SAR (Synthetic Aperture Radar) intermittent sampling forwarding interference suppression method and system based on time-frequency entropy feature and fine mask, and storage medium

Abstract

The invention discloses an SAR (Synthetic Aperture Radar) intermittent sampling forwarding interference suppression method and system based on a time-frequency entropy feature and a fine mask, and a storage medium. The method comprises the following steps: converting an interference-containing SAR echo into a time-frequency domain, calculating a time-frequency information entropy of a pulse, and then carrying out threshold segmentation so as to accurately obtain an interference-containing pulse sequence; and through two times of fine masks with different dimensions, time-frequency domain positioning of the 2D-ISRJ is realized. And designing a time-frequency domain filter, and carrying out time-frequency domain wave trapping on the positioned 2D-ISRJ. And finally, recovering the missing part of the signal after wave trapping by using the time-frequency data of the nearest interference-free pulse. According to the invention, under the conditions of complex background and low interference-to-signal ratio of a city, high interference detection precision can be provided for typical two-dimensional intermittent sampling forwarding interference forwarding types such as two-dimensional intermittent sampling direct forwarding interference, repeated forwarding interference and cyclic forwarding interference, and the suppression method not only can achieve a good suppression effect, but also can be used for detecting the interference of the typical two-dimensional intermittent sampling forwarding interference forwarding types of the typical two-dimensional intermittent sampling forwarding interference forwarding types of the typical two-dimensional intermittent sampling forwarding interference. And useful signals can be better protected.

Description

Technical Field

[0001] This application relates to the field of signal processing technology, and in particular to a method, system and storage medium for suppressing SAR intermittent sampling forwarding interference based on time-frequency entropy characteristics and fine masking. Background Technology

[0002] Synthetic Aperture Radar (SAR) is an active microwave imaging device that improves range resolution by emitting a wide-bandwidth signal and achieves high azimuth resolution by utilizing the equivalent synthesis of a large antenna array through platform motion. Unlike other Earth observation methods such as infrared detection, SAR's unique imaging mechanism gives it advantages such as all-weather, all-day operation, high resolution, certain penetration, and wide mapping range. It has wide applications in fields such as agricultural monitoring, topographic mapping, military reconnaissance, and deep space exploration.

[0003] However, with the continuous emergence of new jamming technologies, especially the widespread application of Digital Radio Frequency Memory (DRFM) technology, jamming devices can quickly acquire radar signals, modulate them, and then retransmit them, effectively achieving suppression and deception jamming effects. This poses a significant challenge to radar target detection and severely impacts radar performance. Interrupted Sampling Repeater Jamming (ISRJ) is an intra-pulse coherent jamming based on DRFM. Its principle is as follows: A small segment of the SAR linear frequency modulated signal is intercepted, sampled with high fidelity, and then retransmitted. The next segment is sampled and retransmitted, and this process continues in a time-division multiplexing manner, alternating between sampling and retransmission until the large-bandwidth signal ends. Furthermore, leveraging the digital processing capabilities of DRFM, ISRJ jammers can flexibly adjust jamming parameters and implement diverse jamming strategies, posing a serious threat to synthetic aperture radar (SAR) systems. In conclusion, the flexibility and adaptability of ISRJ increase the difficulty of radar anti-jamming and significantly impact the normal operation and combat effectiveness of radar systems. Therefore, researching effective methods to combat ISRJ has significant practical implications.

[0004] However, in current research, traditional methods for combating intermittent sampling-forwarding interference mainly focus on two dimensions: transmitter waveform design and receiver signal processing. Improving the anti-interference performance of transmitter waveform design methods relies on sensing the jammer's operating parameters, which are difficult to obtain in practice. The anti-interference effect of receiver signal processing methods is closely related to the accuracy of interference detection and suppression, requiring high-precision interference detection methods and high-performance interference suppression methods. While existing methods can achieve a certain level of interference suppression under strong interference scenarios, they still have significant limitations under weak interference conditions, directly restricting the overall interference suppression effect. Furthermore, existing methods are primarily geared towards traditional detection radars. For two-dimensional imaging radars such as synthetic aperture radar (SAR), due to their unique azimuth characteristics, current methods are difficult to apply, and related research is limited.

[0005] In conclusion, it is urgent and of great practical significance to conduct in-depth research on synthetic aperture radar anti-interference technology for intermittent sampling and forwarding. Summary of the Invention

[0006] The purpose of this application is to provide a method, system, and storage medium for suppressing SAR intermittent sampling forwarding interference based on time-frequency entropy characteristics and fine masking, so as to solve or alleviate the problems existing in the prior art.

[0007] To achieve the above objectives, this application provides the following technical solution:

[0008] This application provides a SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy characteristics and fine masking, including: step S101, performing short-time Fourier transform on each pulse of echo data; step S102, calculating and obtaining the time-frequency information entropy of each pulse of the obtained short-time Fourier transform result, and then thresholding to obtain the interference-containing pulse sequence; step S103, performing short-time Fourier transform on the obtained interference-containing pulse to obtain a two-dimensional time-frequency amplitude matrix; step S104, calculating the energy threshold based on the local mean and standard deviation, and segmenting the time-frequency amplitude matrix to obtain a preliminary masking result; step S105, calculating the area, eccentricity, and aspect ratio thresholds of the preliminary masking result, performing secondary segmentation to obtain a fine masking result; step S106, performing left and right dilation on the fine masking result to obtain time-frequency domain localization of the interference main lobe and side lobes; step S107, designing a time-frequency domain filter, assigning the interference location to the mean of the data corresponding to the two adjacent interference-free sides; step S108, performing the above steps on each interference-containing pulse to achieve interference suppression.

