An airborne SAR ship imaging method based on hierarchical error decoupling

Abstract

The application discloses a layered error decoupling-based airborne SAR ship imaging method, which comprises the following steps: firstly, acquiring SAR echo data and platform initial navigation information, taking segmented acceleration correction parameters as to-be-estimated variables, reconstructing a high-precision platform trajectory and completing preliminary compensation of platform common geometric errors through kinematic integration and iterative optimization, and obtaining a coarse focusing imaging result; then, establishing a low-order polynomial model for residual spatial variable slant range errors caused by ship swinging, selecting a strong scattering constraint unit, solving local parameters in a local focusing optimization mode, fitting a full-scene error distribution through a least square method, and finally, unifying the corrected platform trajectory and the residual spatial variable slant range errors into a final compensation slant range history, and completing joint re-imaging.

Description

Technical Field

[0001] This invention relates to the field of signal processing, specifically to an airborne SAR ship imaging method based on hierarchical error decoupling, which is applicable to airborne SAR ship imaging on various flight platforms, including UAVs. Background Technology

[0002] Synthetic Aperture Radar (SAR) possesses all-weather, all-time imaging capabilities and high resolution, making it invaluable for monitoring ships at sea. SAR ship imaging is applicable to various airborne platforms, including UAVs.

[0003] Airborne SAR imaging is a flexible method for ship imaging, especially UAV-borne SAR, which has advantages such as flexible deployment, fast response and low cost. Therefore, it has good application prospects in the detection and imaging of ship targets on the sea surface.

[0004] In airborne, especially UAV-borne, SAR imaging of ships on the sea surface, platform motion errors and ship rolling errors jointly affect image quality. For UAV-borne SAR imaging, the deviation between the actual and ideal trajectory of the platform leads to target slant range history mismatch, which further manifests as imaging geometric mismatch and spatially varying defocus. At the same time, the ship's rolling motion under sea conditions causes different error compensation requirements for different scattering positions of the target at the same azimuth and time, further exhibiting obvious spatially varying defocus characteristics.

[0005] There are various existing methods for compensating for airborne SAR ship imaging. One known approach is to correct the platform motion error and the ship motion error separately, that is, to first compensate for the platform motion error and then further correct the residual error caused by the ship target.

[0006] For example, prior art 1 (CN111880180A) discloses a self-focusing method for high-resolution SAR imaging of moving ships. Firstly, to address platform motion errors, it primarily uses inertial navigation data to coarsely compensate the baseband echo signal, completing range-matched filtering and coarse focusing imaging. Secondly, to address errors caused by ship motion, it selects multiple strong scattering regions from the original image, estimates the local phase error of each strong scattering region using a weighted phase gradient self-focusing method, and then fits the phase error matrix of the entire image using a second-order polynomial to compensate the original image. This scheme can achieve unified estimation and compensation for the spatially varying errors of moving ships, but its platform motion error compensation is mainly based on coarse compensation using inertial navigation data. Due to measurement errors and accuracy limitations in inertial navigation information, in UAV-borne SAR imaging scenarios of ships on the sea surface, significant defocusing may still exist in the coarsely compensated image, indicating that platform-related residuals have not been sufficiently weakened. This residual will interact with errors caused by ship swaying, making subsequent error manifestations more complex. Meanwhile, image defocus also affects the accurate extraction of strong scattering regions in the image domain and the stability of local error estimation, thus increasing the difficulty of subsequent unified compensation for ship errors. Furthermore, its compensation for ship errors is mainly achieved by estimating local phase errors and fitting the phase error matrix of the entire image, with the focus on phase error compensation. However, in UAV-borne SAR sea surface ship imaging scenarios, the combined effect of platform errors and ship rolling errors, in addition to phase mismatch, may also be accompanied by residual range migration. Therefore, the current scheme still has certain limitations in correcting this type of residual range migration and cannot fully remove the residual influence.

[0007] Existing technology 2 (CN114384520A) discloses a method for refined radar imaging of ships on the sea surface using a mobile platform. This paper addresses platform motion errors primarily by using range pulse compression, travel correction, Doppler center compensation, and an acceleration compensation reference function to compensate for the influence of the mobile platform, resulting in a coarsely focused image of the ship target. Secondly, to address errors caused by the ship target itself, after detecting candidate ship targets, variable transformation and decoupling processing are performed on the target data, combined with PGA compensation and Decirp-clean processing to achieve refined imaging of the ship target. This scheme can achieve refined processing of ship targets under mobile platform conditions, but its platform motion error compensation is mainly based on the acceleration compensation reference function. It lacks targeted correction for spatial defocusing caused by platform trajectory deviations in UAV-borne SAR imaging of ships on the sea surface, therefore, complex platform-related residuals may still be retained after compensation. Meanwhile, the scheme still requires multiple steps, including candidate target detection, target data extraction, variable transformation, decoupling processing, distance-by-distance PGA compensation, and Decirp-clean processing. The overall processing chain is long and computationally intensive, resulting in a heavy imaging burden.

