When designing the baseline, the longest baseline is less than or equal to .

[0058] See also Figure 2 S1. Set N small UAV flight paths, where the X-axis and Y-axis coordinates of the starting and ending points of each small UAV remain unchanged, and the Z-axis value increases step by step along the vertical direction. Obtain N tracks of echo data through the airborne SAR of the small UAV according to the established path, and use the BP algorithm to process the data to obtain N SAR images. , , 2, 3 .

[0059] Based on the designed baseline, a small drone-mounted SAR image is used to obtain N SAR images. , , 2, 3 .

[0060] S2, SAR images acquired by a small UAV with adjacent short baselines As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio as control points, set the matching window with the coordinates of the control points as the center, and A search window is set with the coordinates of the control point as the center, and the search window is larger than the matching window.

[0061] First, the image is preprocessed and the SAR image is obtained by a small UAV with adjacent short baselines. As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio (SNR) as control points, take the coordinates of the control points as the center, and set the matching window size to , and in the auxiliary image With the coordinates of the control point as the center, set the search window size to , where the search window must be large enough to find a matching window within it.

[0062] See also Figure 2 , S3, calculate the coherence coefficient of the matching window and the search window pixel by pixel, and use the coherence coefficient to derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point, and obtain the coordinate offset of the control point in the matching window and the search window through the gradient algorithm.

[0063] The estimation of the control point offset is based on the coefficient that can determine the offset, such as the wave function, spectral function, etc. This patent uses the coherence coefficient as an example to calculate the coherence coefficient of the matching window and the search window pixel by pixel, which can be expressed as:

[0064] ;

[0065] in, Represent the distance coordinate and azimuth coordinate of the control point respectively, Represents control points The offset in range and azimuth, represents the complex conjugate, Represents the coherence coefficient of the matching window and the search window. When the offset is correctly estimated When the main image and auxiliary images Correlation coefficient Reach the maximum.

[0066] The coherence coefficient is used to derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point, and the calculation method is expressed as follows:

[0067] ;

[0068] in, express right Derivative, and in the above derivation process, the coordinates of the control points are calculated by bilinear interpolation Value . And the control points can be obtained by BFGS gradient algorithm Coordinate offset in the match window and search window.

[0069] S4, repeat step S3 until the coordinate offset estimation of M control points is completed, and the main image is obtained by fitting using the least squares method according to the coordinate offset and auxiliary images The offset of .

[0070] See also Figure 2, S5, SAR images of adjacent short baselines The offset is resampled and the resampled SAR image is The offset between them is directly accumulated until the offset between the first image and the Nth image in the control point SAR image is obtained, and the SAR image is fitted using the least squares method. The offset of SAR image is obtained by bilinear interpolation. Perform resampling.

[0071] Since the coordinates of the control points will change with the change of the baseline, the offsets of adjacent baseline SAR images cannot be directly accumulated. In this case, the offsets need to be resampled, that is:

[0072] ;

[0073] in, and Represent control points In N SAR images, The offset between the first image and the first image; and Represent control points In N SAR images, images and The offset between images; and Represent control points In N SAR images, images and The resampling value of the offset between images.

[0074] The offsets between the resampled SAR images are directly accumulated, and the calculation formula is expressed as:

[0075] ;

[0076] When you get the control point When calculating the offset between the first and Nth images in a SAR image, the least squares method is used to fit the offset of the entire image, that is:

[0077] ;

[0078] in: is the fitting parameter, are the pixel coordinates in the main image, The unknown relative coordinate offsets are used to register the primary and secondary images. The values ​​of the various parameters are obtained by fitting the offsets of the control points. An offset is calculated for each pixel in the primary image, and its matching position is found in the secondary image, achieving coordinate transformation between the two images.

[0079] When the offset of the auxiliary image relative to the main image is obtained, the interpolation resampling process of the auxiliary image is completed using the bilinear interpolation method, that is:

[0080] .

[0081] In another embodiment, specifically:

[0082] S1. Set N small UAV flight paths, where the X-axis and Y-axis coordinates of the starting and ending points of each small UAV remain unchanged, and the Z-axis value increases step by step along the vertical direction. Obtain N tracks of echo data through the airborne SAR of the small UAV according to the established path, and use the BP algorithm to process the data to obtain N SAR images. , , 2, 3 ;

[0083] S2, SAR images acquired by a small UAV with adjacent short baselines As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio as control points, set the matching window with the coordinates of the control points as the center, and A search window is set with the coordinates of the control point as the center, and the search window is larger than the matching window;

[0084] S3. Calculate the matching index of the matching window and the search window, which is the coherence coefficient in this patent, and search for the optimal matching index pixel by pixel. To improve matching efficiency, use the matching index as the objective function and estimate the coordinate offset of the matching window in the search window through a gradient algorithm.