[0009] Preferably, in step S101, a short-time Fourier transform is performed on each pulse of the echo data, specifically as follows:

[0010] Assume the echo signal of each pulse received by the SAR is represented as Performing a Short-Time Fourier Transform (STFT) operation on it, the result can be expressed as:

[0011]

[0012] in, Indicator signal Complex representation in the joint time-frequency domain, Indicated by A time-centric sliding window function. Represents frequency variables. This is the integration variable. Furthermore, to obtain high-precision time-frequency resolution and suppress spectral leakage, the window function... Choosing the Hamming window as the primary window, its discrete form can be expressed as:

[0013]

[0014] in, Indicates the length of the window function. This serves as the index for the sampling points within the window. Finally, taking the modulus of the transform result yields the two-dimensional time-frequency amplitude matrix for subsequent processing, which can be represented as:

[0015]

[0016] in, This refers to the two-dimensional time-frequency amplitude distribution of the echo signal, which reflects the variation of signal energy with time and frequency.

[0017] Finally, by performing the above process pulse by pulse, the basic data of time-frequency information entropy can be obtained.

[0018] Preferably, in step S102, after calculating the time-frequency information entropy sequence of the obtained short-time Fourier transform result pulse by pulse, and then performing threshold segmentation to obtain the interference-containing pulse sequence, the specific steps are as follows:

[0019] Time-frequency information entropy is an important statistical indicator describing the sparsity and focus of a signal's energy distribution in the joint time-frequency domain. Theoretically, it utilizes the probability distribution characteristics of signal energy to characterize the degree of disorder or order in the data, providing a quantitative basis for distinguishing between intermittent sampling interference with specific structural characteristics and randomly distributed background clutter. First, construct the... The normalized energy probability density distribution of a pulse in the time-frequency domain can be calculated using the following formula:

[0020]

[0021] in, Indicates the first The two-dimensional time-frequency matrix of a pulse after short-time Fourier transform and These represent the time sampling point index and the frequency sampling point index, respectively. and These represent the total number of sampling points on the time axis and the frequency axis, respectively. Representing time and frequency points The proportion of energy at a given point to the total energy of the pulse.

[0022] Furthermore, according to the definition of Shannon entropy, the time-frequency information entropy of this pulse is calculated, and the result can be expressed as:

[0023]

[0024] in, Indicates the first The time-frequency entropy value of each pulse. Since the intermittent sampling interference signal exhibits a strip-like or block-like structure with highly concentrated energy in the time-frequency domain, its entropy value is significantly lower than that of background clutter with randomly distributed energy. Therefore, this index has high identification accuracy for interference pulses.

[0025] By performing the above processing pulse by pulse, the time-frequency entropy sequence of all pulses can be obtained, which can be represented as:

[0026]

[0027] in, This represents the total number of pulses in the azimuth direction. Finally, the K-means clustering algorithm is used to... Binary segmentation is performed to cluster the entropy value sequence into interference and non-interference classes, yielding the final detection result, which can be represented as:

[0028]

[0029] in, Indicates the first The detection markers for each pulse, with constants 1 and 0 representing pulses containing 2D-ISRJ and pulses not containing 2D-ISRJ, respectively.

[0030] Preferably, in step S103, a short-time Fourier transform is performed on the obtained interference-containing pulse to obtain a two-dimensional time-frequency amplitude matrix, specifically as follows:

[0031] The pulse echo signal marked as containing interference is denoted as... Because Intermittent Sample-and-Forward Interference (ISRJ) exhibits non-stationary characteristics, the Short-Time Fourier Transform (STFT) can effectively reveal its local features in the joint time-frequency domain. Its transform formula can be expressed as:

[0032]

[0033] in, Indicates a pulse containing interference. The complex time-frequency distribution, Centered on The sliding window function at time step is chosen here, specifically the Hamming window, which has low sidelobe characteristics, to reduce the impact of spectral leakage on interference edge localization. and These represent frequency and time variables, respectively.

[0034] To construct the energy feature plane for subsequent morphological processing, the complex time-frequency result needs to be moduloed to obtain a two-dimensional time-frequency amplitude matrix, calculated as follows:

[0035]

[0036] in, This is the resulting two-dimensional time-frequency amplitude matrix. and These represent the operations of extracting the real and imaginary parts, respectively. This matrix intuitively reflects the degree of concentration and geometric shape of the interference signal energy in the time-frequency plane.

[0037] Preferably, in step S104, the energy threshold is calculated based on the local mean and standard deviation, and the time-frequency amplitude matrix is ​​segmented to obtain preliminary masking results, specifically:

[0038] Regarding the obtained two-dimensional time-frequency amplitude matrix To adaptively identify high-energy interference regions in a dynamically changing clutter background, the statistical mean and standard deviation of the current pulse time-frequency distribution are first calculated. The calculation formulas can be expressed as follows:

[0039]

[0040]

[0041] in, The mean of the time-frequency amplitude matrix represents the average background clutter level of the current pulse. The standard deviation of the time-frequency amplitude matrix represents the degree of fluctuation in background energy. and These represent the number of sampling points for the time axis and the frequency axis, respectively.

[0042] Furthermore, based on statistics A criterion is established to construct an adaptive energy decision threshold to distinguish anomalous high-energy interference signals from background clutter. This threshold can be expressed as:

[0043]

[0044] in, This represents the calculated energy decision threshold. This is the sensitivity adjustment coefficient, used to control the false alarm rate of detection.

[0045] Finally, using this threshold pair Perform pixel-by-pixel binarization segmentation to obtain preliminary interference mask results. Its expression is:

[0046]

[0047] The value 1 indicates that the time-frequency point is identified as a potential interference or strong scattering point area, while the value 0 indicates that the time-frequency point is identified as a background clutter or noise area.

[0048] Preferably, in step S105, the area, eccentricity, and aspect ratio threshold of the preliminary masking result are calculated, and the result is divided into two parts to obtain the fine masking result, specifically:

[0049] For the obtained preliminary mask First, we perform connected component labeling analysis. Assume that the initial mask contains... Let the nth independent connected component be denoted as . The connected components are To accurately distinguish between block-shaped interference signals, linearly distributed strong point targets, and discretely distributed noise, it is necessary to extract the morphological feature parameters of each connected component.