[0008] Therefore, even though existing technologies employ a dual-correction approach of first compensating for platform errors and then for ship errors, these methods still have shortcomings. On the one hand, platform error compensation often remains at the level of coarse inertial navigation compensation or correction of the compensation reference function, which is insufficient to fully mitigate the spatial variation and focus degradation caused by platform trajectory deviations in UAV-borne SAR sea surface ship imaging scenarios. On the other hand, ship error compensation is usually based on the platform coarse compensation results. If the platform-related residuals are still strong in the early stages, it will increase the difficulty of subsequent ship error modeling and estimation. In addition, due to the complexity of some correction methods, there are also problems such as long processing procedures and large computational loads.

[0009] In other words, existing technologies still struggle to reduce the complexity of subsequent residual errors while simultaneously addressing residual distance migration correction and azimuth matching compensation.

[0010] Therefore, in dual correction of airborne SAR ship imaging, how to coordinate and hierarchically process the common geometric errors of the airborne platform and the residual spatial variable slant range errors caused by ship rolling, in order to improve the focusing stability and imaging quality of ship targets, has become a technical problem that needs to be solved by existing technologies. Summary of the Invention

[0011] The purpose of this invention is to propose an airborne SAR ship imaging method based on layered error decoupling. This method addresses the problem in airborne SAR, especially UAV-borne SAR, ship imaging methods for the sea surface, where platform motion error and ship roll error are coupled, making it difficult to simultaneously achieve global consistent focusing and local fine focusing. The method combines two-stage residual compensation results for joint imaging.

[0012] To achieve this objective, the present invention adopts the following technical solution: An airborne SAR ship imaging method based on layered error decoupling includes the following steps: Acquire ship target echo data collected by the airborne SAR system and obtain the initial navigation information of the airborne platform corresponding to each position and time; The platform's initial acceleration parameters are obtained based on the initial navigation information. The synthetic aperture time is divided into several time sub-intervals, and a constant acceleration correction parameter is introduced as a variable to be estimated in each sub-interval. The platform's corrected velocity and corrected trajectory are reconstructed sequentially through kinematic integration. The target slant range history is calculated based on the corrected trajectory. The platform's acceleration correction parameters are iteratively optimized using the slant range history to obtain the optimal acceleration correction parameters and the corrected platform trajectory. Based on the corrected platform trajectory, the corrected slant range history of each target point at each position and time is recalculated. Compensation and correction are performed based on the corrected slant range history. After back projection imaging, the compensated coarse focusing result is obtained. A second-order polynomial model with azimuth and time as variables is established. Multiple strong scattering constraint elements are selected from the ship target. For each strong scattering constraint element, the local residual spatial variable slant range error is used as the variable to be estimated for optimization. The spatial positional relationship of each scattering constraint element is used to establish the solution equations for the parameters of the full-scene residual spatial variable slant range error model. Solving these equations yields the full-scene residual spatial variable slant range error at any target position. ; The correction slant distances of the target point at each direction and time were calculated based on the trajectory of the correction platform. The residual space of the entire scene is transformed into slant range error. As an additional slope range compensation term, it is added to the corrected slope range to obtain the final compensated slope range history. Based on the final compensated slant range history, back projection imaging yields a finely focused image of the ship target.

[0013] Optionally, the initial navigation information of the platform includes the platform position, platform velocity, and other inertial navigation measurements, provided using GPS, INS, or a combination thereof, to construct the initial values ​​of the platform acceleration parameters.

[0014] Optionally, the step of obtaining platform acceleration parameters based on the platform's initial navigation information, dividing the synthetic aperture time into several time sub-intervals, and introducing a constant acceleration correction parameter as a variable to be estimated in each sub-interval, specifically involves: Based on the initial navigation information of the platform, the three-dimensional acceleration parameters of the platform in the lateral, track, and vertical directions are obtained. The synthetic aperture time is divided into several time sub-intervals. In each sub-interval, constant three-dimensional acceleration correction parameters in the lateral, track, and vertical directions are introduced as variables to be estimated, and the piecewise acceleration correction amount is obtained.

[0015] Based on the initial navigation information of the platform, the three-dimensional acceleration parameters of the platform in the lateral, track, and vertical directions are obtained. The synthetic aperture time is divided into several time sub-intervals. In each sub-interval, constant three-dimensional acceleration correction parameters in the lateral, track, and vertical directions are introduced as variables to be estimated, and the piecewise acceleration correction amount is obtained.

[0016] Optionally, the step of calculating the target slant range history based on the corrected trajectory, and iteratively optimizing the platform acceleration correction parameters using the slant range history to obtain the optimal acceleration correction parameters and the corrected platform trajectory includes: The slant range history is introduced into the back projection imaging operator to obtain the corresponding complex image result. The platform acceleration correction parameter is iteratively optimized and solved using the image focusing quality evaluation index as the objective function until the objective function converges or meets the preset termination condition.

[0017] Optionally, the focus quality evaluation index includes one or more of image entropy, image contrast, image sharpness, and image acuity.

