SAR periodic missing signal recovery method based on echo structure priori knowledge

Abstract

The invention discloses an SAR periodic missing signal recovery method based on echo structure priori knowledge. The method comprises the following steps: acquiring an original echo signal sampled by a synthetic aperture radar; based on the sampling structure of the original echo signal, partitioning the original echo signal along the distance direction to obtain a plurality of distance blocks; each distance block comprises a plurality of sampling signals; determining a target distance unit with the maximum signal-to-noise ratio in each distance block, and calculating a recovery coefficient of each target distance unit; preprocessing the recovery coefficient of each target distance unit to obtain a target recovery coefficient of a distance block where each target distance unit is located; and determining a missing signal in the original echo signal according to the target recovery coefficient of each distance block, a predetermined target estimation error compensation parameter and the original echo signal. The method can be popularized to a full-distance section only by estimating the recovery coefficient of a part of distance units, and the operation time is greatly shortened while the pulse recovery precision is ensured.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing technology, specifically relating to a method for recovering SAR periodically missing signals based on prior knowledge of echo structure. Background Technology

[0002] Ideally, Synthetic Aperture Radar (SAR) imaging echoes are acquired by the radar through periodic and uniform sampling in both range and azimuth dimensions along a pre-defined trajectory. However, due to various factors in reality, such as single-radar multi-task switching, time-frequency synchronization signal transmission, and cross-reception of signals in the same frequency band, some echoes may be periodically missing or unusable, resulting in a non-ideal signal. Simply zeroing out the non-ideal signal would destroy the coherence of the uniformly sampled signal, leading to numerous high-energy raster lobe ghost images in the direction of the missing signal, obscuring the original scene information and affecting the visibility and interpretability of the SAR image. In practical applications, to ensure that the quality of the acquired SAR image is not compromised, a fast and effective algorithm for recovering missing signals is needed to restore the signal coherence. Since the sampling time in the azimuth direction is much longer than that in the range direction, the recovery of missing signals in the azimuth direction is the primary consideration.

[0003] In recent years, compressed sensing algorithms have become a new research hotspot due to their high performance and robustness, capable of reconstructing signals even under conditions of poor signal-to-noise ratio and high signal loss rate. Compressed sensing algorithms mainly reconstruct the signal by representing the coefficients of the signal to be reconstructed, constructing an observation matrix, and then using appropriate algorithms. Typical examples include the OMP algorithm and the st-OMP algorithm, which reduces the number of iterations. However, despite their excellent reconstruction results, these algorithms come at the cost of requiring iterative optimization of the estimation matrix, with an indefinite number of iterations and a huge computational burden, limiting their application to post-processing stages where algorithmic time consumption is not a concern.

[0004] Besides compressed sensing, mainstream algorithms can be divided into two main types: interpolation and spectral estimation. Interpolation algorithms are often used to perform equivalent homogenization of non-uniformly sampled data. Representative examples include sinc interpolation, cubic interpolation, and linear optimal unbiased interpolation. Interpolation algorithms typically determine interpolation weights based on the relative positional relationships between data points or on an optimization objective based on interpolation accuracy, and then complete the interpolation region by combining the surrounding area with the interpolation weights. Interpolation algorithms usually achieve good accuracy when the signal satisfies the Nyquist sampling theorem. However, to reduce data redundancy, the SAR azimuth sampling rate is usually set slightly larger than the azimuth bandwidth. Therefore, in local areas lacking data, since the Nyquist sampling rate is no longer satisfied, low-order algorithms like interpolation are insufficient to acquire enough information and are no longer usable.

[0005] Spectral estimation algorithms typically estimate spectral parameters based on the non-missing portion of the signal, using the information obtained from this estimation to guide signal recovery. The Burg algorithm, a stable autoregressive algorithm, obtains linear prediction coefficients by minimizing the average power error of the power spectrum in each order of forward and backward predictions. The GAPES algorithm, by estimating complex amplitudes and preserving phase information, achieves higher estimation accuracy and uses least squares interpolation to recover missing data. The MIAA algorithm also estimates the complex amplitude of the signal, calculating optimal weights for each frequency point to be estimated, iteratively alternating between spectral estimation and signal recovery until the recovery achieves the desired result. These spectral estimation algorithms require sequential estimation of each missing portion, resulting in a computationally intensive workload for the range-azimuth sampling number of SAR echoes. In summary, while many existing algorithms address the problem of missing data, those effective for low oversampling rates are limited by high computational demands, lacking a fast and effective method for recovering missing signals. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides a method for recovering periodically missing SAR signals based on prior knowledge of echo structure.