[0085] S4, repeat step S3 until the coordinate offset calculation of M control points is completed, and the auxiliary image is obtained by fitting using the least squares method according to the coordinate offset Relative to the main image The first offset ;

[0086] S5. Repeat steps S2-S4 and select the SAR image again in As the main image, select the SAR image in As auxiliary image, calculate auxiliary image Relative to the Second offset of the main image ;

[0087] S6, with the first offset As a benchmark, bilinear interpolation is used to achieve the second offset With the first offset Resampling, get the third offset after resampling and the first offset And the third offset is directly accumulated to get and The fourth offset , using the fourth offset combined with bilinear interpolation, complete and Interpolation resampling;

[0088] S7, loop steps S2-S5, complete the SAR image and Offset Calculate, using offset Combined with the bilinear interpolation method, the SAR image is obtained Perform resampling.

[0089] See also Figure 3-Figure 5 , respectively designed the baseline length to be 15 meters, and the unregistered image result diagram, the traditional method registration result diagram and the present invention registration result diagram can be seen. Figure 5 The interference coefficient map of the registration result of the present invention is the clearest, and pixel-level offset registration and sub-pixel-level offset registration are completed, thereby improving the real-time performance of the registration.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used only 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 terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0091] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is intended to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for unmanned aerial vehicle (UAV) long baseline InSAR image registration using multiple short baseline InSAR images, characterized in that: include: S1. Set N small UAV flight paths, where the X-axis and Y-axis coordinates of the starting and ending points of each small UAV remain unchanged, and the Z-axis value increases step by step along the vertical direction. Obtain N tracks of echo data through the airborne SAR of the small UAV according to the established path, and use the BP algorithm to process the data to obtain N SAR images. , , 2, 3 ; S2, SAR images acquired by a small UAV with adjacent short baselines As the main image, as a secondary image and from the main image Extract M scattering points with high signal-to-noise ratio as control points, set the matching window with the coordinates of the control points as the center, and A search window is set with the coordinates of the control point as the center, and the search window is larger than the matching window; S3. Calculate the coherence coefficient of the matching window and the search window pixel by pixel, and use the coherence coefficient to derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point, and obtain the coordinate offset of the control point in the matching window and the search window by a gradient algorithm, wherein the calculation method of the coherence coefficient is expressed as follows: ； in, Represent the distance coordinate and azimuth coordinate of the control point respectively, Represents control points The offset in range and azimuth, represents the complex conjugate, represents the coherence coefficient of the matching window and the search window; S4, repeat step S3 until the coordinate offset estimation of M control points is completed, and the main image is obtained by fitting using the least squares method according to the coordinate offset and auxiliary images The offset of S5. SAR images of adjacent short baselines The offset is resampled and the resampled SAR image is The offset between them is directly accumulated until the offset between the first image and the Nth image in the control point SAR image is obtained, and the SAR image is fitted using the least squares method. The offset of SAR image is obtained by bilinear interpolation. Perform resampling.

2. The method for UAV-mounted long baseline InSAR image registration based on multiple short baseline InSAR images as claimed in claim 1, wherein: The step of using the coherence coefficient to respectively derive the offset in the range direction and the offset in the azimuth direction to obtain the coordinates of the control point includes: Coherence coefficient respectively and The calculation method is expressed as: ； in, express right Derivative, and in the above derivation process, the coordinates of the control points are calculated by bilinear interpolation Value .

3. The method for UAV-mounted long baseline InSAR image registration based on multiple short baseline InSAR images as claimed in claim 2, wherein: The SAR images of adjacent short baselines Resampling is performed by the offset, including: SAR images of adjacent short baselines The offset is resampled, and its calculation method is expressed as: ； in, and Represent control points In N SAR images, The offset between the first image and the first image; and Represent control points In N SAR images, images and The offset between images; and Represent control points In N SAR images, images and The resampling value of the offset between images.

4. The method for UAV-borne long baseline InSAR image registration using multiple short baseline InSAR images as claimed in claim 3, wherein: The resampled SAR image The offsets between them are directly accumulated, including: The offset between the resampled SAR images is directly accumulated, and the calculation formula is expressed as: 。

Images