[0018] Optionally, the step of recalculating the corrected slant range history of each target point at each azimuth time based on the corrected platform trajectory, performing compensation and correction based on the corrected slant range history, and obtaining the compensated coarse focusing result after back-projection imaging, specifically involves: Based on the corrected platform trajectory, the corrected slant range history of each target point at each position and time is recalculated. Phase compensation and sampling correction are then performed based on the corrected slant range history. Subsequently, the corrected slant range history is introduced into the back projection imaging operator to coherently accumulate the echo data and obtain the compensated coarse focusing result.

[0019] Optional, residual space variable slant range error across the entire scenario The solution is as follows: Let the imaging coordinates of the target point be... Location and time are The residual spatial slant distance error that still exists after preliminary compensation is denoted as: , in, , and It is a quadratic coefficient function related to the target spatial position, used to characterize the spatial variation characteristics of the residual spatial variable slant range error; Selecting reference points from ship targets As the unfolding center, the coordinate offsets of the target point relative to this reference point are defined as follows: , Let the total selection be The scattering confinement unit, the th scattering confinement unit, The center coordinates of the scattering confinement unit are: Its coordinate offset relative to the reference point is: , For each scattering constraint unit, within the preset parameter search range, the residual space variable slant range error parameters corresponding to its local region are solved, and the parameter combination that enables the local region to achieve the optimal focusing state is solved by the local focusing evaluation and optimization method. Let the first The local residual space variable slant range error corresponding to each scattering constraint element is: , in, , and The first Local observation coefficients corresponding to each scattering constraint element After getting all Local observation coefficients corresponding to each scattering constraint unit , , Then, the solution equations for the residual spatial variable slant range error model parameters of the whole scene are established by using the spatial positional relationship of each scattering constraint unit.

[0020] Optionally, the residual spatial variable slant range error model parameters for the entire scenario include... , And adopting about and The quadratic polynomial representation of the equations is used to establish a global set of equations based on the local observation coefficients and spatial positions of all scattering constraint units. The least squares method is then used to solve for the parameters of the variable slant range error model of the residual space in the whole scene. Obtained using parameters from the full-scene residual space variable slant range error model , and The final expression; according to , and Recover the residual spatial variable slant range error at any target location across the entire scene: .

[0021] Optionally, phase compensation, sampling correction, and coherent accumulation processing are performed on the echo data based on the final compensated slant range history, and a finely focused image of the ship target is obtained using the back projection imaging method.

[0022] In summary, the present invention has the following advantages: 1. This invention employs layered error decoupling to first accurately compensate for the platform's common geometric errors, and then models the variable slant range error of the ship's swaying residual space, significantly reducing error coupling interference. This reduces the interference of platform residuals on the ship's error model, avoids the model complexity and solution divergence problems caused by mixed errors, and helps reduce the model's degrees of freedom, improving the stability of scattering constraint extraction and parameter estimation.

[0023] 2. This invention aligns with the actual physical model of airborne SAR observation, focusing on trajectory reconstruction and slant range correction while considering both global consistent focusing and local fine focusing. Compared to traditional methods that only perform phase compensation or single error correction, this invention uses the platform trajectory and the ship's residual spatial variable slant range error as a unified compensation object, and combines it with the imaging process for joint compensation and re-imaging. This approach better balances residual range migration correction and azimuth matching compensation, improving the overall focusing effect and imaging quality of ship targets.

[0024] 3. This invention adopts a low-order modeling and unified compensation architecture, which simplifies the processing flow, makes the computation more controllable, and enhances engineering feasibility. By fusing the two layers of error correction into the same slant range history for joint re-imaging, the problem of error propagation and inconsistency in compensation caused by step-by-step compensation is avoided. While ensuring the targeted nature of error compensation, the algorithm complexity is effectively reduced, making it more suitable for the real-time and lightweight engineering application requirements of UAV platforms. Attached Figure Description

[0025] Figure 1 This is an airborne SAR ship imaging method based on hierarchical error decoupling according to a specific embodiment of the present invention; Figure 2 This is a SAR imaging result image of the existing phase gradient autofocusing method; Figure 3 This is a SAR imaging result diagram of an airborne SAR ship imaging method based on layered error decoupling according to a specific embodiment of the present invention. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0027] The invention is as follows: In the first stage, when performing geometric error compensation on the airborne platform, the acceleration correction parameter is used as the variable to be estimated to reconstruct the platform trajectory, and the slant range history is introduced into the back projection imaging operator for optimization; In the second stage, the residual spatial variable slant range error caused by ship swaying is modeled and estimated, and the error propagation and compensation inconsistency caused by the separate correction of the slant range history are reduced by a unified slant range history compensation method; Finally, the compensation results of the two stages are combined to complete the joint re-imaging, which avoids the problem of error propagation and compensation inconsistency caused by separate correction of different processing stages, and improves the focusing stability and imaging quality of the ship target.

[0028] For details, see Figure 1 This paper illustrates an airborne SAR ship imaging method based on layered error decoupling according to the present invention, comprising the following steps: Echo and navigation information acquisition step S110: The system acquires ship target echo data collected by the airborne SAR system and obtains the initial navigation information of the airborne platform corresponding to each position and time. The initial navigation information of the platform includes the platform position, platform velocity, inertial navigation measurement, or other parameters that can characterize the platform motion state, and is used to construct the initial values ​​of the platform acceleration parameters, i.e., the initial acceleration parameters.