[0007] The technical problem to be solved by this invention is achieved through the following technical solution: This invention provides a method for recovering periodically missing SAR signals based on prior knowledge of echo structure, comprising: Acquire the raw echo signal obtained from synthetic aperture radar sampling; Based on the sampling structure of the original echo signal, the original echo signal is divided into blocks along the range direction to obtain multiple range blocks; each range block includes multiple sampled signals. Identify the target range cell with the highest signal-to-noise ratio in each range block and calculate the recovery coefficient of each target range cell; The recovery coefficients of each target range cell are preprocessed to obtain the target recovery coefficients of the range block to which each target range cell belongs; Based on the target recovery coefficient of each distance block, the pre-determined target estimation error compensation parameters, and the original echo signal, the missing signal in the original echo signal is determined.

[0008] This invention provides a method for recovering periodically missing SAR signals based on prior knowledge of echo structure. It only requires estimating the recovery coefficients of some range cells and can be extended to the entire range. While ensuring the accuracy of pulse recovery, it significantly reduces the computation time. It can be applied to the rapid idealization of non-ideal SAR echoes with some pulse problems, and avoids the impact of interference factors such as grating lobes on the subsequent use of SAR images.

[0009] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the SAR periodic missing signal recovery method based on prior knowledge of echo structure provided in an embodiment of the present invention. Figure 2 This is a geometric model diagram of the ideal SAR signal acquisition in an embodiment of the present invention; Figure 3 This is a schematic diagram showing the location of the grating lobe in an embodiment of the present invention; Figure 4 This is a schematic diagram comparing SAR images obtained from periodically missing signals and ideal signals according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the sampling of the original echo data in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the experimental results of how distance pulse compression processing improves the recovery of missing signals according to an embodiment of the present invention. Figures 7A to 7D These are schematic diagrams illustrating the experimental results of SAR images obtained using different algorithms in the embodiments of the present invention. Figure 8A and Figure 8B These are obtained by different algorithms in the embodiments of the present invention. Figure 7A A schematic diagram of the experimental results for the azimuth side lobes of the images in region 1 and region 2. Detailed Implementation

[0011] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0012] First, the technical background of this invention will be explained to facilitate understanding of the content of this invention.

[0013] I. Ideal SAR Imaging Model The model of ideal SAR imaging is as follows Figure 2 As shown, the radar moves relative to the target along a preset trajectory. In the range direction, it uses a fast time... Describe the sampling time interval, and in slow time. Describes the azimuth sampling time interval and the range sampling rate. The azimuth sampling rate (PRF) must be greater than their respective bandwidths to satisfy the Nyquist sampling theorem, so as to effectively sample the two-dimensional signal without ambiguity.

[0014] Radar in all directions slow time Location available This indicates that there are N scattering points in the scene. Then the slant range between the radar and each point, which varies with azimuth and time, can be defined as: (0-1) Based on the defined slant range, assuming the radar transmits a linear frequency modulated signal, the echo received by the radar after coherent detection can be represented as the signal at each point. Coherent superposition of echoes: (0-2) in, The range modulation window function represents the signal. This represents the azimuth modulation window function, which changes depending on the imaging mode. For the frequency modulation of the signal, The wavelength corresponding to the center carrier frequency.

