Priori time-frequency information assisted satellite-borne synthetic aperture radar broadband interference suppression method and system, and storage medium

Abstract

The invention discloses a priori time-frequency information assisted satellite-borne synthetic aperture radar broadband interference suppression method and system and a storage medium, and the method comprises the steps: selecting a reference pulse, and constructing a time-frequency domain local amplitude feature through short-time Fourier transform; based on the reference pulse, an absolute median difference method is adopted to extract the frequency band and duration of interference as prior information; time-frequency sub-matrixes are divided for interference-containing pulses, strong and weak scattering areas are distinguished, K-means clustering is carried out respectively, and interference is distinguished preliminarily; the preliminary result is restrained and corrected in combination with prior information, and the integrity and consistency of interference positioning under a complex background are improved; and finally, block subspace filtering is adopted in the interference area to suppress interference components. According to the invention, by introducing the interference prior and time-frequency division mechanism, the method adapts to different scenes, reduces the damage to useful signals while effectively suppressing the broadband radio frequency interference, and facilitates the improvement of the imaging quality.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a method and system for broadband interference suppression of spaceborne synthetic aperture radar assisted by prior time and frequency information. Background Technology

[0002] Synthetic Aperture Radar (SAR) is an active microwave remote sensing device that detects ground targets by emitting microwave signals and receiving reflected waves. It generates a virtual large aperture based on the movement of the radar platform and combines this with pulse compression technology to achieve high-resolution imaging. SAR signals are unaffected by day-night cycles and inclement weather, and can penetrate obstacles such as clouds, rain, and fog, thus providing all-day, all-weather observation capabilities.

[0003] With the rapid advancement of modern communication technology, spectrum resources are becoming increasingly scarce. As an open radar system, SAR shares the same frequency band with other devices, making its electromagnetic environment more complex and exacerbating radio frequency interference (RFI). RFI can cause disturbances in the impulse response, leading to problems such as bright lines, fog artifacts, and image blurring in imaging. Simultaneously, RFI can also cause errors in spatial and radiometric measurements, as well as distortions in polarization and phase, thus affecting the application of SAR images. From the perspective of signal bandwidth, radio frequency interference can be divided into narrowband and wideband. With the development of modern communication technology, the bandwidth of unintentional ground-based interference sources is increasing, greatly increasing the likelihood of wideband radio frequency interference. Furthermore, wideband interference covers most of the signal spectrum, making it difficult to distinguish between interference and the signal. Traditional suppression methods easily damage the useful signal, making suppression even more challenging.

[0004] Processing SAR echoes with broadband interference to protect useful target signals is an important research direction in the face of today's complex electromagnetic environment. To address the interference problems suffered by SAR systems, since the 1990s, scholars both domestically and internationally have conducted in-depth research on broadband interference suppression for SAR, proposing various algorithms, which can be broadly categorized as follows: parametric, semi-parametric, and non-parametric methods.

[0005] The main principle of parameterized methods is to establish an accurate mathematical model of the interference signal, and then extract the interference signal from the SAR echo data to achieve the suppression effect. For example, maximum likelihood estimation, least squares estimation, and parameter maximum likelihood and minimum mean square estimation algorithms are typical parameterized interference suppression methods. Liu Zhiling et al. proposed using the Iterative Adaptive Approach (IAA) to estimate the model parameters of narrowband interference, obtaining interference suppression results that balance speed and efficiency. Yang Zhiwei et al., combining time-frequency analysis techniques, extended the IAA narrowband interference suppression method to make it applicable to broadband interference suppression.

[0006] Semi-parametric methods transform complex signal separation problems into hyperparameter optimization problems for solution. Nguyen et al., starting from the theory of sparse reconstruction and low-rank decomposition, pioneered the concept of semi-parametric interference suppression, achieving narrowband interference suppression by solving the sparse reconstruction optimization problem. Liu Hongqing et al. extended the sparse reconstruction algorithm to broadband interference, providing new ideas for the extended application of this type of method. Huang Yan et al. proposed an adaptive notch filter semi-parametric (ANSP) method, combining the advantages of semi-parametric methods and frequency domain notch filters (FNF), achieving good suppression results. Semi-parametric methods utilize optimization models to constrain interference, which can better protect useful signals; however, their performance depends heavily on the values ​​of hyperparameters and the selection of the optimization model, and the computational load is much larger than the other two types of methods.

[0007] Non-parametric methods directly apply the energy characteristics of the interference signal for suppression, without needing to establish a parametric model of the interference. Interference extraction is achieved through filter design or subspace construction, making it easy to implement and highly applicable. However, the filtering process can easily lead to the loss of useful data. Classic non-parametric methods include notch filtering, adaptive filtering, and subspace projection algorithms. The main principle of notch filtering is to set nulls in the interference region to eliminate the interference signal. The main principle of adaptive filtering is to construct an adaptive filter in a suitable domain to separate the useful signal from the interference; it is still more suitable for narrowband interference. Subspace projection can better handle both narrowband and broadband interference and effectively protects the useful signal, but it is limited by a fixed threshold, and its suppression performance still has considerable room for improvement. Summary of the Invention

[0008] The purpose of this invention is to provide a method, system, and storage medium for suppressing broadband interference in spaceborne synthetic aperture radar (SAR) with prior time-frequency information assistance, which can obtain high-quality SAR images and effectively suppress broadband interference to the SAR system.

[0009] The technical solution adopted in this invention is: a broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time and frequency information, comprising the following steps:

[0010] Step S101: Perform a distance-to-Fourier transform on each pulse of the interference echo data, calculate the spectral flatness, sort the pulses according to the magnitude of the spectral flatness, and select the pulses with the lowest spectral flatness as reference pulses.

[0011] Step S102: Based on the reference pulse, perform a short-time Fourier transform on it, divide it into several sub-matrices in the time-frequency domain, and construct local amplitude features in the time-frequency domain.

[0012] Step S103: Based on the time-frequency local amplitude characteristics of the pulse, locate the interference region using the absolute median difference method, extract the occupied frequency band and duration of the interference in the time-frequency domain, and construct prior information about the interference.

[0013] Step S104: Based on the method described in step S102, the time-frequency representation of the interference pulse is divided into sub-matrices and the local amplitude characteristics are calculated. The absolute median difference method is used to divide the time-frequency representation of the pulse into a strong scattering region and a weak scattering region.

[0014] Step S105: Perform K-means clustering on the local amplitude features in the time-frequency domain in the strong scattering region and the weak scattering region respectively to achieve a preliminary distinction between the interference region and the non-interference region, and obtain the initial localization result of the time-frequency domain interference.

[0015] Step S106: Combine the interference prior information constructed in step S103 to constrain and correct the initial interference localization result; at the same time, at the boundary between the strong scattering region and the weak scattering region, supplement and correct the time-frequency connectivity structure that satisfies the interference prior conditions to obtain the final time-frequency domain interference localization result.