[0029] In this invention, it is preferred to use an unmanned aerial vehicle (UAV) platform to carry a SAR payload to observe ship targets on the sea surface and obtain corresponding echo data. The echo data is either range-compressed data or raw echo data suitable for subsequent imaging processing.

[0030] In an optional embodiment, the initial navigation information of the platform is platform navigation information provided by GPS, INS, or a combination thereof, which serves as the initial basis for platform acceleration parameter estimation and trajectory reconstruction.

[0031] Acceleration-based platform trajectory reconstruction step S120: To improve the physical interpretability of the first-stage platform common geometric error compensation and reduce the high-dimensional solution complexity caused by directly optimizing the platform trajectory time-by-time and point-by-point, this step uses the platform acceleration correction parameter as the variable to be estimated in the first stage, and corrects and reconstructs the platform trajectory based on the initial navigation information. Acceleration is used as the parameter to be optimized; the initial acceleration parameters of the platform are given by the initial navigation information, and the platform acceleration is continuously iterated. This parameter form is not only easy to obtain from the initial navigation information of the airborne platform, but also more closely reflects the mechanism of the platform trajectory changes.

[0032] The steps are as follows: obtain the initial acceleration parameters of the platform based on the initial navigation information of the platform; divide the synthetic aperture time into several time sub-intervals; introduce constant acceleration correction parameters as variables to be estimated in each sub-interval; reconstruct the platform correction velocity and correction trajectory sequentially through kinematic integration; calculate the target slant range history based on the correction trajectory; use the slant range history to iteratively optimize and solve the platform acceleration correction parameters to obtain the optimal acceleration correction parameters and the corresponding corrected platform trajectory.

[0033] Therefore, in this step, the initial acceleration parameter represents the initial acceleration parameter obtained from the initial navigation information of the airborne platform at the start of the iteration, the acceleration correction parameter represents the acceleration correction amount during the iteration process, the optimal acceleration correction parameter represents the acceleration correction amount obtained at the end of the iteration, and the initial acceleration parameter is superimposed with the optimal acceleration correction amount, that is, the optimal acceleration correction parameter, to obtain the final acceleration value.

[0034] Specifically, the initial acceleration parameters of the platform are obtained based on the initial navigation information of the platform, and the synthetic aperture time is divided into several time sub-intervals. A constant acceleration correction parameter is introduced as a variable to be estimated in each sub-interval. Based on the initial navigation information of the platform, the three-dimensional acceleration parameters of the platform in the lateral, track, and vertical directions are obtained. The synthetic aperture time is divided into several time sub-intervals. In each sub-interval, constant three-dimensional acceleration correction parameters in the lateral, track, and vertical directions are introduced as variables to be estimated, and the piecewise acceleration correction amount is obtained.

[0035] The specific steps of reconstructing the platform correction velocity and correction trajectory sequentially through kinematic integration are as follows: the corrected acceleration is obtained from the initial reference acceleration and the piecewise acceleration correction, and the platform correction velocity and platform correction trajectory are reconstructed sequentially through kinematic integration.

[0036] The initial reference acceleration can be obtained from the platform's initial navigation information.

[0037] In one specific embodiment Let the reference acceleration corresponding to the reference trajectory given by the initial navigation information of the platform be... ,

[0038] in, , and These represent the reference acceleration components of the reference trajectory in the lateral, track, and vertical directions, respectively.

[0039] Let the time interval for synthesizing aperture be... Divide it into Each time sub-interval introduces a platform acceleration correction parameter to characterize the platform's motion error relative to the reference trajectory within that time interval. The platform acceleration correction parameter within each time sub-interval can be expressed as: ,

[0040] in, , and They represent the first The acceleration corrections in the lateral, track, and vertical directions within each time sub-interval. Preferably, the acceleration corrections are approximated as constant values ​​within each time sub-interval.

[0041] From this, the total acceleration after platform correction can be obtained. , in, It is a three-dimensional acceleration correction function composed of acceleration correction parameters segmented within each time sub-interval.

[0042] The corrected total acceleration Then, the platform velocity is reconstructed based on kinematic relationships. Let the initial time be... The platform speed is Then at time The platform speed can be expressed as: ,

[0043] Furthermore, let the initial time be... The platform location is Then, based on the reconstructed platform speed, the corrected platform trajectory can be obtained: .

[0044] The step of calculating the target slant range history based on the corrected trajectory, and iteratively optimizing the platform acceleration parameters using the slant range history to obtain the optimal acceleration correction parameters and the corrected platform trajectory includes: The slant range history is introduced into the back projection imaging operator to obtain the corresponding complex image result. The platform acceleration correction parameter is iteratively optimized and solved using the image focusing quality evaluation index as the objective function until the objective function converges or meets the preset termination condition.

[0045] The focus quality evaluation indicators include one or more of image entropy, image contrast, image sharpness, and image acuity.

[0046] The iterative optimization solution can be obtained using gradient descent, adaptive moment estimation (Adam optimization algorithm), or other numerical optimization methods.