[0015] In the range direction, SAR achieves high range resolution by transmitting a large-bandwidth modulated signal and then performing pulse compression. Range resolution is related to signal bandwidth. Related, represented as The distance pulse compression echo data is represented as follows: (0-3) The focusing position of each target in the range direction is related to its relative radar slant range. During the relative motion between the radar and the target in the azimuth direction, samples are taken at equal intervals to form an equivalent large antenna. By selecting a suitable imaging algorithm for effective coherent accumulation in the azimuth direction (such as RD, CS, RMA, PFA, BP, etc.), high azimuth resolution is obtained. Since the signal's azimuth bandwidth is based on the Doppler change, the SAR azimuth resolution is related to the azimuth observation bandwidth formed by the angular difference between the radar and the target in the azimuth direction. Related. Represented as The SAR image obtained after azimuth dimension processing can be represented as: (0-4) Different imaging algorithms may process azimuth data in different domains due to their characteristics, but the ideal resolution and focus position of the resulting image have the same correlation with the true distribution. For the sake of simplicity, this paper takes focusing on the azimuth time domain as an example and gives the expression for a SAR image, where... Representing a point target The azimuth focusing position corresponds to a slower time. Observing equation (0-4), it can be seen that the focusing position of each target in the processed ideal SAR image is related to its true two-dimensional distribution position, and is the position of each target after focusing. The superposition of functions enables them to be distinguished in SAR images, thus completing the observation of the scene.

[0016] Because range pulses are acquired in a shorter time, the probability of problems is generally lower. However, azimuth processing requires combining signals across the entire synthetic aperture time (SAP), resulting in a processing time span several orders of magnitude longer than that of range signals. This makes it more susceptible to interference from external non-ideal factors, such as platform motion errors and signal loss. SAR is based on phase-level processing, demanding extremely high accuracy; non-ideal SAR signals cannot yield effective imaging results. The impact of platform motion errors and their processing algorithms will be addressed in later chapters, while signal loss disrupts signal coherence, generating numerous high-energy false targets in the azimuth direction due to the grating lobe effect.

[0017] II. SAR Missing Signal Model To achieve sufficient azimuth resolution, SAR imaging typically requires a long synthetic aperture time and continuous signal transmission and reception. In practical applications, complex mission requirements, such as switching between multiple observation tasks, time-frequency synchronization, and interference from external signals, can temporarily occupy radar resources, causing some azimuth pulses of the SAR echo to be periodically lost or unavailable. In such cases, the received SAR echo will exhibit a periodic absence of certain SAR echoes. It can be represented as: (0-5) in, Let be the azimuth slow-time sampling function, used to represent missing echoes based on ideal echoes. Let the azimuth sampling time interval of the signal be . The period of loss of the echo azimuth pulse is That is, interval M The pulse loss phenomenon occurred repeatedly, and the number of lost pulses was [number missing]. K Let the first period in each cycle be... k The lost locations are , ,but It can be represented as: (0-6) To facilitate the description of the periodic changes in the sampling function, its duration is set here to... After sampling the ideal echo, the duration of the missing echo remains unchanged and consistent with the true value. Observe equation (0-6), its first term... For the ideal sampling function, the second term This is an additional sampling function applied to the signal after orientation loss. The two components are separable during imaging, and therefore can be analyzed later. The effect of periodic signal loss on imaging was obtained, and the first [signal] was taken separately. k The additional sampling function generated by the missing locations can be expanded using a Fourier series as follows: (0-7) in, , where is the pulse loss frequency, and the Fourier coefficients of each term are... It can be obtained through integration: (0-8) Substituting equation (0-8) into equations (0-6) and (0-7), and performing an azimuth FFT, we obtain... Frequency domain representation: (0-9) Based on equation (0-9), the frequency domain expression of the additional sampled signal applied to the ideal signal after periodic omission is given as follows: (0-10) Observing the above formula, we can see that in the additional sampled signal, the first term determines the accumulation location of the additional signal, and the second term is... K The frequency modulation function composed of the missing pulses only affects the amplitude of the additional signal. Therefore, after the missing pulses, for each scattering point, except for the ideal SAR image... Additional features were introduced, including accumulation in the azimuth direction. , The grating lobe energy at the specified location. The following simulation demonstrates the grating lobe effect caused by missing data to verify the above inference. The parameters used in the simulation are shown in Table 1: Table 1 Simulation parameters for periodic missing data raster lobe phenomenon

[0018] When there is only one point in the azimuth direction, the location of its grating lobe is as follows: Figure 3 As shown in (a), based on the derived lobe occurrence positions and the parameter list given in Table 1, the nearest lobe position is... At that time, it should be located 30 units to the left and right of the actual target. m Place. Figure 3 Figure (b) shows the grating lobe configuration when there are three point targets in the azimuth direction. It can be seen that the three actual targets and their respective grating lobes are marked with different colors. Under the current parameters, the positions of each target and its grating lobe are periodically interspersed. The distance between point target 2 and its nearest lobe can be obtained by using the distance between point target 1 and point target 2, i.e., a half-width of 50. m This verifies the correctness of the estimated grating lobe position.