[0050] First, calculate the area, eccentricity, and aspect ratio of each connected component. Their mathematical definitions are described below:

[0051] Area characteristics : indicates the first The total number of non-zero pixels contained within a connected region is used to eliminate isolated noise points in small areas.

[0052]

[0053] Aspect Ratio Characteristics : Represents the ratio of the major axis length to the minor axis length of the circumscribed ellipse of the connected domain. Strong point targets (LFM signals) exhibit a slender structure with an extremely high aspect ratio in the time-frequency domain; while intermittent sampling interference exhibits a blocky structure with a relatively low aspect ratio.

[0054]

[0055] in, and They represent the first The major and minor axis lengths of the equivalent ellipse for each connected region.

[0056] Eccentricity characteristics : Describes the degree to which the shape of a connected component deviates from a circle, with values ​​ranging from 1 to 10. The eccentricity of linear strong point targets approaches 1, while the eccentricity of blocky interference is relatively small.

[0057]

[0058] Furthermore, based on the morphological differences between the interference signal, strong point targets, and background clutter, an area decision threshold is set. Aspect Ratio Upper Threshold and the upper limit of eccentricity threshold A quadratic segmentation decision is performed on each connected component, retaining only regions that simultaneously satisfy the characteristics of being "large area" and "not a thin, elongated line". The resulting fine-grained interference mask is then obtained. It can be represented as:

[0059]

[0060] In this algorithm, a value of 1 indicates that the pixel is identified as a core interference region, while a value of 0 indicates that the pixel is background, noise, or a strong target that has been removed. This step effectively overcomes the false alarm problem inherent in the single energy threshold method, enabling accurate extraction of interference structures.

[0061] Preferably, in step S106, the fine masking result is dilated horizontally to obtain time-frequency domain localization including the interfering main lobe and side lobes, specifically as follows:

[0062] The resulting fine mask Due to the inherent spectral leakage effect of the Short-Time Fourier Transform (STFT) and the time-varying characteristics of intermittent sampling interference, interference signals often exhibit weak side lobes or edge spread around the main lobe in the time-frequency domain. To eliminate the picket fence effect after imaging, the mask must be morphologically extended to cover these edge regions.

[0063] First, construct a linear structuring element along the time axis (i.e., the left-right direction), denoted as... The length of this structuring element Based on the length of the STFT window function and the number of overlap points, its mathematical form can be expressed as:

[0064]

[0065] Subsequently, morphological dilation is performed on the fine mask $M_{fine}(t, f)$ using the structuring element $B$. This operation causes the mask region to extend to the left and right in the time dimension, thereby filling the tiny gaps between interfering slices and covering the side lobes. The final positioning mask $M_{final}(t, f)$ after dilation can be represented as:

[0066]

[0067] in, Representation of morphological dilation operator Represents the mask by vector The result of the translation.

[0068] After the above processing, the final result is This is a complete time-frequency domain localization result including the main lobe and side lobes of the interference, where the region with a value of 1 represents the determined interference distribution range, and the region with a value of 0 represents a safe signal and background region.

[0069] Preferably, in step S107, a time-frequency domain filter is designed, and the location of the interference is assigned the average value of the data corresponding to the two adjacent interference-free locations, specifically:

[0070] First, the obtained time-frequency domain fine-range positioning mask is... The two-dimensional time-frequency data of the current pulse is divided into a set of interference regions. Set of interference-free regions For any time frequency point ,like Then it is determined to belong to ;like Then it is determined to belong to .

[0071] Secondly, for each time-frequency point identified as interference... Search along the time axis to the left and right respectively, finding the nearest interference-free valid data point, and define it as the interpolation reference point. Record the current frequency. Below, the nearest interference-free moment to the left of the interference point is... The nearest interference-free moment on the right is Its mathematical expression can be described as follows:

[0072]

[0073] Furthermore, extract the original short-time Fourier transform data corresponding to these two reference points. and The complex arithmetic mean of the two values ​​is calculated and used as the recovery estimate for the current disturbance point. The calculation formula is:

[0074]

[0075] Finally, a time-frequency domain suppression filter is constructed to update the original data point by point. The original information of the interference-free regions is preserved, while the data in the interference regions is replaced with the calculated mean, thus obtaining the suppressed, clean time-frequency matrix. , can be represented as:

[0076]

[0077] Through the above processing, the area covered by the interference is smoothly filled by the effective object information on both sides of the interference, and the missing echo signal is recovered while filtering out the interference energy.

[0078] Preferably, in step S108, the above steps are performed on each interference pulse to achieve interference suppression, specifically as follows:

[0079] The above process is executed sequentially for each pulse containing interference until the traversal is complete, thus obtaining the raw SAR echo data without 2D-ISRJ.

[0080] This application also provides a SAR intermittent sampling forwarding interference suppression system based on time-frequency entropy clustering and fine masking, comprising: a feature extraction unit configured to perform short-time Fourier transform on pulse-by-pulse from synthetic aperture radar data input to the system, and then calculate and obtain a time-frequency information entropy sequence on pulse-by-pulse; an interference pulse detection unit configured to perform binary segmentation on the obtained time-frequency information entropy sequence using a K-means clustering algorithm to accurately obtain the interference-containing pulse sequence; and an interference slice localization unit configured to perform short-time Fourier transform on the obtained interference-containing pulses to obtain a two-dimensional time-frequency amplitude matrix, and then calculate an energy threshold based on the local mean and standard deviation to suppress the time-frequency amplitude. The matrix is ​​segmented to obtain a preliminary mask result. Then, the area, eccentricity, and aspect ratio threshold are calculated. After secondary segmentation, a fine mask result is obtained and then expanded to achieve time-frequency domain localization of 2D-ISRJ. The time-frequency domain filter design unit is configured to design a time-frequency domain filter for the time-frequency domain location of the interference, that is, the data of the non-interference location remains unchanged, and the interference location is assigned the mean value of the data corresponding to the two adjacent non-interference locations. The interference suppression unit is configured to update the time-frequency data for each interference pulse using the filter response (i.e., the result after mean replacement), and then perform the inverse short-time Fourier transform (ISTFT) to obtain interference-free SAR echo data.

