A Tomographic SAR Three-Dimensional Imaging Method for Low-Altitude Platforms

Abstract

This invention discloses a tomographic SAR three-dimensional imaging method for low-altitude platforms, comprising the following steps: acquiring SAR images; performing sparse inversion on the SAR images within a lower viewpoint constraint range based on a cylindrical wave TomoSAR imaging model to determine the scattering coefficients of the scatterer signal along the elevation direction in the pixels; calculating the lower viewpoint parameters based on the scattering coefficients; and generating the position coordinates of the scatterer in the ground coordinate system based on the lower viewpoint parameters. This invention, by constructing a cylindrical wave TomoSAR imaging model for sparse inversion, can correct the geometric deformation of the scatterer, significantly improving the three-dimensional imaging accuracy of the scatterer. Simultaneously, by adaptively determining the elevation search range based on the elevation constraint, elevation deblurring is achieved, further enhancing imaging accuracy.

Description

Technical Field

[0001] This invention belongs to the field of tomographic SAR three-dimensional imaging technology, and particularly relates to a tomographic SAR three-dimensional imaging method for a low-altitude platform. Background Technology

[0002] TomoSAR imaging technology reconstructs the true SAR 3D scene by retrieving the scattered signals at different heights (perpendicular to the azimuth and range planes) of the same target object through multiple observations. TomoSAR imaging requires multiple baseline data, and its development was initially slow due to limitations such as a limited number of baselines and unstable orbits. However, with the increasing maturity of spaceborne and airborne SAR systems, high-quality multi-baseline SAR images have been successfully acquired, making it possible to reconstruct the height structure of ground features using TomoSAR technology, thus achieving 3D imaging.

[0003] TomoSAR imaging technology has significant advantages in reconstructing the three-dimensional structure of buildings due to its unique elevation imaging capabilities. TomoSAR can reconstruct the scattering profile within a single pixel along the elevation direction and extract the elevation position and intensity information of each scatterer, thereby reconstructing the overall three-dimensional structure of the building and obtaining a high-precision three-dimensional city model within the scene.

[0004] In recent years, much high-resolution SAR data has been acquired from low-altitude platforms such as low-flying aircraft or drones, with flight altitudes of only a few thousand or even a few hundred meters. Compared to spaceborne platforms, low-altitude platform SAR has a resolution advantage. However, the slant range (distance from the SAR antenna to the ground object) of SAR data from low-altitude platforms is also smaller, leading to a reduction in the maximum unambiguous altitude, sometimes approaching the height of buildings in the scene, resulting in elevation ambiguity that is difficult to filter out. Elevation ambiguity in TomoSAR has become a significant problem affecting 3D building imaging. Furthermore, the flight altitude of low-altitude platforms renders the plane wave assumption of the original TomoSAR imaging model inapplicable, causing certain geometric distortions in the 3D imaging results. Summary of the Invention

[0005] The purpose of this invention is to provide a tomographic SAR three-dimensional imaging method for low-altitude platforms to solve the problems of elevation ambiguity and geometric deformation in three-dimensional imaging results.

[0006] This invention adopts the following technical solution: a tomographic SAR three-dimensional imaging method for low-altitude platforms, comprising the following steps:

[0007] Acquire SAR images;

[0008] Based on the cylindrical wave TomoSAR imaging model, sparse inversion of SAR images is performed within the lower viewpoint constraint range to determine the scattering coefficient of the scatterer signal in the pixel along the elevation direction.

[0009] Calculate the viewing angle parameters based on the scattering coefficient;

[0010] The position coordinates of the scatterer in the ground coordinate system are generated based on the downward viewpoint parameters.

[0011] Furthermore, the cylindrical wave TomoSAR imaging model is as follows:

[0012]

[0013] Among them, g n Let θ′ represent the pixel value of the image formed by the nth antenna element, θ′ = θ - θ0, where θ is the downward viewing angle of the scatterer, θ0 is the downward viewing angle of the reference point (the ground point with the smallest slant range in the detection area), Δθ′ represents the constraint range, r is the slant range of the scatterer, γ(rθ′) represents the scattering coefficient of the signal along the elevation direction of the scatterer with slant range r and downward viewing angle parameter θ′, and ξ n Let ε be the elevation frequency of the nth antenna element. n The image noise is for the nth antenna element.