[0019] from Figure 4 As can be seen, the energy of the ghost image of the strong scattering point grating lobe is extremely high. In practical applications, it will cover the original weak energy area and have a negative impact on the subsequent application of SAR images. In the future, the ability to suppress grating lobe energy will be used as a comparative standard to evaluate the recovery performance of azimuth missing signals.

[0020] Based on the aforementioned analysis of the grating lobe phenomenon, the periodic loss of azimuth signal has a significant impact on SAR images, necessitating effective processing methods. To address this issue, several missing signal recovery algorithms exist, such as the linear unbiased optimal interpolation algorithm, the Burg algorithm, the GAPES algorithm, and the MIAA algorithm. Table 2 compares the recovery performance and computation time of these four existing algorithms. Table 2. Analysis of the amplitude of false targets in point target simulation

[0021] The two-dimensional size of the echo in the simulation is 3200. The computation time of the above algorithms (3000) was based on processing on a workstation equipped with an Intel Xeon Gold 6234 CPU @ 3.30GHz and 128GB RAM. It can be seen that, except for the failure of the linear unbiased optimal interpolation algorithm, the other algorithms effectively suppressed the raster energy, proving the accuracy of the recovered periodically missing data. Among them, the Burg algorithm effectively suppressed raster energy while having a significantly shorter computation time than the GAPES and MIAA algorithms, making it the most suitable recovery algorithm for practical applications. However, when faced with larger data volumes, the Burg algorithm's computation time still becomes difficult to apply directly as the data volume increases proportionally.

[0022] Traditional Burg's algorithm recovers missing parts based on spectral estimation of one-dimensional signals. Therefore, in practical applications, it is necessary to recover the coefficients cell-by-cell. The estimation of azimuth pulses is complex. While the traditional Burg algorithm, as analyzed above, is the most computationally efficient algorithm for effectively recovering data and suppressing grating lobes, it still falls short of real-time requirements in practical applications. This is because the recovery correlation coefficients differ across range cells at each missing azimuth pulse location, requiring sequential estimation and recovery. Simply repeating the Burg algorithm fails to consider the two-dimensional coherence of SAR echoes. Although the missing data is in the azimuth direction, the range coherence and distribution structure of the SAR echoes can be utilized to help achieve more accurate and faster estimation and recovery of missing azimuth data.

[0023] First, in the ideal case of point simulation, the echo signal consists only of the coherent superposition of the feedback from point targets in the scene to the transmitted waveform. However, in practical applications, SAR echoes often contain interference factors such as noise. The model can be described as follows: (0-11) When the signal-to-noise ratio is poor, i.e., the noise power spectrum is... At higher energies, the echo power spectrum estimation will also be affected, as shown in equation (0-16). Change to Therefore, the Burg algorithm is highly sensitive to changes in signal-to-noise ratio (SNR), and the accuracy of power spectrum estimation drops sharply as the SNR decreases. Considering that the signal loss is only in the azimuth dimension, while SAR echoes are sampled in two dimensions, the coherence of the range echo is not fully utilized. Range pulse compression can significantly improve the signal-to-noise ratio. In the above equation, after range pulse compression, it can be expressed as: (0-12) After range pulse compression, the range energy of the signal undergoes coherent accumulation, while the incoherent noise and noise energy mismatched with the matched filter remain unchanged. Therefore, from the perspective of a certain range cell, the signal-to-noise ratio can be significantly improved. Subsequent simulations will compare the data recovery performance using the Burg algorithm under different noise levels, without considering SAR echo characteristics, with that after range pulse compression, to provide simulation verification of this derivation.