[0081] This application also provides a computer-readable storage medium storing a computer program thereon, the program being the SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine mask as described above.

[0082] This application also provides an electronic device, including: a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine mask as described above.

[0083] Beneficial effects:

[0084] The SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine masking provided in this application first transforms the interfering SAR echo to the time-frequency domain. Utilizing the sparsity difference in the time-frequency distribution between the interfering signal and background clutter, the time-frequency information entropy of the pulses is calculated and thresholded to accurately obtain the interfering pulse sequence. Subsequently, a short-time Fourier transform is performed on the interfering pulses, and then two fine masks of different dimensions are applied to achieve time-frequency domain localization of 2D-ISRJ. A time-frequency domain filter is then designed to perform time-frequency domain notch filtering on the located 2D-ISRJ. Finally, the missing signal portion after notch filtering is recovered using the time-frequency data of the nearest interference-free pulse. This invention achieves high interference detection accuracy for typical 2D intermittent sampling forwarding interference types such as 2D-ISDRJ, 2D-ISPRJ, and 2D-ISCRJ, even in complex urban backgrounds and under low interference-to-signal ratio conditions. Furthermore, the suppression method not only achieves good suppression effects but also better protects useful signals. Attached Figure Description

[0085] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. Wherein:

[0086] Figure 1 This is a schematic diagram of the process of the present invention;

[0087] Figure 2 These are experimental comparison images described in the embodiments of the present invention;

[0088] Figure 3 This is a unit configuration diagram according to this application. Detailed Implementation

[0089] The present application will now be described in detail with reference to the accompanying drawings and embodiments. Various examples are provided by way of interpretation and not by way of limitation. In fact, those skilled in the art will recognize that modifications and variations can be made to the present application without departing from the scope or spirit thereof. For example, a feature shown or described as part of one embodiment may be used in another embodiment to produce yet another embodiment. Therefore, it is desirable that the present application encompass such modifications and variations that fall within the scope of the appended claims and their equivalents.

[0090] Exemplary methods

[0091] like Figure 1 As shown, the SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine masking includes:

[0092] In step S101, a short-time Fourier transform is performed on each pulse of the echo data, specifically as follows:

[0093] Assume the echo signal of each pulse received by the SAR is represented as Performing a Short-Time Fourier Transform (STFT) operation on it, the result can be expressed as:

[0094]

[0095] in, Indicator signal Complex representation in the joint time-frequency domain, Indicated by A time-centric sliding window function. Represents frequency variables. This is the integration variable. Furthermore, to obtain high-precision time-frequency resolution and suppress spectral leakage, the window function... Choosing the Hamming window as the primary window, its discrete form can be expressed as:

[0096]

[0097] in, Indicates the length of the window function. This serves as the index for the sampling points within the window. Finally, taking the modulus of the transform result yields the two-dimensional time-frequency amplitude matrix for subsequent processing, which can be represented as:

[0098]

[0099] in, This refers to the two-dimensional time-frequency amplitude distribution of the echo signal, which reflects the variation of signal energy with time and frequency.

[0100] Finally, by performing the above process pulse by pulse, the basic data of time-frequency information entropy can be obtained.

[0101] In step S102, the time-frequency information entropy sequence is calculated pulse-by-pulse from the obtained short-time Fourier transform result, and then threshold segmentation is performed to obtain the interference-containing pulse sequence, specifically as follows:

[0102] Time-frequency information entropy is an important statistical indicator describing the sparsity and focus of a signal's energy distribution in the joint time-frequency domain. Theoretically, it utilizes the probability distribution characteristics of signal energy to characterize the degree of disorder or order in the data, providing a quantitative basis for distinguishing between intermittent sampling interference with specific structural characteristics and randomly distributed background clutter. First, construct the... The normalized energy probability density distribution of a pulse in the time-frequency domain can be calculated using the following formula:

[0103]

[0104] in, Indicates the first The two-dimensional time-frequency matrix of a pulse after short-time Fourier transform and These represent the time sampling point index and the frequency sampling point index, respectively. and These represent the total number of sampling points on the time axis and the frequency axis, respectively. Representing time and frequency points The proportion of energy at a given point to the total energy of the pulse.

[0105] Furthermore, according to the definition of Shannon entropy, the time-frequency information entropy of this pulse is calculated, and the result can be expressed as:

[0106]

[0107] in, Indicates the first The time-frequency entropy value of each pulse. Since the intermittent sampling interference signal exhibits a strip-like or block-like structure with highly concentrated energy in the time-frequency domain, its entropy value is significantly lower than that of background clutter with randomly distributed energy. Therefore, this index has high identification accuracy for interference pulses.

[0108] By performing the above processing pulse by pulse, the time-frequency entropy sequence of all pulses can be obtained, which can be represented as:

[0109]

[0110] in, This represents the total number of pulses in the azimuth direction. Finally, the K-means clustering algorithm is used to... Binary segmentation is performed to cluster the entropy value sequence into interference and non-interference classes, yielding the final detection result, which can be represented as:

[0111]

[0112] in, Indicates the first The detection markers for each pulse, with constants 1 and 0 representing pulses containing 2D-ISRJ and pulses not containing 2D-ISRJ, respectively.

[0113] In step S103, a short-time Fourier transform is performed on the obtained interference-containing pulse to obtain a two-dimensional time-frequency amplitude matrix, specifically:

[0114] The pulse echo signal marked as containing interference is denoted as... Because Intermittent Sample-and-Forward Interference (ISRJ) exhibits non-stationary characteristics, the Short-Time Fourier Transform (STFT) can effectively reveal its local features in the joint time-frequency domain. Its transform formula can be expressed as:

[0115]

[0116] in, Indicates a pulse containing interference. The complex time-frequency distribution, Centered on The sliding window function at time step is chosen here, specifically the Hamming window, which has low sidelobe characteristics, to reduce the impact of spectral leakage on interference edge localization. and These represent frequency and time variables, respectively.