[0014] Furthermore, the method for generating the lower-view constraint range is as follows:

[0015] Determine the maximum height threshold h1 and minimum height threshold h2 of buildings in the detection area;

[0016] Calculate the upper and lower bounds of the adaptive elevation search range based on the maximum height threshold h1 and the minimum height threshold h2.

[0017] Generate the lower viewpoint constraint range based on the upper and lower bounds.

[0018] Furthermore, before determining the scattering coefficient of the scatterer signal in the pixel along the elevation direction, the following steps are also included:

[0019] Scattering object detection is performed on TomoSAR images.

[0020] Furthermore, a three-threshold method is used to detect scatterers in SAR images;

[0021] The three thresholds include the coherence coefficient threshold, the average amplitude threshold, and the amplitude deviation index threshold.

[0022] Furthermore, the target problem when performing sparse inversion on SAR images is:

[0023]

[0024] Where γ is the set of all scattering coefficients, g is the set of pixel values ​​of the image formed by N antenna elements, G is the observation matrix, and N is the total number of antenna elements. It is the standard deviation of the SAR image observation noise ε.

[0025] Furthermore, generating the position coordinates of the scatterer in the ground coordinate system based on the downward viewing angle parameters includes:

[0026] Y P =k·Xbin,

[0027] X P =r i ·sin(θ0+θ′ P )-X,

[0028] Z P =-r i ·cos(θ0+θ′ P )+H,

[0029] Among them, X P Y P and Z P θ′ represents the three-axis coordinates of the scatterer P in the ground coordinate system, k represents the column number of the scatterer P in the azimuth direction of the SAR image, Xbin is the azimuth resolution, and θ′ is the coordinates of the scatterer P. P Here, r represents the downward viewing angle parameter of the scatterer P, X is the X-axis coordinate of the reference point, and r is the downward viewing angle parameter of the scatterer P. i H is the slant range of the scatterer in the i-th column of the range direction in the SAR image, and H is the height of the SAR antenna phase center from the ground.

[0030] Furthermore, the OMP greedy algorithm, basis pursuit algorithm, or atomic norm minimization algorithm are used to solve the target problem.

[0031] Furthermore, the range of lower viewpoint constraints generated based on the upper and lower bounds includes:

[0032] Δθ′ i =[s′ i min / r i ,s′ i max / r i ],

[0033] Where, Δθ′ i Let s′ be the downward view constraint range in the i-th distance direction. i min Let s′ be the lower bound corresponding to the i-th distance direction. i max Let r be the upper bound corresponding to the i-th distance direction. i denoted as the slant range of the scatterer in the i-th column pixel of the SAR image range direction.

[0034] Another technical solution of the present invention is a tomographic SAR three-dimensional imaging device for a low-altitude platform, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the above-described method when executing the computer program.

[0035] The beneficial effects of this invention are: by constructing a cylindrical wave TomoSAR imaging model for sparse inversion, this invention can correct the geometric deformation of the scatterer and greatly improve the three-dimensional imaging accuracy of the scatterer. At the same time, by adaptively determining the elevation search range based on height constraints, elevation de-ambiguity can be achieved, further improving the imaging accuracy. Attached Figure Description

[0036] Figure 1 This is a geometric schematic diagram of TomoSAR three-dimensional imaging in an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of the coherence coefficient distribution of TomoSAR images before and after preprocessing in an embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram showing the calculation results of the boundary of the interval Δθ' in each pixel in an embodiment of the present invention;

[0039] Figure 4 This is a diagram showing the estimation results of the number of scatterers in each pixel in an embodiment of the present invention;

[0040] Figure 5 This is a schematic diagram of the inversion results of the scattering coefficient γ in an embodiment of the present invention;

[0041] Figure 6 The above are amplitude maps and Google Earth optical maps of the data in the embodiments of this invention;

[0042] Figure 7 This is a side view of the three-dimensional point cloud implementation result in an embodiment of the present invention;

[0043] Figure 8 This is a comparison diagram of the front view of the three-dimensional point cloud data obtained by the method of the present invention and the existing method. Detailed Implementation