[0016] Step S107: For the finally determined time-frequency domain interference location area, use the block subspace filtering method to perform interference suppression processing, and output the interference-suppressed echo data as the final interference suppression result.

[0017] In step S101, a range-to-Fourier transform is performed on each pulse of the interference-containing data, and the spectral flatness is calculated. Several pulses with the lowest spectral flatness are selected as reference pulses to extract broadband interference components with prominent interference characteristics. Specifically:

[0018] To measure the uniformity of the energy distribution in the echo spectrum, let the distance direction be... The number of azimuth sampling points is For each pulse Perform a Fourier transform to obtain its frequency domain representation:

[0019]

[0020] in, Indicates the first A time-domain sampling sequence of pulses in the range direction. For distance-oriented sampling point index; Represents the Discrete Fourier Transform operator; For the corresponding frequency domain complex spectral components, Represent the frequency index and calculate its amplitude spectrum:

[0021]

[0022] in, Indicates the first The pulse in the first The amplitude values ​​at each frequency point are used to calculate the spectral flatness (SF), which is defined as the ratio of the geometric mean to the arithmetic mean of the amplitude spectrum.

[0023]

[0024] Spectral flatness characterizes the uniformity of spectral energy distribution: when the echo spectrum contains only useful scattered signals and noise, its energy is usually uniformly distributed within the frequency band, resulting in a large flatness value; when broadband interference exists in the echo, the interference energy is often distributed within a frequency band, causing the amplitude of the spectral portion to rise, thus leading to a decrease in spectral flatness. The spectral flatness corresponding to the interfering pulse is then calculated. Sort the data and select the smallest value. A set of pulses serves as a reference pulse. These reference pulses typically contain more pronounced broadband interference characteristics, with more prominent frequency domain occupancy and time duration characteristics, making them suitable for extracting prior interference information.

[0025] In step S102, let the reference pulse set obtained in step S101 be... ,in Indicates the first The time-domain echo signal of the reference pulse in the range direction, The number of reference pulses is given. Time-frequency analysis is performed on each of the reference pulses. In this embodiment, a short-time Fourier transform is used to process the reference pulses, and its time-frequency representation can be expressed as:

[0026]

[0027] in, To analyze window functions, and These represent the time index and frequency index, respectively. The amplitude value is taken from the time-frequency representation result to obtain the corresponding time-frequency domain amplitude distribution. It is used to characterize the distribution of signal energy in the time and frequency dimensions.

[0028] After obtaining the time-frequency domain amplitude distribution, the time-frequency representation result is divided within the time-frequency plane according to a preset time scale and frequency scale, dividing the entire time-frequency plane into several non-overlapping time-frequency sub-matrices. Each time-frequency sub-matrix corresponds to a local region in the time-frequency plane, used to describe the local energy characteristics of echoes and interference within that region. Let the first... The index range of each time-frequency submatrix in the time dimension is: The index range in the frequency dimension is Then the first The summation of the amplitudes within each time-frequency submatrix can be expressed as:

[0029]

[0030] The statistics, as local amplitude features in the time-frequency domain, are used to characterize the overall level of energy distribution within that time-frequency region.

[0031] In step S103, based on the local amplitude features corresponding to all time-frequency sub-matrices... The absolute median difference method was used to divide the region into interference and non-interference regions. First, the median of the local amplitude characteristics was calculated.

[0032]

[0033] And based on this, the absolute median difference is calculated, and its expression is:

[0034]

[0035] The segmentation threshold was then determined and set as follows:

[0036]

[0037] in Using empirical coefficients, submatrices with amplitudes exceeding a set threshold are identified as interference regions, while the rest are considered non-interference regions. Based on the obtained interference regions, their frequency-dimensional occupancy can be represented as intervals.

[0038]

[0039] in Indicates the first The starting frequency position of the interference band in the reference pulse. In the time dimension, the corresponding duration of a single interference event can be expressed as...

[0040]

[0041] in and These represent the start and end times of the interference in the time and frequency domain, respectively.

[0042] To improve the robustness of the statistical results, the set of starting positions of the frequency band is... and duration set Outlier handling is performed. In this embodiment, an outlier detection method based on the median is used, that is, for any set of parameters... Calculate its absolute median and remove those that meet the following conditions.

[0043]

[0044] Abnormal samples, among which This is an empirical coefficient.

[0045] After outlier removal, statistical fusion is performed on the remaining valid samples to determine the final prior parameters of the interference. In this embodiment, the starting position of the interference frequency band is taken as its average value, and the duration of a single interference event is also statistically analyzed in the same way to obtain...

[0046]

[0047]

[0048] in This indicates the number of valid samples remaining after outlier removal. Through the above processing, the number of samples including those occupying interference frequency bands is obtained. and the duration of a single interference Interference prior information.

[0049] In step S104, let the data containing interference echoes be... ,in Indicates the azimuth sampling point. This represents the distance-oriented sampling point. Based on the method described in step S102, a short-time Fourier transform is performed pulse by pulse, and the amplitude of the time-frequency representation result is taken to obtain the corresponding time-frequency domain amplitude distribution. Then, the local amplitude features in the time-frequency domain are constructed. Let the first... The pulse in the first The summation of the amplitudes within each time-frequency submatrix is: ,in Indicates the index of the time-frequency submatrix. This represents the total number of time-frequency submatrices. For the same interfering pulse, the sum of the amplitudes corresponding to all time-frequency submatrices forms a local amplitude feature sequence. To distinguish the scattering intensity corresponding to different time-frequency regions, the local amplitude feature sequence is statistically analyzed using the absolute median difference method. When the amplitude sum corresponding to a certain time-frequency submatrix is ​​greater than the threshold range determined by the absolute median difference, the time-frequency submatrix is ​​identified as a strong scattering region; the rest are identified as weak scattering regions. Through the above processing, adaptive division of scattering intensity at the time-frequency submatrix scale is achieved, thus providing a basis for subsequent interference localization in different scattering regions.

[0050] In step S105, it is assumed that step S104 has already divided the time-frequency sub-matrix set into a strong scattering sub-matrix set. With the set of weak scattering submatrices For any pulse containing interference Take its corresponding local amplitude sum in each time-frequency submatrix. As local amplitude features in the time-frequency domain, sample sequences composed of these local amplitude features are constructed within the sets of strong and weak scattering sub-matrices, respectively. K-means clustering is then used to automatically classify these sample sequences within each scattering region. In this embodiment, the number of clusters is set to 2 to divide the time-frequency sub-matrices into interference and non-interference classes. The objective function of K-means clustering is defined as minimizing the sum of the squared Euclidean distances between a sample and its cluster center, i.e.