[0047] The objective function converges or meets the preset termination conditions, including: the change in the objective function value over multiple iterations is less than the preset convergence threshold, the update magnitude of the parameter to be optimized over multiple iterations is less than the preset parameter threshold, or the number of iterations reaches the preset maximum number of iterations.

[0048] This invention uses minimizing image entropy as the optimization objective for illustrative purposes.

[0049] For any target point on the imaging plane According to the corrected platform trajectory Calculate its position and time. Corresponding slant distance history: .

[0050] Let the amplitude image of the imaging result be... , Given the row and column indices of the image pixels, the normalized energy distribution is: ,

[0051] The image entropy can then be expressed as: .

[0052] By iteratively updating the platform acceleration correction parameters, the objective function is improved. The acceleration is gradually reduced until the preset termination condition is met, thereby obtaining the optimal platform acceleration correction parameters and the corresponding corrected platform trajectory.

[0053] Therefore, in this step, the platform acceleration correction parameters are not directly used to construct the signal domain compensation reference function, but rather serve as physical parameters for platform trajectory correction. These parameters correct the platform velocity and trajectory through kinematic relationships and are further used to calculate a more accurate target slant range history. Compared to existing methods that directly compensate signals based on acceleration compensation reference functions, this embodiment starts from the actual motion geometry of the platform and performs parameterized correction on the platform trajectory, which is more consistent with the physical observation process of airborne SAR imaging, especially UAV-borne SAR imaging.

[0054] Step S130: Coarse focusing imaging based on the corrected platform trajectory: This step involves using the corrected platform trajectory obtained in the previous step to compensate for the echo data, thereby reducing the common geometric error of the platform caused by the platform motion error and thus obtaining a coarse focusing result.

[0055] This includes: recalculating the corrected slant range history of each target point at each position and time according to the corrected platform trajectory, performing compensation and correction according to the corrected slant range history, and obtaining the compensated coarse focusing result after back projection imaging.

[0056] In one specific embodiment, the corrected slant range history of each target point at each position time is recalculated based on the corrected platform trajectory, and phase compensation and sampling correction are performed based on the corrected slant range history. Subsequently, the corrected slant range history is introduced into the back projection imaging operator to coherently accumulate the echo data to obtain the compensated coarse focusing result.

[0057] In this invention, the backprojection operator (BP operator) mentioned is a core operator for time-domain imaging and a common projection algorithm in the field of radar imaging. This algorithm maps the received echo data in reverse according to the geometric path of electromagnetic wave propagation and coherently superimposes it onto the image grid, thereby reconstructing the reflectivity distribution of the target scene.

[0058] Specifically, in the back projection algorithm, the radar platform obtains the time delay by using the slant distance from the transmitting platform to the pixel, constructs a phase compensation factor by using the two-way time delay, and uses the phase compensation factor to compensate for the phase error of the pulse compression echo signal to obtain the phase-compensated signal. The phase-compensated signal is then coherently superimposed to obtain the imaging result.

[0059] Assuming the radar echo signal obtained after range compression is , For distance to time, This is the azimuth time. The slant range history for a given target point is... Then its corresponding delay for , The speed of light. The phase compensation factor is constructed based on the time delay. for , The imaging results are obtained by coherently superimposing the radar carrier frequencies. .

[0060] The above description is for illustrative purposes only, and the slant range history is incorporated into the back projection operator imaging process.

[0061] This step primarily aims to mitigate the common geometric errors of the platform caused by trajectory deviations of the airborne platform, rather than eliminating all imaging errors at once. Through this first-stage compensation, the magnitude of subsequent residual spatial variable slant range errors can be reduced, their changes smoother, and a more accurate geometric reference can be provided for the second-stage modeling of residual spatial variable slant range errors.

[0062] Because the ship target may still experience swaying, turning, and other movements under sea conditions, residual spatial variable slant range errors related to the target's spatial position may still be retained in the image after the first-stage compensation by the airborne platform. This coarse focusing result and its corresponding geometric reference will serve as the basis for subsequent modeling and estimation of the residual spatial variable slant range error.

[0063] Steps S140 for solving the ship's sway residual space variable slant distance error model: Through steps S120-S130 above, the coarse focusing result and the corrected platform geometric reference have been obtained. At this point, the residual error mainly originates from the residual spatial variable slant range error caused by the ship's rolling motion. Compared with directly processing mixed errors in the context of strong platform residuals, the subsequent stages face smaller error amplitudes, smoother changes, and more regular error forms based on the compensation in the previous stage. Furthermore, due to the ship's rolling, pitching, and heave in the waves, different slant range errors exist at different scattering points. The error varies with spatial position (x, y), which is a spatially variable error and cannot be compensated using a globally unified phase compensation method.

[0064] Therefore, in this step, under the premise that the platform error has been significantly reduced, a low-order parametric model is performed on the ship's residual air-variable slant range error to reduce the amount of computation and control the complexity of subsequent processing; and the full-scene error is solved by using local strong scattering point constraints; focusing on the slant range deviation caused by the ship's own motion, and constructing the final compensated slant range history with the corrected platform trajectory from the previous step, and completing joint compensation re-imaging under the same back-projection imaging framework, which is beneficial to improving the consistency of the compensation process and the final imaging effect.