[0117] To construct the energy feature plane for subsequent morphological processing, the complex time-frequency result needs to be moduloed to obtain a two-dimensional time-frequency amplitude matrix, calculated as follows:

[0118]

[0119] in, This is the resulting two-dimensional time-frequency amplitude matrix. and These represent the operations of extracting the real and imaginary parts, respectively. This matrix intuitively reflects the degree of concentration and geometric shape of the interference signal energy in the time-frequency plane.

[0120] In step S104, the energy threshold is calculated based on the local mean and standard deviation, and the time-frequency amplitude matrix is ​​segmented to obtain preliminary masking results, specifically:

[0121] Regarding the obtained two-dimensional time-frequency amplitude matrix To adaptively identify high-energy interference regions in a dynamically changing clutter background, the statistical mean and standard deviation of the current pulse time-frequency distribution are first calculated. The calculation formulas can be expressed as follows:

[0122]

[0123]

[0124] in, The mean of the time-frequency amplitude matrix represents the average background clutter level of the current pulse. The standard deviation of the time-frequency amplitude matrix represents the degree of fluctuation in background energy. and These represent the number of sampling points for the time axis and the frequency axis, respectively.

[0125] Furthermore, based on statistics A criterion is established to construct an adaptive energy decision threshold to distinguish anomalous high-energy interference signals from background clutter. This threshold can be expressed as:

[0126]

[0127] in, This represents the calculated energy decision threshold. This is the sensitivity adjustment coefficient, used to control the false alarm rate of detection.

[0128] Finally, using this threshold pair Perform pixel-by-pixel binarization segmentation to obtain preliminary interference mask results. Its expression is:

[0129]

[0130] The value 1 indicates that the time-frequency point is identified as a potential interference or strong scattering point area, while the value 0 indicates that the time-frequency point is identified as a background clutter or noise area.

[0131] In step S105, the area, eccentricity, and aspect ratio threshold of the preliminary masking result are calculated, and the result is divided into two parts to obtain the fine masking result, specifically:

[0132] For the obtained preliminary mask First, we perform connected component labeling analysis. Assume that the initial mask contains... Let the nth independent connected component be denoted as . The connected components are To accurately distinguish between block-shaped interference signals, linearly distributed strong point targets, and discretely distributed noise, it is necessary to extract the morphological feature parameters of each connected component.

[0133] First, calculate the area, eccentricity, and aspect ratio of each connected component. Their mathematical definitions are described below:

[0134] Area characteristics : indicates the first The total number of non-zero pixels contained within a connected region is used to eliminate isolated noise points in small areas.

[0135]

[0136] Aspect Ratio Characteristics : Represents the ratio of the major axis length to the minor axis length of the circumscribed ellipse of the connected domain. Strong point targets (LFM signals) exhibit a slender structure with an extremely high aspect ratio in the time-frequency domain; while intermittent sampling interference exhibits a blocky structure with a relatively low aspect ratio.

[0137]

[0138] in, and They represent the first The major and minor axis lengths of the equivalent ellipse for each connected region.

[0139] Eccentricity characteristics : Describes the degree to which the shape of a connected component deviates from a circle, with values ​​ranging from 1 to 10. The eccentricity of linear strong point targets approaches 1, while the eccentricity of blocky interference is relatively small.

[0140]

[0141] Furthermore, based on the morphological differences between the interference signal, strong point targets, and background clutter, an area decision threshold is set. Aspect Ratio Upper Threshold and the upper limit of eccentricity threshold A quadratic segmentation decision is performed on each connected component, retaining only regions that simultaneously satisfy the characteristics of being "large area" and "not a thin, elongated line". The resulting fine-grained interference mask is then obtained. It can be represented as:

[0142]

[0143] In this algorithm, a value of 1 indicates that the pixel is identified as a core interference region, while a value of 0 indicates that the pixel is background, noise, or a strong target that has been removed. This step effectively overcomes the false alarm problem inherent in the single energy threshold method, enabling accurate extraction of interference structures.

[0144] In step S106, the fine masking result is dilated horizontally to obtain time-frequency domain localization including the interfering main lobe and side lobes, specifically as follows:

[0145] The resulting fine mask Due to the inherent spectral leakage effect of the Short-Time Fourier Transform (STFT) and the time-varying characteristics of intermittent sampling interference, interference signals often exhibit weak side lobes or edge spread around the main lobe in the time-frequency domain. To eliminate the picket fence effect after imaging, the mask must be morphologically extended to cover these edge regions.

[0146] First, construct a linear structuring element along the time axis (i.e., the left-right direction), denoted as... The length of this structuring element Based on the length of the STFT window function and the number of overlap points, its mathematical form can be expressed as:

[0147]

[0148] Subsequently, morphological dilation is performed on the fine mask $M_{fine}(t, f)$ using the structuring element $B$. This operation causes the mask region to extend to the left and right in the time dimension, thereby filling the tiny gaps between interfering slices and covering the side lobes. The final positioning mask $M_{final}(t, f)$ after dilation can be represented as:

[0149]

[0150] in, Representation of morphological dilation operator Represents the mask by vector The result of the translation.

[0151] After the above processing, the final result is This is a complete time-frequency domain localization result including the main lobe and side lobes of the interference, where the region with a value of 1 represents the determined interference distribution range, and the region with a value of 0 represents a safe signal and background region.

[0152] In step S107, a time-frequency domain filter is designed, and the location of the interference is assigned the average value of the data corresponding to the two adjacent interference-free locations, specifically:

[0153] First, the obtained time-frequency domain fine-range positioning mask is... The two-dimensional time-frequency data of the current pulse is divided into a set of interference regions. Set of interference-free regions For any time frequency point ,like Then it is determined to belong to ;like Then it is determined to belong to .