[0024] Secondly, the traditional method of recovering SAR signals using spectral estimation algorithms on a range-by-range basis does not take into account the continuity of the SAR signal in the range direction. In fact, if prior knowledge of the SAR signal structure can be derived, the recovery correlation coefficients of only a portion of the cells in the range direction can be estimated, and then, combined with the prior knowledge of the signal structure, the recovery correlation coefficients of the entire range direction can be generalized to obtain the recovery correlation coefficients.

[0025] Based on this, embodiments of the present invention provide a method for recovering periodically missing SAR signals based on prior knowledge of the echo structure. See also Figure 1 The method includes the following steps: S10. Obtain the raw echo signal obtained from the synthetic aperture radar sampling.

[0026] For example, the original echo signal contains a portion of the azimuth sampling signal, and there are missing sampling signals in some azimuth directions, which are the missing signals that are completed using the method of the present invention.

[0027] The sampling process of the original echo signal is as follows: Figure 5 As shown, in the distance direction, the interval between each sampling point is... In the azimuth direction, although the sampling time interval in the azimuth direction is However, due to the additional Doppler increment caused by the radar's motion relative to the scene during sampling, the azimuth interval between sampling points corresponding to each azimuth sampling time is an angle related to the Doppler increment. Each directional sampling point represents the current moment, and scene azimuth information centered on the current sampling point is collected on the corresponding distance cell. Therefore, the sampling structure of the original echo signal can be represented as: a fan-shaped region obtained by limiting the radius of concentric circles from near to far in the distance direction, and then limiting the angle range by the sampling start and end times in the azimuth direction.

[0028] In the sampling structure of the original echo signal, the azimuth sampling interval of each range cell is different. The slant range of the radars in each range cell is Then, within this distance cell, the azimuth radian interval between adjacent azimuth sampling points satisfies: (1-1) From the above formula, we can see It is only related to the distance cell where the sampling point is located, therefore the one-dimensional prediction relationship in the traditional Burg algorithm can be generalized to a two-dimensional prediction relationship of distance-azimuth: (1-2) Combining equations (1-1) and (1-2), let the distance sampling time interval be... If the distance cell containing equation (1-2) obtains an accurate estimated coefficient of restitution... For distance interval m The recovery relationship of a sampled distance cell based on the unlost signal (i.e., the sampled signal) can be expressed as: (1-3) In the formula, Indicates the first The slope distance corresponding to each distance unit Indicates the first The slant distance corresponding to each distance unit.

[0029] Therefore, theoretically, by accurately estimating the restitution coefficient of a certain range cell, and based on prior knowledge of the SAR echo structure, this can be extended to the entire range. However, in practice, due to the single... To assess the possibility of errors, the following steps can be used to recover the relationship across the entire distance segment.

[0030] S20. Based on the sampling structure of the original echo signal, the original echo signal is divided into blocks along the range direction to obtain multiple range blocks.

[0031] Each distance block includes multiple sampled signals.

[0032] Specifically, each range block comprises multiple range cells. A range cell refers to the smallest range interval at which a SAR system can distinguish two adjacent targets in the range direction, and is also the physical spatial range corresponding to a single sampling point in the fast time domain.

[0033] Optionally, step S20 may specifically include: S201. Perform range pulse compression processing on the original echo signal to obtain a preprocessed echo signal.

[0034] S202. Divide the preprocessed echo signal into blocks along the range direction to obtain multiple range blocks.

[0035] For example, the original echo signal is first subjected to range pulse compression to obtain a preprocessed echo signal, which can improve the signal-to-noise ratio and thus improve the accuracy of the restoration coefficient estimation. Then, for each missing aperture, it is divided along the range direction... L This yields multiple distance blocks.

[0036] S30. Determine the target range cell with the highest signal-to-noise ratio in each range block, and calculate the restoration coefficient of each target range cell.

[0037] For example, the energy after range pulse compression is concentrated, and within each range block, the signal-to-noise ratio (SNR) of each range cell is calculated. The range cell with the highest SNR is selected as the target range cell. Then, the target range cells are estimated. coefficient of recovery .