[0065] Therefore, the steps are as follows: Establish a low-order polynomial model with azimuth and time as variables, where the model coefficients are expressed as quadratic polynomials of the target spatial coordinates; select multiple strong scattering constraint elements from the ship target, and optimize the solution for each strong scattering constraint element using the local residual space variable slant range error as the variable to be estimated; and establish the solution equations for the parameters of the full-scene residual space variable slant range error model using the spatial positional relationships of each scattering constraint element, solving for the full-scene residual space variable slant range error model parameters, and then obtaining the full-scene residual space variable slant range error at any target position. .

[0066] The strong scattering constraint unit refers to a scattering point or scattering region that has high energy, stable structure, and can reflect the residual focusing error in its local area in the coarse focusing result. It can be selected manually or by setting a certain threshold for screening.

[0067] Specifically, a low-order parametric model relating to azimuth and time is used to describe the residual spatial variable slant range error. Let the imaging coordinates of the target point be... Location and time are The residual spatial slant distance error that still exists after compensation in step S130 is denoted as: , in, , and It is a coefficient function related to the target spatial position, used to characterize the spatial variation characteristics of the residual spatial variable slant range error.

[0068] A reference point within the ship target is selected as the unfolding center, and the coordinate offsets of the target point relative to this reference point are defined as follows: , in, The coordinates are used as a reference point. Preferably, the coefficient function... , and Can adopt about and The low-order function representation is preferred in this embodiment, which preferably uses a quadratic polynomial form.

[0069] After obtaining the first-stage coarse focusing results, multiple scattering constraint units are selected from the ship target image as observation units for subsequent local parameter solving. These scattering constraint units are local high-energy scattering points or regions containing local main scattering structures. This invention does not limit the scattering constraint units to a single form; any unit that can stably characterize the local residual error features in the coarse focusing results can be used as a constraint unit for subsequent parameter solving. The parameters of the full-scene residual spatial variable slant range error model are solved using the local optimal focusing parameters corresponding to each scattering constraint unit and their spatial positional relationships, as detailed below: Let the total selection be The scattering confinement unit, the th scattering confinement unit, The center coordinates of the scattering confinement unit are: Its coordinate offset relative to the reference point is:

[0070] For each scattering constraint unit, within the preset parameter search range, the residual space variable slant range error parameters corresponding to its local region are solved. In this embodiment, the local residual space variable slant range error model parameters are directly used as the variables to be estimated, and the parameter combination that makes the local region achieve the optimal focusing state is solved through the local focusing evaluation and optimization method.

[0071] Let the first The local residual space variable slant range error corresponding to each scattering constraint element is:

[0072] in, , and The first The local observation coefficients corresponding to each scattering constraint element.

[0073] The local focus evaluation and optimization method is the minimum entropy method, the image contrast optimization method, the image sharpness optimization method, or other traditional autofocus methods.

[0074] In a preferred embodiment, the minimum entropy method is used to solve for the local residual space variable slant range error parameters. That is, in the first... Within the local region corresponding to each scattering constraint unit, the candidate parameters , and A search is performed to optimize the focusing evaluation index of the imaging results in this local area, thereby obtaining the first... The local observation coefficients corresponding to each scattering constraint element.

[0075] After getting all Local observation coefficients corresponding to each scattering constraint unit , , Then, the spatial positional relationships of each scattering constraint unit can be used to establish the solution equations for the parameters of the residual space variable slant range error model for the entire scene. Specifically, we have:

[0076] when , and Adopting about and When expressing the quadratic polynomial, a corresponding global equation system can be established based on the local observation coefficients and spatial positions of all scattering constraint units. The least squares method is then used to solve for the parameters of the full-scene residual space variable slant range error model, thus obtaining the parameters of the full-scene residual space variable slant range error model.

[0077] in, , , , The parameters are the residual space variable slant distance error model parameters for the entire scene to be solved.

[0078] Using the parameters of the variable slant range error model in the residual space of the entire scene, we obtain , and The final expression;

[0079] This allows for the recovery of the full-scene residual spatial variable slant range error at any target location: .

[0080] This step does not directly address the complex overall error caused by the combined effects of platform and ship errors. Instead, it focuses on low-order modeling and parameter solving for the residual spatial variable slant range error caused by ship swaying, given that the common geometric error of the platform has been mitigated first. This helps reduce the degrees of freedom of the subsequent model, improves the stability of parameter solving, and provides a more accurate residual slant range correction for subsequent joint compensation re-imaging.

[0081] Steps S120 and S140 are sequentially linked, providing the final compensated slant range history for the final imaging in step S150.

[0082] Joint compensation imaging step S150: This step combines platform error compensation and ship error compensation into the final slant range, and uses the back projection operator to re-image, avoiding the error propagation caused by segmented compensation. It achieves a comprehensive unification of global trajectory accuracy and local ship air-variable error correction, resulting in a final high-quality imaging result.