[0044] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0045] The existing general method for realizing TomoSAR three-dimensional imaging is based on the plane wave scattering model. Its imaging steps can be summarized as follows: (1) Select an appropriate elevation search range to ensure that all buildings in the scene are covered within the three-dimensional imaging boundary; (2) Invert the elevation scattering signal through a super-resolution iterative algorithm, such as the Orthogonal Matching Pursuit combined with Bayesian Information Criterion (OMP-BIC); (3) Calculate the position of the scatterer in the elevation direction by detecting the peak value of the scattering signal; (4) Obtain the three-dimensional imaging result through coordinate transformation.

[0046] The plane wave scattering model assumes that the probe signal arriving at the imaging scene is a plane wave and performs elevation-direction scattering of the scatterer signal. OMP-BIC is an elevation-direction sparse inversion algorithm, consisting of two parts: OMP and BIC. The OMP algorithm, or Orthogonal Matching Pursuit algorithm, is a computationally efficient compressed sensing algorithm commonly used for sparse signal estimation. Compared to traditional spectral estimation algorithms, it can suppress ambiguity to some extent in terms of sidelobe suppression and super-resolution performance. However, the OMP algorithm requires sparsity as input, so the BIC algorithm is needed to estimate the sparsity. In TomoSAR, sparsity equals the number of scatterers, so the OMP-BIC algorithm is frequently used for elevation-direction sparse inversion in TomoSAR.

[0047] In TomoSAR imaging, while the OMP-BIC algorithm offers certain super-resolution advantages over traditional spectral estimation algorithms, it cannot distinguish the number of ambiguities in the TomoSAR elevation direction, still retaining elevation ambiguity. Furthermore, the BIC algorithm's accuracy in estimating the number of scatterers is insufficient, introducing noise, necessitating improvement. Moreover, in TomoSAR applications on low-altitude SAR platforms, the plane-wave-based scattering model cannot address geometric deformation issues, limiting its effectiveness in filtering elevation ambiguity in TomoSAR imaging from low-altitude platforms.

[0048] To address the elevation ambiguity and geometric deformation issues in TomoSAR imaging of low-altitude platforms, existing TomoSAR imaging methods are insufficient in elevation deambiguity, do not consider geometric deformation correction, and have low accuracy in estimating the number of scatterers.

[0049] This invention utilizes the imaging geometry characteristics of low-altitude SAR platforms, proposes a TomoSAR imaging model based on the cylindrical wave assumption to correct geometric deformation, uses height constraints as geometric constraints in the solution space to achieve elevation de-ambiguity, and proposes a refined scatterer pre-estimation method to improve the accuracy of scatterer number estimation.

[0050] In this embodiment of the invention, an airborne SAR system is taken as an example, such as... Figure 1The diagram illustrates its application scenario. In this diagram, an aircraft flies from one end of the detection area to the other, capturing images. The shaded area represents the ground region corresponding to the radar detection area. The direction of the aircraft's movement is the azimuth direction; the aircraft and its projection point on the ground form one line, and it forms another line with a point in the detection area; the angle between these two lines is the downward viewing angle; the direction the airborne SAR faces towards the point in the detection area is the range direction, i.e., the radar's line of sight, and the range direction is perpendicular to the azimuth direction; the aircraft's projection point on the ground and the point in the detection area form the ground-to-range direction; the range direction and the azimuth direction form a plane, and the direction perpendicular to this plane and pointing towards the sky is the elevation direction.

[0051] Specifically, this invention discloses a tomographic SAR three-dimensional imaging method for low-altitude platforms, comprising the following steps: acquiring SAR images; performing sparse inversion on the SAR images within the lower viewpoint constraint range based on a cylindrical wave TomoSAR imaging model to determine the scattering coefficients of the scatterer signals in the pixels along the elevation direction; calculating the lower viewpoint parameters based on the scattering coefficients; and generating the position coordinates of the scatterer in the ground coordinate system based on the lower viewpoint parameters.