[0051]

[0052] in Indicates the first A cluster, This represents the center value of the corresponding cluster, calculated as the mean of all samples within that cluster. During the clustering iteration process, the objective function is gradually reduced until convergence by alternately performing sample assignment and cluster center update operations. Specifically, for any local amplitude feature sample... It is assigned to the cluster corresponding to the cluster center with the smallest Euclidean distance to it, that is...

[0053]

[0054] After clustering is completed, the clustering results are judged based on the mean of the local amplitude features corresponding to each cluster. Clusters with larger mean local amplitude features are judged as interference candidate regions, while clusters with smaller mean local amplitude features are judged as non-interference regions.

[0055] In step S106, the initial time-frequency domain interference localization result obtained in step S105 is constrained and corrected by combining the interference prior information constructed in step S103; at the same time, the time-frequency connectivity structure that satisfies the interference prior conditions is supplemented and corrected at the boundary between the strong scattering region and the weak scattering region to obtain the final time-frequency domain interference localization result.

[0056] Suppose that the initial time-frequency domain interference localization result obtained in step S105 can be expressed as a binary indicator function. The value is 1 when the corresponding time-frequency submatrix is ​​determined to be an interference region, and 0 otherwise. The interference prior information constructed in step S103 includes the frequency band occupied by the interference in the frequency dimension. and the duration range of a single interference. Based on the prior information about the interference, a consistency constraint is applied to the initial positioning results. That is, only time-frequency regions that simultaneously satisfy the condition that the frequency position falls within the frequency band occupied by the interference and the duration of the continuous time matches the characteristics of a single interference event are retained, thereby eliminating false alarms caused by noise or strong scattering points.

[0057] During the constraint process, when a certain time-frequency region satisfies

[0058]

[0059] If a region is connected to an identified interference region in the time-frequency domain, it is confirmed as an interference region; otherwise, time-frequency regions that do not meet the prior conditions for interference or are only isolated are removed from the interference localization results.

[0060] Furthermore, at the boundary between strong and weak scattering regions, due to abrupt changes in background energy distribution, the interference structure may appear discontinuous or partially missing in the initial localization results. To address this, the time-frequency connectivity structure located at the boundary between strong and weak scattering regions is analyzed. When it meets the aforementioned interference prior conditions in terms of frequency range and temporal duration, the corresponding time-frequency region is supplemented and marked; for connectivity structures that significantly deviate from the interference prior characteristics, they are corrected or deleted.

[0061] In step S107, the interference time-frequency localization result determined in step S106 is assumed to correspond to a local data matrix in the pulse time-domain representation, denoted as...

[0062]

[0063] in The bandwidth occupied by the interference area. For the duration of the interference, It includes useful echo components, interference components, and noise components. A sample covariance matrix is ​​constructed for the data block.

[0064]

[0065] Performing eigenvalue decomposition on the covariance matrix, we have:

[0066]

[0067] in Let be an eigenvalue matrix, and satisfy... , This is the corresponding eigenvector matrix.

[0068] Based on the interference localization results and the characteristics of feature value distribution, the feature vectors corresponding to larger feature values ​​are identified as the interference subspace, which can be represented as follows:

[0069]

[0070] in Let be the dimension of the interference subspace. Based on this, construct the orthogonal projection matrix of the interference subspace.

[0071]

[0072] in Accordingly, the projection matrix of the non-interference subspace can be expressed as:

[0073] in It is an identity matrix.

[0074] The data block is filtered using the non-interference subspace projection matrix to obtain interference-suppressed echo data.

[0075]

[0076] By performing the block subspace filtering process described above, interference components are projected from the original data blocks and suppressed, while useful echo components in the non-interference subspace are retained, thereby effectively suppressing the location interference area. The final interference suppression result can be obtained by performing the aforementioned processing on the data blocks corresponding to each interference area and reconstructing the echo data.

[0077] A broadband interference suppression system for spaceborne synthetic aperture radar assisted by prior time-frequency information includes:

[0078] The interference prior information construction unit is configured to perform Fourier transform on the interference pulse and calculate the spectral flatness, select several pulses with the lowest spectral flatness as reference pulses; then perform short-time Fourier transform on the reference pulses to locate the interference region, extract the frequency occupancy range of the interference and the duration of a single interference, and construct the interference prior information.

[0079] The time-frequency transformation and local partitioning unit is configured to perform short-time Fourier transform on the interference echo to be processed to obtain the energy distribution of the echo signal in the time-frequency domain; and to partition the time-frequency representation result in the time-frequency domain to form several time-frequency sub-matrices, and to statistically analyze the amplitude information of each time-frequency sub-matrices to construct local amplitude features in the time-frequency domain.

[0080] The time-frequency region division unit is configured to analyze the amplitude distribution of each time-frequency sub-matrix based on the local amplitude characteristics of the time-frequency domain using the absolute median difference method, dividing the time-frequency domain into strong scattering regions and weak scattering regions to distinguish echo characteristics under different scattering conditions.

[0081] The partitioned interference localization unit is configured to perform cluster analysis on the local amplitude features in the time-frequency domain in the strong scattering region and the weak scattering region respectively, so as to achieve preliminary segmentation of the interference region and the non-interference region and obtain the initial localization result of the time-frequency domain interference.

[0082] The prior constraint correction unit is configured to combine the interference prior information to constrain and correct the initial interference localization result; at the boundary between the strong scattering region and the weak scattering region, it supplements and corrects the time-frequency connectivity structure that satisfies the interference prior conditions to obtain the final time-frequency domain interference localization result.

[0083] The interference suppression unit is configured to perform interference suppression processing using a block subspace filtering method for the finally determined time-frequency domain interference location area, so as to suppress broadband interference components and retain useful echo signals.

[0084] The pulse combining unit is configured to sequentially combine the echo data after interference suppression processing to obtain the final interference suppression result for subsequent imaging processing.

[0085] A storage device storing multiple programs, which, when loaded and executed by a processor, are used to implement the prior time-frequency information-assisted broadband interference suppression method for spaceborne synthetic aperture radar as described above.

[0086] An electronic device includes a storage device and a processor; the processor is adapted to execute various programs; the memory is adapted to store multiple programs; when the memory executes the programs on the processor, it implements the aforementioned a priori time-frequency information-assisted broadband interference suppression method for spaceborne synthetic aperture radar.