[0154] Secondly, for each time-frequency point identified as interference... Search along the time axis to the left and right respectively, finding the nearest interference-free valid data point, and define it as the interpolation reference point. Record the current frequency. Below, the nearest interference-free moment to the left of the interference point is... The nearest interference-free moment on the right is Its mathematical expression can be described as follows:

[0155]

[0156] Furthermore, extract the original short-time Fourier transform data corresponding to these two reference points. and The complex arithmetic mean of the two values ​​is calculated and used as the recovery estimate for the current disturbance point. The calculation formula is:

[0157]

[0158] Finally, a time-frequency domain suppression filter is constructed to update the original data point by point. The original information of the interference-free regions is preserved, while the data in the interference regions is replaced with the calculated mean, thus obtaining the suppressed, clean time-frequency matrix. , can be represented as:

[0159]

[0160] Through the above processing, the area covered by the interference is smoothly filled by the effective object information on both sides of the interference, and the missing echo signal is recovered while filtering out the interference energy.

[0161] In step S108, the above steps are performed on each interference pulse to achieve interference suppression, specifically as follows:

[0162] The above process is executed sequentially for each pulse containing interference until the traversal is complete, thus obtaining the raw SAR echo data without 2D-ISRJ.

[0163] The method provided by this invention exhibits significant technical advantages in each processing stage:

[0164] First, in steps S101 to S102, this invention abandons the traditional single-domain (time domain or frequency domain) energy detection and introduces time-frequency information entropy as an evaluation index. Since intermittent sampling interference exhibits highly concentrated energy sparsity in the time-frequency domain, by calculating the time-frequency information entropy and combining it with K-means clustering, the sensitivity and accuracy of interference detection in complex urban backgrounds or under low interference-to-signal ratio conditions can be greatly improved, effectively reducing the false alarm rate.

[0165] Secondly, in the interference localization stage of steps S103 to S106, this method designs a fine masking mechanism that combines "preliminary local energy threshold masking" with "secondary segmentation based on multi-dimensional morphological features (area, eccentricity, aspect ratio)". This multi-level joint strategy effectively overcomes the defect of strong scattering points being misjudged as interference; at the same time, combined with morphological dilation operations, it not only accurately locks the main lobe of the interference, but also fully covers the weak side lobe edges, eliminating the fence effect and interference residue that may occur after imaging from the root.

[0166] Finally, in steps S107 to S108, this invention innovatively designs a time-frequency domain suppression filter, which uses the mean of effective interference-free data adjacent to both sides of the interference slice for smooth filling. Compared with traditional direct notch filtering or nulling, this step thoroughly filters out 2D-ISRJ energy while maximally restoring and protecting the integrity of the original SAR echo signal, thereby ensuring the final high-quality imaging effect.

[0167] like Figure 2 As shown, the experimental comparison diagrams of this application are as follows: the images in the leftmost column are experimental data containing 2D-ISDRJ, 2D-ISPRJ, and 2D-ISCRJ respectively; the images in the middle two columns are the processing results of existing methods; and the images in the rightmost column are the processing results of this invention. As can be seen from the comparison of the above results, under the same scene data conditions, the suppression effect of this invention on data containing 2D-ISDRJ, 2D-ISPRJ, and 2D-ISCRJ is more significant.

[0168] Exemplary System

[0169] Figure 3This application also provides a SAR intermittent sampling forwarding interference suppression system based on time-frequency entropy clustering and fine masking, comprising: a feature extraction unit configured to perform short-time Fourier transform on pulse-by-pulse from synthetic aperture radar data input to the system, and then calculate and obtain a time-frequency information entropy sequence on pulse-by-pulse; an interference pulse detection unit configured to perform binary segmentation on the obtained time-frequency information entropy sequence using a K-means clustering algorithm to accurately obtain the interference-containing pulse sequence; and an interference slice localization unit configured to perform short-time Fourier transform on the obtained interference-containing pulses to obtain a two-dimensional time-frequency amplitude matrix, and then calculate an energy threshold based on the local mean and standard deviation to suppress the time-frequency amplitude. The degree matrix is ​​segmented to obtain preliminary masking results. Then, the area, eccentricity, and aspect ratio threshold are calculated. After secondary segmentation, a fine masking result is obtained and then expanded to achieve time-frequency domain localization of 2D-ISRJ. The time-frequency domain filter design unit is configured to design a time-frequency domain filter for the time-frequency domain location of the interference, that is, the data of the non-interference location remains unchanged, and the interference location is assigned the mean value of the data corresponding to the two adjacent non-interference locations. The interference suppression unit is configured to update the time-frequency data for each interference pulse using the filter response (i.e., the result after mean replacement), and then perform the inverse short-time Fourier transform (ISTFT) to obtain interference-free SAR echo data.

[0170] The SAR intermittent sampling forwarding interference suppression system based on time-frequency entropy clustering and fine mask provided in this application embodiment can realize any of the above-mentioned SAR intermittent sampling forwarding interference suppression steps and processes, and achieve the same technical effect, which will not be described in detail here.

[0171] Exemplary device

[0172] 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. The feature is that when the memory executes the programs on the processor, it implements the SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine mask.

[0173] 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 SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy characteristics and fine masking as described above. 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 types of memory.

[0174] Since the steps for SAR intermittent sampling and forwarding interference suppression based on time-frequency entropy clustering and fine masking have been described in detail in the specific implementation method examples, they will not be repeated here.

[0175] 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.

[0176] The processor can be specifically configured as follows: It performs a short-time Fourier transform on each pulse of the synthetic aperture radar data from the input system, then calculates and obtains a time-frequency information entropy sequence on each pulse; it performs binary segmentation on the obtained time-frequency information entropy sequence using the K-means clustering algorithm to accurately obtain the interference-containing pulse sequence; it performs a short-time Fourier transform on the obtained interference-containing pulses to obtain a two-dimensional time-frequency amplitude matrix, then calculates the energy threshold based on the local mean and standard deviation, segments the time-frequency amplitude matrix to obtain preliminary masking results, then calculates the area, eccentricity, and aspect ratio thresholds, performs secondary segmentation, obtains a refined masking result, and then dilates it to achieve time-frequency domain localization of 2D-ISRJ; it designs a time-frequency domain filter for the time-frequency domain location of the interference, i.e., the data at its non-interference location remains unchanged, and assigns the interference location the mean of the data corresponding to the two adjacent non-interference locations; it updates the time-frequency data for each interference-containing pulse using the filter response (i.e., the result after mean replacement), and then performs an inverse short-time Fourier transform (ISTFT) to obtain interference-free SAR echo data.