[0052] This invention constructs a cylindrical wave-based TomoSAR imaging model for sparse inversion, which can correct the geometric deformation of the scatterer and greatly improve the three-dimensional imaging accuracy of the scatterer. At the same time, it adaptively determines the elevation search range based on height constraints to achieve elevation de-ambiguity and further improve the imaging accuracy.

[0053] In this embodiment of the invention, the aforementioned scattering body refers to buildings in the detection area, such as tall buildings, towers, etc., and may also include some artificial mountains, ground, etc.

[0054] In one embodiment, the SAR dataset (i.e., SAR images) is preprocessed before sparse inversion. The purpose of preprocessing is to enhance the coherence between the SAR master image and other images in the SAR image set, including image registration and amplitude / phase error correction. The SAR master image is a SAR image selected from the image set.

[0055] Specifically, the coherence coefficient of each pixel in the TomoSAR image before and after preprocessing is calculated, and the pixel number distribution of the coherence coefficient is obtained, such as... Figure 2 The figure shows the coherence coefficient distribution of TomoSAR images before and after preprocessing. It can be seen from the figure that the preprocessing operation significantly improves the overall coherence coefficient of the TomoSAR images, laying the foundation for subsequent 3D imaging.

[0056] For low-altitude platform SAR systems, the radar microwaves actually reaching the target are more akin to cylindrical waves, with the wavefront being a coaxial cylindrical surface. The echo signal can be considered a circular wave extending along the azimuth in the range-elevation plane. Therefore, when the distance between the SAR antenna phase center (APC) and the target decreases, the error caused by the plane wave model cannot be ignored and needs to be corrected. Thus, this invention proposes a general model for TomoSAR imaging based on the cylindrical wave assumption:

[0057]

[0058] Among them, g n Let S'(θ) be the pixel value of the image formed by the nth antenna element, and let S′(θ) represent the mapping from the lower viewpoint to the corrected elevation vector s′. This is the first derivative of S′(θ). Unlike tomographic SAR imaging based on the plane wave model, the scattering coefficient in the cylindrical wave model is integral along the lower view angle, not along the elevation direction s. The corrected elevation direction is a correction for the elevation direction, approximating but not coinciding with it.

[0059] Let the direction along the cylindrical wavefront be the corrected elevation direction, and the mapping relationship between the downward angle and the corrected elevation direction be s′=(θ-θ0)r=θ′ r Therefore, the above formula can be further expressed as:

[0060]

[0061] Since all scatterers within a pixel have the same slant range, the scattering profile is an arc centered on the main APC position. θ′=θ-θ0, where θ is the downward viewing angle of the scatterer, θ0 is the downward viewing angle of the reference point (the ground point with the smallest slant range in the detection area), Δθ′ represents the constraint range of θ′, r is the slant range of the scatterer, γ(rθ′) represents the scattering coefficient along the elevation direction of the signal from the scatterer with slant range r and downward viewing angle parameter θ′, and ξ n Let ε be the elevation frequency of the nth antenna element. n Let be the image noise of the nth antenna element, where j represents the imaginary part of the complex number.

[0062] Furthermore, one embodiment of the present invention employs an adaptive method for determining the elevation search range based on height constraints. To accurately invert the scattered signal, a suitable expression for the search range based on a modified elevation direction needs to be found. Assuming that the heights of urban buildings in the same scene are of the same order of magnitude, and that the elevation search range is typically determined by a suitable elevation direction constraint, the present invention proposes an adaptive method for determining the elevation search range based on height constraints.

[0063] Based on prior information about buildings in the imaging scene, the maximum height threshold and minimum height threshold h2 of buildings in the detection area are determined as height threshold pairs. Then, based on the maximum height threshold h1 and minimum height threshold h2, the upper and lower bounds of the adaptive elevation search range are calculated; finally, the lower viewpoint constraint range is generated based on the upper and lower bounds.

[0064] Here, h0 = 0 is taken as the ground height, and h1 can be the roof height of the tallest building. Since ground scattering information is also included in the SAR echo, a redundant value needs to be provided for the three-dimensional inversion of the ground scatterer. Therefore, h2 can be an empirical value below the ground height value h0.