[0087] This invention performs range-to-Fourier transform on pulse-by-pulse synthetic aperture radar (SAR) echo data containing interference to calculate spectral flatness, selecting several echoes with the lowest spectral flatness as reference pulses. It then performs short-time Fourier transform on the interference pulses, generating several sub-matrices in the time-frequency domain and constructing local amplitude features in the time-frequency domain. Based on the local amplitude features of the reference pulses, it uses the absolute median difference method to locate the interference region, extracting the occupied frequency band and duration of the interference in the time-frequency domain to construct prior interference information. Finally, it divides the interference pulses into time-frequency sub-matrices using the same time-frequency division method as the reference pulses and calculates... The invention calculates local amplitude features and uses the absolute median difference method to divide the time-frequency representation of the pulse into strong scattering and weak scattering regions. Within each region, K-means clustering is performed on the local amplitude features in the time-frequency domain to initially distinguish between interfering and non-interfering regions, obtaining initial interference localization results. Interference prior information is then used to constrain and correct the initial interference localization results, improving the completeness and consistency of interference localization results under complex background conditions. Finally, a block subspace filtering method is used to suppress interference components within the finally determined interference region, obtaining interference-suppressed echo data. Compared with existing technologies, this invention can effectively suppress broadband radio frequency interference in synthetic aperture radar echoes under complex background conditions, avoiding the problem of insufficient adaptability of a uniform threshold in different scenarios. By introducing interference prior constraints and a time-frequency divide-and-conquer mechanism, the accuracy of interference suppression is improved while reducing damage to useful echo signals, thereby improving subsequent imaging quality. Attached Figure Description

[0088] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0089] Figure 1 This is a schematic diagram of the process of the present invention;

[0090] Figure 2 Image features of broadband jamming signals for synthetic aperture radar provided in embodiments of the present invention;

[0091] Figure 3 The multi-domain features of the synthetic aperture radar broadband jamming signal provided in the embodiments of the present invention;

[0092] Figure 4 This is a block diagram illustrating the system structure principle of the present invention;

[0093] Figure 5 This is a comparative experimental result of the present invention and existing technology on Sentinel-1 image data. Detailed Implementation

[0094] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0095] Exemplary methods

[0096] like Figure 1 As shown, the present invention includes the following steps:

[0097] Step S101: Calculate the spectral flatness of the data containing interference echo pulse by pulse, and select several pulses with the lowest flatness as reference pulses;

[0098] To measure the uniformity of the energy distribution in the echo spectrum, let and These represent the number of sampling points in the range and azimuth directions, respectively, for each pulse. Perform a Fourier transform to obtain its frequency domain representation.

[0099]

[0100] in, Indicates the first A time-domain sampling sequence of echo pulses in the range direction. For distance-oriented sampling point index; Represents the Discrete Fourier Transform operator; For the corresponding frequency domain complex spectral components, Represent the frequency index and calculate its amplitude spectrum:

[0101]

[0102] in, Indicates the first The pulse in the first The amplitude values ​​at each frequency point are used to calculate the spectral flatness (SF), which is defined as the ratio of the geometric mean to the arithmetic mean of the amplitude spectrum.

[0103]

[0104] Spectral flatness characterizes the uniformity of spectral energy distribution: when the echo spectrum contains only useful scattered signals and noise, its energy is usually uniformly distributed within the frequency band, and the flatness value is relatively large; for example... Figure 3As shown, when broadband interference exists in the echo, the interference energy is often distributed within a frequency band, causing the amplitude of the spectral portion to rise, thus leading to a decrease in spectral flatness. The spectral flatness corresponding to the interfering pulse is then calculated. Sort the data and select the smallest value. A set of pulses serves as a reference pulse. These reference pulses typically contain more pronounced broadband interference characteristics, with more prominent frequency domain occupancy and time duration characteristics, making them suitable for extracting prior interference information.

[0105] Step S102: Based on the reference pulse, perform a short-time Fourier transform on it, divide it into several sub-matrices in the time-frequency domain, and construct local amplitude features in the time-frequency domain.

[0106] Specifically, let the set of reference pulses obtained in step S101 be... ,in Indicates the first The time-domain echo signal of the reference pulse in the range direction, The number of reference pulses is given. Time-frequency analysis is performed on each of the reference pulses. In this embodiment, a short-time Fourier transform is used to process the reference pulses, and its time-frequency representation can be expressed as:

[0107]

[0108] in, To analyze window functions, and These represent the time index and frequency index, respectively. The amplitude value is taken from the time-frequency representation result to obtain the corresponding time-frequency domain amplitude distribution. It is used to characterize the distribution of signal energy in the time and frequency dimensions.

[0109] After obtaining the time-frequency domain amplitude distribution, the time-frequency representation result is divided within the time-frequency plane according to a preset time scale and frequency scale, dividing the entire time-frequency plane into several non-overlapping time-frequency sub-matrices. Each time-frequency sub-matrix corresponds to a local region in the time-frequency plane, used to describe the local energy characteristics of echoes and interference within that region. Let the first... The index range of each time-frequency submatrix in the time dimension is: The index range in the frequency dimension is Then the first The summation of the amplitudes within each time-frequency submatrix can be expressed as:

[0110]

[0111] The statistics, as local amplitude features in the time-frequency domain, are used to characterize the overall level of energy distribution within that time-frequency region.

[0112] Step S103: Based on the time-frequency local amplitude characteristics of the pulse, locate the interference region using the absolute median difference method, extract the occupied frequency band and duration of the interference in the time-frequency domain, and construct prior information about the interference.

[0113] Specifically, based on the local amplitude features corresponding to all time-frequency sub-matrices The absolute median difference method was used to divide the region into interference and non-interference regions. First, the median of the local amplitude characteristics was calculated.

[0114]

[0115] And based on this, the absolute median difference is calculated, and its expression is:

[0116]

[0117] The segmentation threshold was then determined and set as follows:

[0118]

[0119] in Using empirical coefficients, submatrices with amplitudes exceeding a set threshold are identified as interference regions, while the rest are considered non-interference regions. Based on the obtained interference regions, their frequency-dimensional occupancy can be represented as intervals.

[0120]

[0121] in Indicates the first The starting frequency position of the interference band in the reference pulse. In the time dimension, the corresponding duration of a single interference event can be expressed as...

[0122]

[0123] in and These represent the start and end times of the interference in the time and frequency domain, respectively.

[0124] To improve the robustness of the statistical results, the set of starting positions of the frequency band is... and duration set Outlier handling is performed. In this embodiment, an outlier detection method based on the median is used, that is, for any set of parameters... Calculate its absolute median and remove those that meet the following conditions.

[0125]

[0126] Abnormal samples, among which This is an empirical coefficient.

[0127] After outlier removal, statistical fusion is performed on the remaining valid samples to determine the final prior parameters of the interference. In this embodiment, the starting position of the interference frequency band is taken as its average value, and the duration of a single interference event is also statistically analyzed in the same way to obtain...

[0128]

[0129]

[0130] in This indicates the number of valid samples remaining after outlier removal. Through the above processing, the number of samples including those occupying interference frequency bands is obtained. and the duration of a single interference Interference prior information.

[0131] Step S104: Based on the method described in step S102, divide the time-frequency sub-matrix pulse by pulse and calculate the local amplitude characteristics. Use the absolute median difference method to divide the time-frequency representation of the pulse into strong scattering region and weak scattering region to characterize the difference characteristics of echo and interference under different scattering conditions.