[0177] 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.

[0178] 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 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 processed by software stored 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 single-track synthetic aperture radar jamming source localization method 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.

[0179] 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.

[0180] 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.

[0181] 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.

[0182] 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.

Claims

1. A SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy characteristics and fine masking, characterized in that, include: Step S101: Perform short-time Fourier transform on each pulse of the echo data; Step S102: Calculate the time-frequency information entropy of the obtained short-time Fourier transform result pulse by pulse, and then use threshold segmentation to obtain the pulse sequence containing interference; Step S103: Perform a short-time Fourier transform on the obtained interference-containing pulse to obtain a two-dimensional time-frequency amplitude matrix; Step S104: Calculate the energy threshold based on the local mean and standard deviation, segment the time-frequency amplitude matrix, and obtain preliminary masking results; Step S105: Calculate the area, eccentricity, and aspect ratio threshold of the preliminary masking result, perform secondary segmentation, and obtain the fine masking result; Step S106: Dilate the fine masking result horizontally to obtain time-frequency domain localization including the interfering main lobe and side lobes. Step S107: Design a time-frequency domain filter and assign the value of the location of the interference to the average value of the data corresponding to the two adjacent interference-free locations. Step S108: Perform the above steps for each interference pulse to achieve interference suppression.

2. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking as described in claim 1, characterized in that, In step S101, a short-time Fourier transform is performed on each pulse of the echo data, specifically as follows: Assume the echo signal of each pulse received by the SAR is represented as Performing a Short-Time Fourier Transform (STFT) operation on it, the result can be expressed as: ； in, Indicates signal Complex representation in the joint time-frequency domain, Indicates A time-centric sliding window function. Represents frequency variables. For integration variables; in addition, in order to obtain high-precision time-frequency resolution and suppress spectral leakage, Represents the imaginary unit. The window function is the corresponding echo signal function. Choosing the Hamming Window as the primary window, its discrete form can be expressed as: ； in, Indicates the length of the window function. The index of the sampling points within the window is used; finally, taking the modulus of the transformation result yields the two-dimensional time-frequency amplitude matrix for subsequent processing, which can be represented as: ； in, This refers to the two-dimensional time-frequency amplitude distribution of the echo signal, which reflects the variation of signal energy with time and frequency. Finally, by performing the above process pulse by pulse, the basic data of time-frequency information entropy can be obtained.

3. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking as described in claim 1, characterized in that, In step S102, the time-frequency information entropy sequence is calculated pulse-by-pulse from the obtained short-time Fourier transform result, and then threshold segmentation is performed to obtain the interference-containing pulse sequence, specifically as follows: First, construct the first The normalized energy probability density distribution of a pulse in the time-frequency domain is calculated using the following formula: ； in, Indicates the first The two-dimensional time-frequency matrix of a pulse after short-time Fourier transform and These represent the time sampling point index and the frequency sampling point index, respectively. and These represent the total number of sampling points on the time axis and the frequency axis, respectively. Representing time and frequency points The proportion of energy at a given point to the total energy of the pulse; Furthermore, according to the definition of Shannon entropy, the time-frequency information entropy of this pulse is calculated, and the result can be expressed as: ； in, Indicates the first The time-frequency entropy value of each pulse; By performing the above processing pulse by pulse, the time-frequency entropy sequence of all pulses can be obtained, which can be represented as: ； in, Indicates the total number of pulses in the azimuth direction; Finally, the K-means clustering algorithm was used to... Binary segmentation is performed to cluster the entropy value sequence into interference and non-interference classes, yielding the final detection result, which can be represented as: ； in, Indicates the first The detection markers for each pulse, with constants 1 and 0 representing pulses containing 2D-ISRJ and pulses not containing 2D-ISRJ, respectively.

4. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking as described in claim 1, characterized in that, In step S103, a short-time Fourier transform is performed on the obtained interference-containing pulse to obtain a two-dimensional time-frequency amplitude matrix, specifically: The pulse echo signal marked as containing interference is denoted as... Since the Intermittent Sample-and-Forward Interference (ISRJ) has non-stationary characteristics, the Short Time Fourier Transform (STFT) can effectively reveal its local characteristics in the joint time-frequency domain. Its transformation formula can be expressed as: ； in, Indicates a pulse containing interference. The complex time-frequency distribution, Centered on The sliding window function at time t, and These represent frequency and time variables, respectively. The two-dimensional time-frequency amplitude matrix is ​​obtained, and its calculation formula is as follows: ； in, This is the resulting two-dimensional time-frequency amplitude matrix. and These represent the operations of taking the real and imaginary parts, respectively; this matrix intuitively reflects the degree of concentration and geometric shape of the interference signal energy in the time-frequency plane.

5. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking according to claim 1, characterized in that, In step S104, the energy threshold is calculated based on the local mean and standard deviation, and the time-frequency amplitude matrix is ​​segmented to obtain preliminary masking results, specifically: Regarding the obtained two-dimensional time-frequency amplitude matrix To adaptively identify high-energy interference regions in a dynamically changing clutter background, the statistical mean and standard deviation of the current pulse time-frequency distribution are first calculated; their calculation formulas can be expressed as follows: ； ； in, The mean of the time-frequency amplitude matrix represents the average background clutter level of the current pulse. The standard deviation of the time-frequency amplitude matrix represents the degree of fluctuation in background energy. and These represent the number of sampling points for the time axis and the frequency axis, respectively. Furthermore, based on statistics The criterion is to construct an adaptive energy decision threshold to distinguish anomalous high-energy interference signals from background clutter; this threshold can be expressed as: ； in, This represents the calculated energy decision threshold. This is the sensitivity adjustment coefficient, used to control the false alarm rate of detection; Finally, using this threshold pair Perform pixel-by-pixel binarization segmentation to obtain preliminary interference mask results. Its expression is: ； The value 1 indicates that the time-frequency point is identified as a potential interference or strong scattering point area, while the value 0 indicates that the time-frequency point is identified as a background clutter or noise area.

6. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking according to claim 1, characterized in that, In step S105, the area, eccentricity, and aspect ratio threshold of the preliminary masking result are calculated, and the result is divided into two parts to obtain the fine masking result, specifically: For the obtained preliminary mask First, perform connected component labeling analysis; assume that there are connected components in the initial mask. Let the nth independent connected component be denoted as . The connected components are Extract the morphological feature parameters of each connected component: First, calculate the area, eccentricity, and aspect ratio of each connected component. Their mathematical definitions are described below: Area characteristics : indicates the first The total number of non-zero pixels contained in a connected region is used to eliminate isolated noise points in small areas; ； Aspect Ratio Characteristics : Represents the ratio of the major axis length to the minor axis length of the circumscribed ellipse of the connected domain; strong point target LFM signals exhibit a slender structure in the time-frequency domain with an extremely high aspect ratio; while intermittent sampling interference exhibits a blocky structure with a relatively low aspect ratio; ； in, and They represent the first The major and minor axis lengths of the equivalent ellipse of a connected region; Eccentricity characteristics : Describes the degree to which the shape of a connected component deviates from a circle, with values ​​ranging from 1 to 10. The eccentricity of linear strong point targets approaches 1, while the eccentricity of blocky interference is relatively small. ； Furthermore, based on the morphological differences between the interference signal, strong point targets, and background clutter, an area decision threshold is set. Aspect Ratio Upper Threshold and the upper limit of eccentricity threshold For each connected component, a quadratic segmentation decision is performed, retaining only regions that simultaneously satisfy the characteristics of "large area" and "non-slender line shape"; ultimately, a fine-grained interference mask is obtained. It can be represented as: ； The value 1 indicates that the pixel is identified as the core interference region, and the value 0 indicates that the pixel is background, noise, or a strong target that has been removed.

7. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking according to claim 1, characterized in that, In step S106, the fine masking result is dilated horizontally to obtain time-frequency domain localization including the interfering main lobe and side lobes, specifically as follows: The resulting fine mask Perform morphological extensions to cover these edge regions: First, construct a linear structuring element along the time axis, denoted as . ; the length of this structuring element Based on the length of the STFT window function and the number of overlap points, its mathematical form is as follows: ； Subsequently, the structuring element $B$ is used to perform a morphological dilation operation on the fine mask $M_{fine}(t, f)$; the final positioning mask $M_{final}(t, f)$ after dilation is represented as: ； in, Representation of morphological dilation operator Represents the mask by vector The result of the translation.

8. The SAR intermittent sampling and forwarding interference suppression method based on time-frequency entropy features and fine masking according to claim 1, characterized in that, In step S107, a time-frequency domain filter is designed, and the location of the interference is assigned the average value of the data corresponding to the two adjacent interference-free locations, specifically: First, the obtained time-frequency domain fine-range positioning mask is... The two-dimensional time-frequency data of the current pulse is divided into a set of interference regions. Set of interference-free regions For any time and frequency point ,like Then it is determined to belong to ;like Then it is determined to belong to ; Secondly, for each time-frequency point identified as interference... Search along the time axis to the left and right respectively to find the nearest interference-free valid data point, and define it as the interpolation reference point; record the current frequency. Below, the nearest interference-free moment to the left of the interference point is... The nearest interference-free moment on the right is Its mathematical expression can be described as follows: ； Furthermore, extract the original short-time Fourier transform data corresponding to these two reference points. and The complex arithmetic mean of the two values ​​is calculated and used as the recovery estimate for the current disturbance point. The calculation formula is: ； Finally, a time-frequency domain suppression filter is constructed to update the original data point by point: the original information of the interference-free region is preserved, while the data in the interference region is replaced with the calculated mean, thereby obtaining the suppressed clean time-frequency matrix. , is represented as: ； Through the above processing, the area covered by the interference is smoothly filled by the effective object information on both sides of the interference, and the missing echo signal is recovered while filtering out the interference energy.

9. A SAR intermittent sampling and forwarding interference suppression system based on time-frequency entropy clustering and fine masking, characterized in that, include: The feature extraction unit is configured to perform short-time Fourier transform pulse by pulse from the synthetic aperture radar data of the input system, and then calculate and obtain the time-frequency information entropy sequence pulse by pulse. The interference pulse detection unit is configured to perform binary segmentation on the obtained time-frequency information entropy sequence using the K-means clustering algorithm to accurately obtain the sequence containing interference pulses; The interference slice localization unit is configured to perform a short-time Fourier transform on the obtained interference pulse to obtain a two-dimensional time-frequency amplitude matrix, then calculate the energy threshold based on the local mean and standard deviation, segment the time-frequency amplitude matrix to obtain a preliminary mask result, then calculate the area, eccentricity and aspect ratio threshold, perform a second segmentation, obtain a fine mask result and then dilate to achieve time-frequency domain localization of 2D-ISRJ. The time-frequency domain filter design unit is configured to design a time-frequency domain filter for the time-frequency domain location where the interference is located. That is, the data at the non-interference location remains unchanged, and the value at the location of the interference is assigned to the average value of the data corresponding to the two adjacent non-interference locations. The interference suppression unit is configured to update the time-frequency data by using the filter response (i.e., the result after mean replacement) for each interference pulse, and then perform the inverse short-time Fourier transform (ISTFT) to obtain interference-free SAR echo data.

10. A storage medium storing a plurality of programs, characterized in that, The program application is loaded and executed by the processor to implement the SAR intermittent sampling forwarding interference suppression method based on time-frequency entropy features and fine mask as described in any one of claims 1-9.

Images