[0065] Based on the geometric relationships of TomoSAR imaging, we can obtain:

[0066]

[0067]

[0068] Therefore, the boundaries (upper and lower bounds) of the adaptive elevation search range based on the height threshold constraint are:

[0069]

[0070]

[0071] From the above, we can see that Δs′ i =[s′ i min ,s′ i max ] is the elevation search range of all pixels inward from the i-th distance, thus the lower view constraint range can be obtained as Δθ′. i =[s′ i min / r i ,s′ i max / r i ]. Wherein, Δθ′ i Let s′ be the downward view constraint range in the i-th distance direction. i min Let s′ be the lower bound corresponding to the i-th distance direction. i max Let r be the upper bound corresponding to the i-th distance direction. i denoted as the slant range of the scatterer in the i-th column pixel of the SAR image range direction.

[0072] Specifically, such as Figure 3 The image shows a schematic diagram of the calculation results (in radians) for the boundary of the interval Δθ' at each pixel. Figure 3 (a) is a graph showing the calculation results of the upper boundary of the interval Δθ'. Figure 3 (b) is a graph showing the calculation results of the lower boundary of the interval Δθ'.

[0073] In one embodiment, before determining the scattering coefficient of the scatterer signal along the elevation direction in a pixel, the method further includes: scatterer detection of the TomoSAR image. In this embodiment, the three thresholds include a coherence coefficient threshold, an average amplitude threshold, and an amplitude deviation index threshold.

[0074] First, SAR image scatterer detection is performed. Scatterer detection often extracts scatterers by setting thresholds for decision variables. This invention employs a three-threshold scatterer detection method, setting a small threshold related to coherence coefficient and signal-to-noise ratio based on different datasets to remove severely decoherent scatterers. The three thresholds are coherence coefficient, average amplitude, and amplitude deviation index. The three-threshold detection steps are as follows: First, the pixels of the entire region are filtered by coherence coefficient and average amplitude detection. Then, the result of coherence coefficient detection is further filtered by amplitude deviation index detection. Finally, the result of amplitude mean detection is used to compensate for the pixels lost in the previous step. The result of the three-threshold detection can be expressed as:

[0075]

[0076] Among them, P, P γ , and These are the remaining pixels after detection using three thresholds, coherence coefficient, average amplitude, and amplitude deviation index, respectively. After this step, most pixels are discarded as non-scatterers.

[0077] like Figure 4 The image shown is an estimation result of the number of scatterers in each pixel according to an embodiment of the present invention. The image displays the number of scatterers in each pixel. Pixels with 0 scatterers are non-scatterer pixels. A pixel can have a maximum of 3 scatterers.

[0078] Next, scatterer pre-classification based on average amplitude is performed. Whether a pixel contains a two-scatterer or a three-scatterer is determined by a threshold set as a multiple of the image's average amplitude. SAR images rarely contain more than three scatterers, so scatterers with more than three scatterers are not considered.

[0079] In this embodiment of the invention, the TomoSAR model based on the cylindrical wave assumption (i.e., the target problem) can be solved by applying sparsity constraints (e.g., L0 norm regularization):

[0080]

[0081] Where γ is the set of all scattering coefficients, g is the set of pixel values ​​of the image formed by N antenna elements, G is the observation matrix, and N is the total number of antenna elements. It is the standard deviation of the SAR image observation noise ε.

[0082] Next, the target problem is solved using the OMP greedy algorithm, the basis pursuit algorithm, or the atomic norm minimization algorithm. Solving equation (9) using greedy algorithms such as OMP requires sparsity (i.e., the number of scatterers) as input. The final set γ is the set of all scattering coefficients, which is then converted into the corresponding lower viewpoint parameter θ′.

[0083] Taking the inversion result of the scattering coefficient γ of a certain pixel as an example, the result is as follows: Figure 5 As shown in the figure, it can be seen that the method proposed in this invention corrects the peak position of the inverted scattering coefficient under the constraint of the adaptive search range.