[0132] like Figure 3 The interference signal characteristics shown exhibit significant differences in the pulse time-frequency representation of the echo received by spaceborne synthetic aperture radar under complex scene conditions, occurring across different range cells and backgrounds. This manifests as the simultaneous existence of strong and weak scattering regions in the time-frequency domain. Within the weak scattering region, the energy level of the interference component may be significantly lower than the useful echo energy in the strong scattering region, causing the interference characteristics to be masked by the strong scattering echo on a global scale. Most existing interference detection and suppression methods typically use a globally uniform threshold to determine time-frequency domain characteristics. However, under the complex background conditions of coexisting strong and weak scattering, it is difficult to simultaneously address the discrimination requirements of different background regions. This can easily lead to missed interference detection in weak scattering regions or false suppression of useful signals in strong scattering regions, resulting in a significant degrade in method performance.

[0133] Step S105: Perform K-means clustering analysis on the local amplitude features in the time-frequency domain in the strong scattering region and the weak scattering region respectively to achieve a preliminary distinction between the interference region and the non-interference region, and obtain the initial location result of the time-frequency domain interference.

[0134] In step S105, it is assumed that step S104 has already divided the time-frequency sub-matrix set into a strong scattering sub-matrix set. With the set of weak scattering submatrices For any pulse containing interference Take its corresponding local amplitude sum in each time-frequency submatrix. As local amplitude features in the time-frequency domain, sample sequences composed of these local amplitude features are constructed within the sets of strong and weak scattering sub-matrices, respectively. K-means clustering is then used to automatically classify these sample sequences within each scattering region. In this embodiment, the number of clusters is set to 2 to divide the time-frequency sub-matrices into interference and non-interference classes. The objective function of K-means clustering is defined as minimizing the sum of the squared Euclidean distances between a sample and its cluster center, i.e.

[0135]

[0136] in Indicates the first A cluster, This represents the center value of the corresponding cluster, calculated as the mean of all samples within that cluster. During the clustering iteration process, the objective function is gradually reduced until convergence by alternately performing sample assignment and cluster center update operations. Specifically, for any local amplitude feature sample... It is assigned to the cluster corresponding to the cluster center with the smallest Euclidean distance to it, that is...

[0137]

[0138] After clustering is completed, the clustering results are judged based on the mean of the local amplitude features corresponding to each cluster. Clusters with larger mean local amplitude features are judged as interfering candidate regions, while clusters with smaller mean local amplitude features are judged as non-interfering regions.

[0139] Step S106: Combine the interference prior information constructed in step S103 to constrain and correct the initial interference localization result; at the same time, at the boundary between the strong scattering region and the weak scattering region, supplement and correct the time-frequency connectivity structure that satisfies the interference prior conditions to obtain the final time-frequency domain interference localization result.

[0140] In step S106, the initial time-frequency domain interference localization result obtained in step S105 is constrained and corrected by combining the interference prior information constructed in step S103; at the same time, the time-frequency connectivity structure that satisfies the interference prior conditions is supplemented and corrected at the boundary between the strong scattering region and the weak scattering region to obtain the final time-frequency domain interference localization result.

[0141] Suppose that the initial time-frequency domain interference localization result obtained in step S105 can be expressed as a binary indicator function. The value is 1 when the corresponding time-frequency submatrix is ​​determined to be an interference region, and 0 otherwise. The interference prior information constructed in step S103 includes the frequency band occupied by the interference in the frequency dimension. and the duration range of a single interference. Based on the prior information about the interference, a consistency constraint is applied to the initial positioning results. That is, only time-frequency regions that simultaneously satisfy the condition that the frequency position falls within the frequency band occupied by the interference and the duration of the continuous time matches the characteristics of a single interference event are retained, thereby eliminating false alarms caused by noise or strong scattering points.

[0142] During the constraint process, when a certain time-frequency region satisfies

[0143]

[0144] If a region is connected to an identified interference region in the time-frequency domain, it is confirmed as an interference region; otherwise, time-frequency regions that do not meet the prior conditions for interference or are only isolated are removed from the interference localization results.

[0145] Furthermore, at the boundary between strong and weak scattering regions, due to abrupt changes in background energy distribution, the interference structure may appear discontinuous or partially missing in the initial localization results. To address this, the time-frequency connectivity structure located at the boundary between strong and weak scattering regions is analyzed. When it meets the aforementioned interference prior conditions in terms of frequency range and temporal duration, the corresponding time-frequency region is supplemented and marked; for connectivity structures that significantly deviate from the interference prior characteristics, they are corrected or deleted.

[0146] Through the above-described constraint and correction process based on prior interference information, the initial interference localization result is optimized, so that the final time-frequency domain interference localization result maintains structural integrity while reducing false detections and missed detections caused by complex background conditions, providing an accurate and reliable localization basis for subsequent interference suppression processing.

[0147] Step S107: For the finally determined time-frequency domain interference location area, use the block subspace filtering method to perform interference suppression processing, and output the interference-suppressed echo data as the final interference suppression result.

[0148] In step S107, the interference time-frequency localization result determined in step S106 is assumed to correspond to a local data matrix in the pulse time-domain representation, denoted as...

[0149]

[0150] in The bandwidth occupied by the interference area. For the duration of the interference, It includes useful echo components, interference components, and noise components. A sample covariance matrix is ​​constructed for the data block.

[0151]

[0152] Performing eigenvalue decomposition on the covariance matrix, we have:

[0153]

[0154] in Let be an eigenvalue matrix, and satisfy... , This is the corresponding eigenvector matrix.

[0155] Based on the interference localization results and the characteristics of feature value distribution, the feature vectors corresponding to larger feature values ​​are identified as the interference subspace, which can be represented as follows:

[0156]

[0157] in Let be the dimension of the interference subspace. Based on this, construct the orthogonal projection matrix of the interference subspace.

[0158]

[0159] in Accordingly, the projection matrix of the non-interference subspace can be expressed as:

[0160] in It is an identity matrix.

[0161] The data block is filtered using the non-interference subspace projection matrix to obtain interference-suppressed echo data.

[0162]

[0163] By performing the block subspace filtering process described above, interference components are projected from the original data blocks and suppressed, while useful echo components in the non-interference subspace are retained, thereby effectively suppressing the location interference area. The final interference suppression result can be obtained by performing the aforementioned processing on the data blocks corresponding to each interference area and reconstructing the echo data.