[0038] Optionally, the formula for calculating the restitution coefficient of each target distance cell is expressed as: (1-2) in, The restitution coefficient represents the distance cell of each target. This represents the echo after coherent radar detection. Indicates the signal to be recovered Given known signals at a distance of i azimuth units, where n represents the length of the known azimuth sequence used for signal recovery. Represents the ordinal number of a known signal. Indicates a fast time. Indicates slow time. Indicates the azimuth sampling time interval. express, express . conjugate.

[0039] Specifically, in equation (1-2), the upper equation represents extrapolation recovery from the left side of the missing position, and the lower equation represents extrapolation recovery from the right side of the missing position.

[0040] S40. Preprocess the restoration coefficients of each target distance cell to obtain the target restoration coefficients of the distance block where each target distance cell is located.

[0041] Optionally, step S40 may specifically include: S401. Remove the jump values ​​in the restoration coefficients of each target distance cell and replace them with the average value of the restoration coefficients of the target distance cells in the left and right adjacent distance blocks to obtain the preprocessed restoration coefficients of each target distance cell.

[0042] For example, after removing the jump values ​​from the restoration coefficients of each target distance cell, for the target distance cell corresponding to the jump value, the average value of the restoration coefficients of the target distance cells in the two adjacent distance blocks is calculated, and the average value is used to replace the jump value as the restoration coefficient of the corresponding target distance cell. After processing, the preprocessed restoration coefficients of each target distance cell can be obtained.

[0043] S402. Based on the radar slant range of the predetermined reference range cell and the radar slant range of each target range cell, the preprocessed recovery coefficients of each target range cell are normalized to obtain the target recovery coefficients of the range block in which each target range cell is located.

[0044] Optionally, the target restitution coefficient of the range block where each target range cell is located is expressed as: (1-4) in, Indicates the first The target recovery coefficient for each distance block, Indicates the first Target distance unit in each distance block radar slant range, This indicates the radar slant range corresponding to the reference range cell. Represents target distance unit The pretreatment recovery coefficient.

[0045] For example, based on prior knowledge of the SAR echo structure, the preprocessed restoration coefficients of each target range cell are normalized to the same reference range cell to obtain the target restoration coefficients of the range block in which each target range cell is located. That is, the normalized target restoration coefficients of each target range cell are used as the restoration coefficients of all range cells contained in its range block.

[0046] S50. Based on the target recovery coefficient of each distance block, the pre-determined target estimation error compensation parameters, and the original echo signal, determine the missing signal in the original echo signal.

[0047] Optionally, step S50 may specifically include: S501. Based on the least squares method, construct an objective function containing the target recovery coefficients of each range block, the sampled signals in the original echo signal, and the estimation error compensation parameters to be solved, and solve it to obtain the target estimation error compensation parameters.

[0048] Alternatively, the objective function can be expressed as:

[0049] in, Indicates the number of distance blocks. Indicates the first Target distance unit in each distance block The corresponding sampling signal, Indicates the first Target distance unit in each distance block radar slant range, This indicates the radar slant range corresponding to the reference range cell. Indicates the length of the known azimuth sequence used for signal recovery. Represents target distance unit The target recovery coefficient, Indicates the first The missing signal in each distance block The sampled signal corresponding to the distance unit that is i azimuth sampling intervals apart. Indicates the sampling time interval in the azimuth direction. and This represents the estimated error compensation parameter to be solved.

[0050] For example, It can be calculated using equation (1-2). and This represents the linear and constant terms in the estimated error compensation parameters to be solved, used to eliminate the estimation error of the recovery coefficient.

[0051] S502. Based on the target recovery coefficient of each distance block, the target estimation error compensation parameter, and the sampled signal in the original echo signal, determine the missing signal in the original echo signal.

[0052] Optionally, the missing signal in the original echo signal is represented as:

[0053] in, This indicates the missing signal in the original echo signal. Indication of missing signals Distance The sampling signal corresponding to the distance unit of each azimuth sampling interval. This represents the azimuth recovery coefficient of the range cell. and This represents the target estimation error compensation parameter.

[0054] This embodiment presents a SAR periodic missing signal recovery method based on prior knowledge of echo structure. It only requires estimating the recovery coefficients of some range cells and can be extended to the entire range. While ensuring pulse recovery accuracy, it significantly reduces the computation time. It can be applied to the rapid idealization processing of non-ideal SAR echoes with some pulse problems, avoiding the impact of interference factors such as grating lobes on the subsequent use of SAR images.