[0084] Finally, assume that the range-azimuth pixel (i, k) contains the scatterer P. The position of the scattering point P in the corrected elevation direction is s′. P The downward angle of the scattering point P is θ P Taking the perpendicular point of the APC on the ground as the origin of the ground coordinate system, the distance from the ground as the X-axis, and the azimuth as the Y-axis, the Y-coordinate of the scattering point P in the ground coordinate system can be obtained according to the SAR imaging geometry:

[0085] Y P =k·Xbin (8)

[0086] In addition, the following relationship can be derived:

[0087]

[0088]

[0089] Where, θ′ P =θ P -θ0=s′ P / r i The X and Z coordinates of the scatterer P in the ground coordinate system are as follows:

[0090] X P =r i ·sin(θ0+θ′ P )-X (11)

[0091] Z P =-r i ·cos(θ0+θ′ P )+H (12)

[0092] The positions of all scatterers in the ground coordinate system for each pixel are calculated one by one. This allows the construction of a 3D point cloud of a large urban area. In formulas (8)-(12), k represents the column number of scatterer P in the azimuth direction of the SAR image, Xbin is the azimuth resolution, and θ′... PHere, r represents the downward viewing angle parameter of the scatterer P, X is the X-axis coordinate of the reference point, and r is the downward viewing angle parameter of the scatterer P. i H is the slant range of the scatterer in the i-th column of the range direction in the SAR image, and H is the height of the SAR antenna phase center from the ground.

[0093] Taking the publicly available SAR dataset from the Chinese Academy of Sciences in the Emei region as an example, there are 12 images in total, with an image size of 3600×1800. The platform's flight altitude is approximately 1900m, which is consistent with the characteristics of a low-altitude SAR platform. Figure 6 (a) is a graph showing the magnitude of the data used. Figure 6 (b) is an optical map of Google Earth. Figure 7 The image shows a side view of the implementation result of the three-dimensional point cloud of the present invention. Figure 7 (a) is a diagram showing the elevation ambiguity phenomenon in a TomoSAR 3D point cloud. Figure 7 (b) is a side view of the 3D point cloud of the existing method (an imaging model based on the plane wave assumption). Figure 7 (c) is a three-dimensional point cloud side view of the method proposed in this invention (cylindrical wave model + height constraint + improved scatterer pre-estimation). Figure 8 (a) is a 3D point cloud front view of an existing method (an imaging model based on the plane wave assumption). Figure 8 (b) is a three-dimensional point cloud front view of the method of the present invention (cylindrical wave model + height constraint + improved scatterer pre-estimation).

[0094] As can be seen from the above, although existing methods can achieve TomoSAR 3D imaging, some elevation ambiguity and noise points still exist, and point cloud curvature distortion also exists. Compared with existing methods, the method proposed in this invention significantly reduces elevation ambiguity and noise points, makes buildings clearer, and effectively achieves deformation correction. Existing methods produce 3D point clouds with some artifacts, such as… Figure 8 (a) Elliptical bounding box region: The method shown in this invention suppresses artifacts. Meanwhile, existing methods produce 3D point clouds with uneven ground point heights; ground points at scene edges are lower than normal, resulting in some deformation distortion, such as... Figure 8 (a) Rectangular frame region; the method proposed in this invention reduces deformation distortion, such as Figure 8 (b) Rectangular area.

[0095] In summary, this invention modifies the TomoSAR imaging model, which is generally based on the plane wave assumption, into a model based on the cylindrical wave assumption, which facilitates geometric deformation correction in 3D imaging. An adaptive method for determining the elevation search range based on height constraints is proposed, filtering out elevation ambiguity. A refined scatterer pre-estimation method is employed, utilizing three threshold estimations to improve the accuracy of scatterer number pre-estimation and reduce imaging noise.

[0096] This invention studies 3D imaging of low-altitude TomoSAR platforms (i.e., airborne platforms and UAV platforms), considering the application of near-field conditions in TomoSAR imaging and the impact of elevation ambiguity on imaging. It proposes a TomoSAR imaging model based on cylindrical waves, an adaptive method for determining the elevation search range based on height constraints, and a refined scatterer pre-estimation method. Finally, it utilizes the OMP sparse inversion algorithm and the 3D localization of scatterers in the ground coordinate system to complete 3D imaging of multiple building scenes.

[0097] Compared to existing technologies, this invention solves the problems of elevation ambiguity and geometric deformation in TomoSAR imaging of low-altitude platforms. This invention achieves elevation de-ambiguity even when the elevation ambiguity location is very close to building height, and also reduces noise points. This invention improves geometric deformation and enhances the accuracy of 3D point clouds. This invention aligns with the rapid development of low-altitude SAR platforms and has broad application prospects in the research of low-altitude SAR platforms.