[0164] This invention performs a range-to-Fourier transform on pulse-by-pulse synthetic aperture radar (SAR) echo data containing interference to calculate spectral flatness, selecting several pulses with the lowest spectral flatness as reference pulses. A short-time Fourier transform is then performed on the interference-containing pulses to generate several sub-matrices in the time-frequency domain and construct local amplitude features in the time-frequency domain. Based on the local amplitude features of the reference pulses in the time-frequency domain, the absolute median difference method is used to locate the interference region, extract the occupied frequency band and duration of the interference in the time-frequency domain, and construct prior information about the interference. Following the same time-frequency division method as the reference pulses, the interference-containing pulses are divided into time-frequency sub-matrices and calculated... The invention calculates local amplitude features and uses the absolute median difference method to divide the time-frequency representation of the pulse into strong scattering and weak scattering regions. Within each region, K-means clustering is performed on the local amplitude features in the time-frequency domain to initially distinguish between interfering and non-interfering regions, obtaining initial interference localization results. Interference prior information is then used to constrain and correct the initial interference localization results, improving the completeness and consistency of interference localization results under complex background conditions. Finally, a block subspace filtering method is used to suppress interference components within the finally determined interference region, obtaining interference-suppressed echo data. Compared with existing technologies, this invention can effectively suppress broadband radio frequency interference in synthetic aperture radar echoes under complex background conditions, avoiding the problem of insufficient adaptability of a uniform threshold in different scenarios. By introducing interference prior constraints and a time-frequency divide-and-conquer mechanism, the accuracy of interference suppression is improved while reducing damage to useful echo signals, thereby improving subsequent imaging quality.

[0165] Exemplary System

[0166] Figure 4A broadband interference suppression system for spaceborne synthetic aperture radar, provided with prior time-frequency information assistance for embodiments of this application, includes: an interference prior information construction unit, configured to perform range-to-Fourier transform on the interference pulses and calculate the spectral flatness, selecting several pulses with the lowest spectral flatness as reference pulses; subsequently performing short-time Fourier transform on the reference pulses to locate the interference region, extracting the frequency occupancy range and single interference duration information of the interference, and constructing interference prior information; a time-frequency transform and local partitioning unit, configured to perform short-time Fourier transform on the interference echo to be processed to obtain the energy distribution of the echo signal in the time-frequency domain; and partition the time-frequency representation result in the time-frequency domain to form several time-frequency sub-matrices, statistically analyzing the amplitude information of each time-frequency sub-matrices to construct local amplitude features in the time-frequency domain; and a time-frequency region partitioning unit, configured to analyze the amplitude distribution of each time-frequency sub-matrices using the absolute median difference method based on the local amplitude features in the time-frequency domain, dividing the time-frequency domain into strong scattering regions and weak scattering regions. The interference localization unit is configured to perform cluster analysis on the local amplitude features in the time-frequency domain within strong and weak scattering regions, respectively, to achieve preliminary segmentation of the interference and non-interference regions, obtaining the initial localization result of the time-frequency domain interference. The prior constraint correction unit is configured to constrain and correct the initial interference localization result by combining the interference prior information; at the boundary between the strong and weak scattering regions, it supplements and corrects the time-frequency connectivity structure that satisfies the interference prior conditions, obtaining the final time-frequency domain interference localization result. The interference suppression unit is configured to perform interference suppression processing using a block subspace filtering method on the finally determined time-frequency domain interference localization region to suppress broadband interference components and retain useful echo signals. The pulse combining unit is configured to sequentially combine the echo data after interference suppression processing as the final interference suppression result for subsequent imaging processing.

[0167] The prior time-frequency information-assisted satellite-borne synthetic aperture radar broadband interference suppression system provided in this application embodiment can realize any of the above-mentioned prior time-frequency information-assisted satellite-borne synthetic aperture radar broadband interference suppression steps and processes, and achieve the same technical effect, which will not be described in detail here.

[0168] Exemplary device

[0169] A computer-readable storage medium stores a computer program thereon. When executed by a processor, the computer program causes the device containing the computer-readable storage medium to perform the aforementioned a priori time-frequency information-assisted broadband interference suppression method for spaceborne synthetic aperture radar. The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), random access memory, and other memories.

[0170] This application provides an electronic device, including a storage device and a processor. The processor is suitable for executing various programs; the memory is used to store multiple programs; when the memory executes the programs on the processor, it implements the aforementioned a priori time-frequency information-assisted broadband interference suppression method for spaceborne synthetic aperture radar.

[0171] Since the broadband interference suppression steps of spaceborne synthetic aperture radar assisted by prior time and frequency information have been described in detail in the specific implementation method examples, they will not be repeated here.

[0172] The processor includes a Central Processing Unit (CPU), a Network Processor (NP), etc., and can also be a digital signal processor, an application-specific integrated circuit, an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0173] The processor can be specifically configured as follows:

[0174] First, range-to-Fourier transform is performed pulse-by-pulse on the synthetic aperture radar (SAR) echo data containing interference to calculate spectral flatness. Several pulses with the lowest spectral flatness are selected as reference pulses. Prior information such as the frequency occupancy range of broadband interference and the duration of a single interference event is extracted through time-frequency amplitude analysis. Next, short-time Fourier transform is performed on the interference pulses to divide them into sub-matrices in the time-frequency domain and statistically analyze local amplitude characteristics. The absolute median difference method is used to distinguish between strong and weak scattering regions, and cluster analysis is applied in different scattering regions to achieve preliminary location of the interference region. The preliminary interference location results are constrained and corrected based on the prior interference information to improve the completeness and consistency of the interference location results under complex background conditions. Finally, block subspace filtering is used to suppress the interference components within the finally determined interference region to obtain the interference-suppressed echo data.

[0175] It should be noted that, depending on the implementation needs, the various components / steps described in the embodiments of this application can be broken down into more components / steps, or two or more components / steps or parts of the operation of components / steps can be combined into new components / steps to achieve the purpose of the embodiments of this application.

[0176] The methods described in the embodiments of this application can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code downloaded over a network that is originally stored in a remote recording medium or a non-transitory machine storage medium and will be stored in a local recording medium. Thus, the methods described herein can be stored as software processing on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components (e.g., RAM, ROM, flash memory, etc.) capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the synthetic aperture radar interference suppression method based on dynamic threshold subspace projection described herein is implemented. Furthermore, when a general-purpose computer accesses the code used to implement the methods shown herein, the execution of the code transforms the general-purpose computer into a dedicated computer for executing the methods shown herein.

[0177] like Figure 2 and Figure 5As shown, based on the interference-affected measured data obtained by Sentinel-1, a comparative experiment was conducted on the method of the present invention with the existing Cross-Domain Multi-Feature Fusion-Based RFI Detection and Localization Method (CDMFF) and Time-Frequency Domain-Block Subspace Filtering Method (TFD-BSF). Figure 2 The original SAR image affected by radio frequency interference is presented. Figure 5 The comparison results of the interference suppression effects of CDMFF, TFD-BSF and the method of the present invention are shown, and the local areas marked by the rectangles are shown in magnified display on the right side of the figure.