[0083] This includes: firstly, calculating the corrected slant distance of the target point at each position and time based on the trajectory of the correction platform. Then, the obtained full-scene residual space variable slant range error is... As an additional slope range compensation term, combined with the corrected slope range, the final compensated slope range history is obtained:

[0084] in, For target point In terms of location and time The corresponding final compensated slant range history is used to obtain a finely focused image of the ship target through back-projection imaging.

[0085] Furthermore, based on the final compensated slant range history, phase compensation, sampling correction, and coherent accumulation processing are performed on the echo data. Then, using the back projection imaging method, a finely focused image of the ship target is obtained.

[0086] Therefore, this invention applies the first-stage platform common geometric error correction result and the second-stage ship residual space variable slant range error correction result to the same final compensation slant range process, avoiding the problem of inconsistent error transmission and compensation caused by separate correction in different processing stages. This allows for a better balance between global consistent focusing and local fine focusing, thereby improving the overall imaging effect of ship targets.

[0087] Compared with existing technologies that only perform platform motion error compensation or only perform local autofocus compensation, this invention processes platform common geometric error compensation and ship residual space variable slant range error compensation in layers and integrates them uniformly in the final imaging stage. This can simultaneously take into account both global consistent focusing and local fine focusing, thereby improving the overall imaging effect of ship targets.

[0088] Example: In this embodiment, the parameters shown in Table 1 are used to process the actual collected UAV-borne SAR observation data of ships on the sea surface to verify the effectiveness of the method proposed in this invention.

[0089]

[0090] See Figure 2 This paper presents SAR imaging results obtained using the existing phase gradient autofocus method. Figure 3 An imaging result diagram according to the present invention is shown. Figure 2 and Figure 3As can be seen from the comparison, after adopting the method of the present invention, the scattering structure of the ship target is more concentrated, and the target outline and local structure are clearer, indicating that the present invention can effectively improve the focusing effect of ship target imaging. Since the present invention reduces the degree of freedom in solving residual errors through layered modeling and integrates the two-stage compensation quantities into the final slant range history for re-imaging, it has good engineering feasibility.

[0091] In summary, the present invention has the following advantages: 1. This invention employs layered error decoupling to first accurately compensate for the platform's common geometric errors, and then models the variable slant range error of the ship's swaying residual space, significantly reducing error coupling interference. This reduces the interference of platform residuals on the ship's error model, avoids the model complexity and solution divergence problems caused by mixed errors, and helps reduce the model's degrees of freedom, improving the stability of scattering constraint extraction and parameter estimation.

[0092] 2. This invention aligns with the actual physical model of airborne SAR observation, focusing on trajectory reconstruction and slant range correction while considering both global consistent focusing and local fine focusing. Compared to traditional methods that only perform phase compensation or single error correction, this invention uses the platform trajectory and the ship's residual spatial variable slant range error as a unified compensation object, and combines it with the imaging process for joint compensation and re-imaging. This approach better balances residual range migration correction and azimuth matching compensation, improving the overall focusing effect and imaging quality of ship targets.

[0093] 3. This invention adopts a low-order modeling and unified compensation architecture, which simplifies the processing flow, makes the computation more controllable, and enhances engineering feasibility. By fusing the two layers of error correction into the same slant range history for joint re-imaging, the problem of error propagation and inconsistency in compensation caused by step-by-step compensation is avoided. While ensuring the targeted nature of error compensation, the algorithm complexity is effectively reduced, making it more suitable for the real-time and lightweight engineering application requirements of UAV platforms.

[0094] Obviously, those skilled in the art will understand that the various units or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device, or alternatively, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by the computing device. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0095] The above description is a further detailed explanation of the present invention in conjunction with specific preferred embodiments. It should not be considered that the specific embodiments of the present invention are limited to this. For those skilled in the art, several simple deductions or substitutions can be made without departing from the concept of the present invention, and all such deductions or substitutions should be considered to fall within the scope of protection of the present invention as defined by the submitted claims.

Claims

1. An airborne SAR ship imaging method based on layered error decoupling, characterized in that, Includes the following steps: Acquire ship target echo data collected by the airborne SAR system and obtain the initial navigation information of the airborne platform corresponding to each position and time; The platform's initial acceleration parameters are obtained based on the initial navigation information. The synthetic aperture time is divided into several time sub-intervals, and a constant acceleration correction parameter is introduced as a variable to be estimated in each sub-interval. The platform's corrected velocity and corrected trajectory are reconstructed sequentially through kinematic integration. The target slant range history is calculated based on the corrected trajectory. The platform's acceleration correction parameters are iteratively optimized using the slant range history to obtain the optimal acceleration correction parameters and the corrected platform trajectory. Based on the corrected platform trajectory, the corrected slant range history of each target point at each position and time is recalculated. Compensation and correction are performed based on the corrected slant range history. After back projection imaging, the compensated coarse focusing result is obtained. A second-order polynomial model with azimuth and time as variables is established. Multiple strong scattering constraint elements are selected from the ship target. For each strong scattering constraint element, the local residual spatial variable slant range error is used as the variable to be estimated for optimization. The solution equation for the parameters of the full-scene residual spatial variable slant range error model is established using the spatial positional relationship of each scattering constraint element. The parameters of the full-scene residual spatial variable slant range error model are obtained by solving the equation, and then the full-scene residual spatial variable slant range error at any target position is obtained. ; The correction slant distances of the target point at each direction and time were calculated based on the trajectory of the correction platform. The residual space of the entire scene is transformed into slant range error. As an additional slope range compensation term, it is added to the corrected slope range to obtain the final compensated slope range history. Based on the final compensated slant range history, back projection imaging yields a finely focused image of the ship target.