Claims

1. A tomographic SAR three-dimensional imaging method for a low-altitude platform, characterized in that, Includes the following steps: Acquire SAR images; Based on the cylindrical wave TomoSAR imaging model, sparse inversion is performed on the SAR image within the lower viewpoint constraint range to determine the scattering coefficient of the scatterer signal in the pixel along the elevation direction. Calculate the viewing angle parameters based on the scattering coefficient; The position coordinates of the scatterer in the ground coordinate system are generated based on the downward viewing angle parameters.

2. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 1, characterized in that, The cylindrical wave TomoSAR imaging model is as follows: Among them, g n Let θ′ represent the pixel value of the image formed by the nth antenna element, θ′ = θ - θ0, where θ is the downward viewing angle of the scatterer, θ0 is the downward viewing angle of the reference point (the ground point with the smallest slant range in the detection area), Δθ′ represents the constraint range, r is the slant range of the scatterer, γ(rθ′) represents the scattering coefficient of the signal along the elevation direction of the scatterer with slant range r and downward viewing angle parameter θ′, and ξ n Let ε be the elevation frequency of the nth antenna element. n The image noise is for the nth antenna element.

3. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 2, characterized in that, The method for generating the lower viewpoint constraint range is as follows: Determine the maximum height threshold h1 and minimum height threshold h2 of buildings in the detection area; Calculate the upper and lower bounds of the adaptive elevation search range based on the maximum height threshold h1 and the minimum height threshold h2. The lower view constraint range is generated based on the upper and lower bounds.

4. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 3, characterized in that, Before determining the scattering coefficient of the scatterer signal in a pixel along the elevation direction, the following steps are also included: Scatterer detection is performed on the SAR image.

5. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 4, characterized in that, The SAR image was subjected to scatterer detection using a three-threshold method; The three thresholds include the coherence coefficient threshold, the average amplitude threshold, and the amplitude deviation index threshold.

6. A tomographic SAR three-dimensional imaging method for a low-altitude platform as described in any one of claims 2-5, characterized in that, The objective problem when performing sparse inversion on the SAR image is: Where γ is the set of all scattering coefficients, g is the set of pixel values ​​of the image formed by N antenna elements, G is the observation matrix, and N is the total number of antenna elements. It is the standard deviation of the SAR image observation noise ε.

7. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 6, characterized in that, Generating the position coordinates of the scatterer in the ground coordinate system based on the aforementioned downward viewing angle parameters includes: Y P =k·Xbin, X P =r i ·sin(θ0+θ′ P )-X, Z P =-r i ·cos(θ0+θ′ P )+H, Among them, X P Y P and Z P θ′ represents the three-axis coordinates of the scatterer P in the ground coordinate system, k represents the column number of the scatterer P in the azimuth direction of the SAR image, Xbin is the azimuth resolution, and θ′ is the coordinates of the scatterer P. P Here, r represents the downward viewing angle parameter of the scatterer P, X is the X-axis coordinate of the reference point, and r is the downward viewing angle parameter of the scatterer P. i H is the slant range of the scatterer in the i-th column of the range direction in the SAR image, and H is the height of the SAR antenna phase center from the ground.

8. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 6, characterized in that, The target problem is solved using the OMP greedy algorithm, the basis pursuit algorithm, or the atomic norm minimization algorithm.

9. The tomographic SAR three-dimensional imaging method for a low-altitude platform as described in claim 3, characterized in that, The lower view constraint range is generated based on the upper and lower bounds, including: Δθ′ i =[s′ imin / r i ,s′ imax / r i ], Where, Δθ′ i Let s′ be the downward view constraint range in the i-th distance direction. imin Let s′ be the lower bound corresponding to the i-th distance direction. imax Let r be the upper bound corresponding to the i-th distance direction. i denoted as the slant range of the scatterer in the i-th column pixel of the SAR image range direction.

10. A tomographic SAR three-dimensional imaging device for a low-altitude platform, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 9.

Images