[0055] The following simulation experiment further illustrates the method for evaluating the skid resistance of asphalt pavement under aeolian sand conditions provided by this invention.

[0056] I. Simulation Experiment To verify the effectiveness of the proposed fast recovery algorithm for azimuth loss signals, a simulation experiment was conducted to verify the improvement of the algorithm performance by range pulse compression. The parameters used in the simulation are shown in Table 1 above.

[0057] By comparing the effect of the algorithm of this invention on suppressing grid lobes before and after pulse compression by adding 0dB, 5dB, 10dB and 15dB noise respectively, the stability of the algorithm of this invention is higher due to the improvement of signal-to-noise ratio after pulse compression, as shown in Table 3.

[0058] Table 3 Added Noise Levels

[0059] like Figure 6 As shown in Table 3, it can be seen that when there is no noise, the improvement of grating lobe suppression effect by range pulse compression is not significant. However, as the noise level increases, the grating lobe suppression capability of directly using the Burg algorithm based on the original echo decays rapidly. Comparing the effect of Burg after range pulse compression in this invention, it can be seen that due to the improvement of signal-to-noise ratio by pulse compression, the Burg algorithm after pulse compression has a significantly better grating lobe suppression effect than the traditional mode, and the difference in suppression effect increases with the increase of noise level. This also verifies the derivation of the aforementioned equation (0-12).

[0060] II. Processing of Measured Data Next, using the SAR measurement data acquired via airborne flight, 4 pulses were extracted from the azimuth data at 32-pulse intervals to make it equivalent to missing azimuth pulse data. The system parameters corresponding to the measurement data are shown in Table 4. The two-dimensional size of the measurement data is 12288. 32768. Due to the small grazing angle of the radar observation scene, some areas of the scene are blocked, resulting in more shadowed areas in the image. The grating lobe energy of strong scattering points is higher than that of other areas of the scene, making the grating lobe effect caused by the lack of orientation more obvious.

[0061] Table 4 System parameters corresponding to the processed SAR measured data

[0062] Figure 7AFigure 7 shows the SAR images obtained by directly zeroing out missing locations, restoring data cell-by-cell using the traditional Burg algorithm, restoring data using the present invention, and processing the original ideal data. The processed images clearly demonstrate that both the traditional Burg algorithm and the algorithm of this invention effectively suppress raster lobe energy in the images, and ghost images no longer exist in the scene.

[0063] To demonstrate the effectiveness of the algorithm in suppressing gate lobes, we take... Figure 7A The two segments shown are analyzed for the most severe areas of the grating lobe phenomenon. Figure 8A and Figure 8B The sidelobe patterns of SAR images in regions 1 and 2 are presented sequentially using zero-padding of missing data, original data without missing data, the traditional Burg algorithm, and the algorithm of this invention. As can be seen... Figure 8A (d) Figure 8B As shown in (d), the algorithm of the present invention effectively and significantly suppresses the raster lobe energy in the image. Compared with the original data without missing data and the traditional Burg algorithm, the algorithm of the present invention introduces low-energy noise in the azimuth direction after compensating for missing data, but does not affect the imaging results.

[0064] Table 5 provides the following information: Figure 8A and Figure 8B The results of grating lobe suppression and computation time of different algorithms are presented. The platform used is the same as described above. It can be seen that the present invention saves a lot of computation time while maintaining the accuracy of data recovery. Compared with the 2973.60 seconds required by the traditional Burg algorithm, the present invention only requires 128.30 seconds. For reference, the actual measurement data took 260.15 seconds to image using the RMA algorithm. Therefore, for real-time imaging application scenarios with sufficient computing power, the present invention can adapt to its application requirements and quickly recover missing data.

[0065] Table 5. Energy and time consumption for recovering data grating lobes using different algorithms.

[0066] It should be noted that the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention.

[0067] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0068] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings and the disclosure in carrying out the claimed invention. In the description of the invention, the word "comprising" does not exclude other components or steps, "a" or "an" does not exclude a plurality, and "a plurality" means two or more, unless otherwise explicitly specified. Furthermore, while different embodiments may describe certain measures, this does not mean that these measures cannot be combined to produce good results.