2. The airborne SAR ship imaging method according to claim 1, characterized in that: The initial navigation information of the platform includes the platform position, platform velocity, and other inertial navigation measurements, provided using GPS, INS, or a combination thereof, and is used to construct the initial values ​​of the platform's acceleration parameters.

3. The airborne SAR ship imaging method according to claim 1, characterized in that: The process involves obtaining platform acceleration parameters based on the platform's initial navigation information, dividing the synthetic aperture time into several time sub-intervals, and introducing a constant acceleration correction parameter as a variable to be estimated within each sub-interval. Specifically: Based on the initial navigation information of the platform, the three-dimensional acceleration parameters of the platform in the lateral, track, and vertical directions are obtained. The synthetic aperture time is divided into several time sub-intervals. In each sub-interval, constant three-dimensional acceleration correction parameters in the lateral, track, and vertical directions are introduced as variables to be estimated, and the piecewise acceleration correction amount is obtained.

4. The airborne SAR ship imaging method according to claim 3, characterized in that: The specific steps of reconstructing the platform correction velocity and correction trajectory sequentially through kinematic integration are as follows: the corrected acceleration is obtained from the initial reference acceleration and the piecewise acceleration correction, and the platform correction velocity and platform correction trajectory are reconstructed sequentially through kinematic integration.

5. The airborne SAR ship imaging method according to claim 4, characterized in that: The step of calculating the target slant range history based on the corrected trajectory, and iteratively optimizing the platform acceleration correction parameters using the slant range history to obtain the optimal acceleration correction parameters and the corrected platform trajectory includes: The slant range history is introduced into the back projection imaging operator to obtain the corresponding complex image result. The platform acceleration correction parameter is iteratively optimized and solved using the image focusing quality evaluation index as the objective function until the objective function converges or meets the preset termination condition.

6. The airborne SAR ship imaging method according to claim 5, characterized in that: The focus quality evaluation indicators include one or more of image entropy, image contrast, image sharpness, and image acuity.

7. The airborne SAR ship imaging method according to claim 1, characterized in that: The process involves recalculating the corrected slant range history of each target point at each location based on the corrected platform trajectory, performing compensation and correction based on the corrected slant range history, and obtaining the compensated coarse focusing result after back-projection imaging. Specifically: Based on the corrected platform trajectory, the corrected slant range history of each target point at each position and time is recalculated. Phase compensation and sampling correction are then performed based on the corrected slant range history. Subsequently, the corrected slant range history is introduced into the back projection imaging operator to coherently accumulate the echo data and obtain the compensated coarse focusing result.

8. The airborne SAR ship imaging method according to claim 1, characterized in that: Full-scene residual space variable slant distance error The solution is as follows: Let the imaging coordinates of the target point be... Location and time are The residual spatial slant distance error that still exists after preliminary compensation is denoted as: ， in, , and It is a quadratic coefficient function related to the target spatial position, used to characterize the spatial variation characteristics of the residual spatial variable slant range error; Selecting reference points from ship targets As the unfolding center, the coordinate offsets of the target point relative to this reference point are defined as follows: ， Let the total selection be The scattering confinement unit, the th scattering confinement unit, The center coordinates of the scattering confinement unit are: Its coordinate offset relative to the reference point is: ， For each scattering constraint unit, within the preset parameter search range, the residual space variable slant range error parameters corresponding to its local region are solved, and the parameter combination that enables the local region to achieve the optimal focusing state is solved by the local focusing evaluation and optimization method. Let the first The local residual space variable slant range error corresponding to each scattering constraint element is: ， in, , and The first Local observation coefficients corresponding to each scattering constraint element After getting all Local observation system corresponding to each scattering constraint unit , Then, the solution equations for the residual spatial variable slant range error model parameters of the whole scene are established by using the spatial positional relationship of each scattering constraint unit.

9. The airborne SAR ship imaging method according to claim 8, characterized in that: The residual space variable slant range error model parameters for the entire scenario include ， And adopting about and The quadratic polynomial representation of the equations is used to establish a global set of equations based on the local observation coefficients and spatial positions of all scattering constraint units. The least squares method is then used to solve for the parameters of the variable slant range error model of the residual space in the whole scene. Obtained using parameters from the full-scene residual space variable slant range error model , and The final expression; according to , and Recover the residual spatial variable slant range error at any target location across the entire scene: 。 10. The airborne SAR ship imaging method according to claim 9, characterized in that: Based on the final compensated slant range history, phase compensation, sampling correction, and coherent accumulation processing are performed on the echo data. Then, a finely focused image of the ship target is obtained using the back projection imaging method.

Images