[0178] The experimental results show that in the sea area, due to the overall weak scattering intensity, the CDMFF method, which performs interference localization by dividing the time-frequency representations of interfering echoes with those of adjacent interference-free echoes under complex time-frequency background conditions, struggles to adapt to variations in background scattering characteristics. This results in interference components in weakly scattering regions not being effectively detected and suppressed, leaving more noticeable interference artifacts in the sea area compared to land areas. While the TFD-BSF method can mitigate the impact of interference on imaging results to some extent, its limited ability to constrain the subspace filtering area makes it prone to over-suppressing useful echo signals during interference suppression, leading to significant loss of image details and structural information.

[0179] In contrast, the method of this invention, based on the local amplitude difference characteristics in the time-frequency domain, divides the strong scattering region into a strong scattering region and a weak scattering region, and then constructs an adaptive threshold and constraint condition to apply block subspace filtering only to the precisely located interference region, thereby significantly improving the robustness of the interference suppression method under complex background conditions. Experimental results show that, under conditions of drastic fluctuations such as alternating sea and land, the method of this invention can still effectively suppress radio frequency interference while better preserving useful information in synthetic aperture radar images, and its overall interference suppression performance is superior to the comparative methods.

[0180] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application of the technical solution and the constraints involved. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this application.

[0181] It should be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for the device and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments.

[0182] The device and system embodiments described above are merely illustrative. The units referred to as separate entities may or may not be physically separate. The entities mentioned as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0183] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0184] Note that the above description is merely a preferred embodiment and application of the technical principles of the present invention. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the specific embodiments described herein, and may include many other effective embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information, characterized in that: Includes the following steps: Step S101: Perform a distance-to-Fourier transform on each pulse of the interference echo data, calculate the spectral flatness, sort the pulses according to the magnitude of the spectral flatness, and select the pulses with the lowest spectral flatness as reference pulses. Step S102: Based on the reference pulse, perform a short-time Fourier transform on it to generate several sub-matrices in the time-frequency domain and construct local amplitude features in the time-frequency domain. Step S103: Based on the time-frequency local amplitude characteristics of the pulse, locate the interference region using the absolute median difference method, extract the occupied frequency band and duration of the interference in the time-frequency domain, and construct prior information about the interference. Step S104: Based on the method described in step S102, the time-frequency representation of the interference pulse is divided into sub-matrices and the local amplitude characteristics are calculated. The absolute median difference method is used to divide the time-frequency representation of the pulse into a strong scattering region and a weak scattering region. Step S105: Perform K-means clustering on the local amplitude features in the time-frequency domain in the strong scattering region and the weak scattering region respectively to achieve a preliminary distinction between the interference region and the non-interference region, and obtain the initial localization result of the time-frequency domain interference. Step S106: Combine the interference prior information constructed in step S103 to constrain and correct the initial interference localization result; at the same time, at the boundary between the strong scattering region and the weak scattering region, supplement and correct the time-frequency connectivity structure that satisfies the interference prior conditions to obtain the final time-frequency domain interference localization result. Step S107: For the finally determined time-frequency domain interference location area, use the block subspace filtering method to perform interference suppression processing, and output the interference-suppressed echo data as the final interference suppression result.

2. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, In step S101, a range-to-Fourier transform is performed pulse by pulse, and the spectral flatness is calculated. Several pulses with the lowest spectral flatness are selected as reference pulses to extract broadband interference components with prominent interference characteristics. Specifically: To measure the uniformity of the energy distribution in the echo spectrum, let and These represent the number of sampling points in the range and azimuth directions, respectively. A range Fourier transform is performed pulse-by-pulse to obtain its frequency domain representation. ； in, Indicates the first A time-domain sampling sequence of pulses in the range direction. For distance-oriented sampling point index; Represents the Discrete Fourier Transform operator; For the corresponding frequency domain complex spectral components, Represent the frequency index and calculate its amplitude spectrum: ； in, Indicates the first The pulse in the first The amplitude values ​​at each frequency point are used to calculate the spectral flatness (SF), which is defined as the ratio of the geometric mean to the arithmetic mean of the amplitude spectrum. ； Spectral flatness corresponding to the interfering pulse Sort the data and select the smallest value. The pulses are used as a reference pulse set.

3. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, Step S102 specifically includes the following steps: Let the reference pulse set obtained in step S101 be: ,in Indicates the first The time-domain echo signal of the reference pulse in the range direction, Reference pulse count; Perform time-frequency analysis on each of the reference pulses, and process the reference pulses using short-time Fourier transform. The time-frequency representation can be expressed as follows: ； in, To analyze window functions, and The time index and frequency index are represented respectively; the amplitude value is taken from the time-frequency representation result to obtain the corresponding time-frequency domain amplitude distribution. It is used to characterize the distribution of signal energy in the time and frequency dimensions; After obtaining the time-frequency domain amplitude distribution, the time-frequency representation result is divided within the time-frequency plane according to a preset time scale and frequency scale, dividing the entire time-frequency plane into several non-overlapping time-frequency sub-matrices; each time-frequency sub-matrix corresponds to a local region in the time-frequency plane, used to describe the local energy characteristics of echoes and interference within that region; let the first... The index range of each time-frequency submatrix in the time dimension is: The index range in the frequency dimension is: Then the first The summation of amplitudes within each time-frequency submatrix can be expressed as: ； The statistics, as local amplitude features in the time-frequency domain, are used to characterize the overall level of energy distribution within that time-frequency region.

4. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, In step S103, based on the local amplitude features corresponding to all time-frequency sub-matrices... The absolute median difference method is used to divide the region into interference and non-interference regions: First, the median of the local amplitude characteristics is calculated. ； Based on this, the absolute median difference is calculated, and its expression is: ； The segmentation threshold was then determined and set as follows: ； in Using empirical coefficients, submatrices with amplitudes exceeding a set threshold are identified as interference regions, while the rest are considered non-interference regions. Based on the obtained interference regions, their frequency-dimensional occupancy range can be represented as intervals: ； in Indicates the first The starting frequency position of the interference band in the reference pulse; in the time dimension, the corresponding duration of a single interference can be expressed as: ； in and These represent the start and end times of the interference in the time-frequency domain, respectively. For the set of starting positions of the frequency band and duration set Outlier handling is performed; an outlier detection method based on the median is used, that is, for any set of parameters... Calculate its absolute median and remove those that satisfy: Abnormal samples, among which This is an empirical coefficient; After outlier removal, statistical fusion was performed on the remaining valid samples to determine the final prior parameters of the interference. The average value of the starting position of the interference frequency band was taken, and the duration of each interference event was also statistically analyzed in the same way, resulting in: ； ； in This indicates the number of valid samples remaining after outlier removal; through the above processing, the number of samples including those occupying interference frequency bands is obtained. and the duration of a single interference Interference prior information.

5. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, Step S104 specifically includes the following steps: Based on the method described in step S102, for the first... The pulse in the first The summation of the amplitudes within each time-frequency submatrix is: ,in Indicates the index of the time-frequency submatrix. The total number of time-frequency sub-matrices; for the same echo pulse, the sum of the amplitudes corresponding to all time-frequency sub-matrices constitutes a local amplitude feature sequence. To distinguish the scattering intensity corresponding to different time-frequency regions, the local amplitude feature sequence is statistically analyzed using the absolute median difference method to divide it into strong scattering regions and weak scattering regions. When the sum of the amplitudes corresponding to a certain time-frequency submatrix is ​​greater than the threshold range determined by the absolute median difference, the time-frequency submatrix is ​​determined to be a strong scattering region; the rest are determined to be weak scattering regions.

6. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, In step S105, it is assumed that step S104 has already divided the time-frequency sub-matrix set into a strong scattering sub-matrix set. With the set of weak scattering submatrices ; For any pulse containing interfering echo Take its corresponding local amplitude sum in each time-frequency submatrix. As local amplitude features in the time-frequency domain; within the sets of strong and weak scattering sub-matrices, sample sequences composed of the aforementioned local amplitude features are constructed respectively, and the K-means clustering method is used to automatically classify the sample sequences within each scattering region; the number of clusters is set to 2 to divide the time-frequency sub-matrices into interference and non-interference classes; the objective function of K-means clustering is defined as minimizing the sum of squared Euclidean distances between a sample and its cluster center, i.e.: ； in Indicates the first A cluster, This represents the center value of the corresponding cluster, which is calculated as the mean of all samples within that cluster; for any local amplitude feature sample It is assigned to the cluster corresponding to the cluster center with the smallest Euclidean distance from it, that is: ； ； After clustering is completed, the clustering results are judged based on the mean of the local amplitude features corresponding to each cluster. Clusters with larger mean local amplitude features are judged as interference candidate regions, while clusters with smaller mean local amplitude features are judged as non-interference regions.

7. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, Step S106 specifically includes the following steps: Suppose that the initial time-frequency domain interference localization result obtained in step S105 can be expressed as a binary indicator function. The value is 1 when the corresponding time-frequency submatrix is ​​determined to be an interference region, and 0 otherwise; the interference prior information constructed in step S103 includes the frequency band occupied by the interference in the frequency dimension. and the duration range of a single interference. Based on the prior information of the interference, the initial positioning results are subject to consistency constraints, that is, only time-frequency regions that simultaneously satisfy the condition that the frequency position falls within the frequency band occupied by the interference and the duration of the continuous time meets the characteristics of the duration of a single interference are retained, thereby eliminating false alarms caused by noise or strong scattering points. During the constraint process, when a certain time-frequency region satisfies: ； If a region is connected to an identified interference region in the time-frequency domain, it is identified as an interference region; otherwise, time-frequency regions that do not meet the prior conditions for interference or are only isolated are removed from the interference localization results. Furthermore, at the boundary between the strong scattering region and the weak scattering region, the time-frequency connectivity structure located at the boundary between the strong and weak scattering regions is analyzed. When it meets the aforementioned interference prior conditions in terms of frequency range and time duration, the corresponding time-frequency region is supplemented and marked. For connectivity structures that significantly deviate from the interference prior characteristics, they are corrected or deleted.

8. The broadband interference suppression method for spaceborne synthetic aperture radar assisted by prior time-frequency information according to claim 1, characterized in that, Step S107 specifically includes the following steps: Assume the interference time-frequency localization result determined in step S106 corresponds to a local data matrix in the pulse time-domain representation, denoted as: ； in The bandwidth occupied by the interference area. For the duration of the interference, It includes useful echo components, interference components, and noise components; a sample covariance matrix is ​​constructed for the data block: ； Performing eigenvalue decomposition on the covariance matrix, we have: ； in Let be an eigenvalue matrix, and satisfy... , This is the corresponding eigenvector matrix; Based on the interference localization results and the characteristics of feature value distribution, the feature vectors corresponding to larger feature values ​​are identified as the interference subspace, which can be represented as follows: ； in Let be the dimension of the interference subspace; thus, construct the orthogonal projection matrix of the interference subspace: ； in Correspondingly, the projection matrix of the non-interference subspace can be expressed as: in It is the identity matrix; The data block is filtered using the non-interference subspace projection matrix to obtain interference-suppressed echo data: ； By performing the above-described block subspace filtering process, interference components are projected from the original data blocks and suppressed, while useful echo components in the non-interference subspace are retained, thereby achieving effective suppression of the location interference area. By performing the above processing on the data blocks corresponding to each interference area and reconstructing the echo data, the final interference suppression result can be obtained.

9. A broadband interference suppression system for spaceborne synthetic aperture radar assisted by prior time-frequency information, characterized in that, include: The interference prior information construction unit is configured to perform Fourier transform on the interference pulses and calculate the spectral flatness, and select several pulses with the lowest spectral flatness as reference pulses. Subsequently, a short-time Fourier transform is performed on the reference pulse to locate the interference region, extract the frequency range occupied by the interference and the duration of a single interference, and construct the prior information of the interference. The time-frequency transformation and local partitioning unit is configured to perform short-time Fourier transform on the interference echo to be processed to obtain the energy distribution of the echo signal in the time-frequency domain; and to partition the time-frequency representation result in the time-frequency domain to form several time-frequency sub-matrices, and to statistically analyze the amplitude information of each time-frequency sub-matrices to construct local amplitude features in the time-frequency domain. The time-frequency region division unit is configured to analyze the amplitude distribution of each time-frequency sub-matrix based on the local amplitude characteristics of the time-frequency domain using the absolute median difference method, and divide the time-frequency domain into strong scattering regions and weak scattering regions to distinguish the echo characteristics under different scattering conditions. The partitioned interference localization unit is configured to perform cluster analysis on the local amplitude features in the time-frequency domain in the strong scattering region and the weak scattering region respectively, so as to achieve preliminary segmentation of the interference region and the non-interference region and obtain the initial localization result of the time-frequency domain interference. The prior constraint correction unit is configured to combine the interference prior information to constrain and correct the initial interference localization result; at the boundary between the strong scattering region and the weak scattering region, the time-frequency connectivity structure that satisfies the interference prior conditions is supplemented and corrected to obtain the final time-frequency domain interference localization result. The interference suppression unit is configured to use a block subspace filtering method to suppress interference for the finally determined time-frequency domain interference location area, so as to suppress broadband interference components and retain useful echo signals. The pulse combining unit is configured to sequentially combine the echo data after interference suppression processing to obtain the final interference suppression result for subsequent imaging processing.

10. A storage device storing a plurality of programs, characterized in that, The program is loaded and executed by a processor to implement the broadband interference suppression method for spaceborne synthetic aperture radar with prior time-frequency information as described in any one of claims 1-7.

Images