[0069] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for recovering periodically missing SAR signals based on prior knowledge of echo structure, characterized in that, include: Acquire the raw echo signal obtained from synthetic aperture radar sampling; Based on the sampling structure of the original echo signal, the original echo signal is divided into blocks along the range direction to obtain multiple range blocks; each range block includes multiple sampled signals; Identify the target range cell with the highest signal-to-noise ratio in each range block and calculate the recovery coefficient of each target range cell; The recovery coefficients of each target distance unit are preprocessed to obtain the target recovery coefficients of the distance block where each target distance unit is located; Based on the target recovery coefficient of each distance block, the pre-determined target estimation error compensation parameter, and the original echo signal, the missing signal in the original echo signal is determined.

2. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 1, characterized in that, The step of determining the missing signal in the original echo signal based on the target recovery coefficient of each distance block, the pre-determined target estimation error compensation parameter, and the original echo signal includes: Based on the least squares method, an objective function is constructed that includes the target recovery coefficient of each range block, the sampled signal in the original echo signal, and the estimation error compensation parameter to be solved, and then solved to obtain the target estimation error compensation parameter; Based on the target recovery coefficient of each distance block, the target estimation error compensation parameter, and the sampled signal in the original echo signal, the missing signal in the original echo signal is determined.

3. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 1, characterized in that, The preprocessing of the recovery coefficients of each target distance unit to obtain the target recovery coefficients of each target distance unit includes: The jump values ​​in the recovery coefficients of each target distance unit are removed, and the average value of the recovery coefficients of the target distance units in the left and right adjacent distance blocks is used to replace them to obtain the preprocessed recovery coefficients of each target distance unit. Based on the radar slant range of the predetermined reference range cell and the radar slant range of each target range cell, the preprocessed recovery coefficients of each target range cell are normalized to obtain the target recovery coefficients of the range block in which each target range cell is located.

4. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 1, characterized in that, The sampling structure based on the original echo signal divides the original echo signal into multiple range blocks along the range direction, including: The original echo signal is subjected to range pulse compression processing to obtain a preprocessed echo signal; Based on the sampling structure of the original echo signal, the preprocessed echo signal is divided into blocks along the range direction to obtain multiple range blocks.

5. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 1, characterized in that, The formula for calculating the restitution coefficient of each target range cell is as follows: in, This represents the restitution coefficient of each target distance cell. This represents the echo after coherent radar detection. Indication of missing signals Distance The sampling signal corresponding to the distance unit of each azimuth sampling interval. This indicates the length of the known azimuth sequence used for signal recovery. Represents the ordinal number of a known signal. Indicates a fast time. Indicates slow time. Indicates the azimuth sampling time interval. express . conjugate.

6. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 2, characterized in that, The objective function is expressed as: in, Indicates the number of distance blocks. Indicates the first Target distance unit in each distance block The corresponding sampling signal, Indicates the first Target distance unit in each distance block radar slant range, This indicates the radar slant range corresponding to the reference range cell. This indicates the length of the known azimuth sequence used for signal recovery. Represents target distance unit The target recovery coefficient, Indicates the first The missing signal in each distance block Distance The sampling signal corresponding to the distance unit of each azimuth sampling interval. Indicates the sampling time interval in the azimuth direction. and This represents the estimated error compensation parameter to be solved.

7. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 6, characterized in that, The missing signal in the original echo signal is represented as follows: in, This indicates the missing signal in the original echo signal. Indication of missing signals Distance The sampling signal corresponding to the distance unit of each azimuth sampling interval. This represents the azimuth recovery coefficient of the range cell. and This represents the target estimation error compensation parameter.

8. The SAR periodic missing signal recovery method based on prior knowledge of echo structure according to claim 3, characterized in that, The target recovery coefficient of each target range cell in the range block is expressed as: in, Indicates the first The target recovery coefficient for each distance block, Indicates the first Target distance unit in each distance block radar slant range, This indicates the radar slant range corresponding to the reference range cell. Represents target distance unit The pretreatment recovery coefficient.

Images