Radar interference suppression method based on time-frequency image dynamic threshold search strategy

Abstract

The invention belongs to the technical field of radar interference suppression, and particularly relates to a radar interference suppression method based on a time-frequency image dynamic threshold search strategy. According to the method, a time-frequency analysis and image processing method is adopted, radio frequency receiving, dechirp processing and short-time Fourier transform are carried out on echo signals containing intermittent sampling forwarding interference, the echo signals are converted into a time-frequency domain two-dimensional matrix, and a target can be analyzed and extracted conveniently. And subsequently, an image processing module and a dynamic threshold search algorithm are utilized, so that the target area in the time-frequency two-dimensional matrix can be accurately positioned. And in combination with a target reconstruction method, a target echo signal is further restored, so that the interference filtering effect is better realized.

Description

Technical Field

[0001] The invention belongs to the technical field of radar interference suppression, and in particular relates to a radar interference suppression method based on a time-frequency image dynamic threshold search strategy. Background Art

[0002] In recent years, the advent of Digital Radio Frequency Memory (DRFM) has brought severe challenges to radar anti-interference technology. The intermittent sampling and forwarding interference pattern based on DRFM technology is based on the "store-forward-store-forward" interference concept, and utilizes the intra-pulse coherence of the interference signal and the linear frequency modulation pulse compression radar, which can cause the radar to be interfered by multiple realistic false targets at the same time, achieving the effect of deception and suppression. Therefore, it is difficult for the radar system to effectively identify and suppress the interference of this intermittent sampling and forwarding interference method based on DRFM.

[0003] Since there is serious aliasing between the intermittent sampling forwarding interference signal and the real target echo signal in both time domain and frequency domain distribution, the time domain or frequency domain analysis commonly used in the field of traditional signal processing is difficult to suppress this type of interference. Summary of the invention

[0004] The purpose of the present invention is to construct a new radar interference suppression method based on a dynamic threshold search strategy of a time-frequency image for an intermittent sampling and repeated forwarding interference mode based on radio frequency storage and forwarding. The present invention utilizes the characteristics of the two-dimensional joint analysis of the time-frequency domain of the short-time Fourier transform (STFT) to obtain the frequency distribution information of the signal in different time periods, and proposes a set of interference suppression algorithms based on the discontinuous distribution characteristics of the intermittent sampling and forwarding interference signal in time, combined with the image processing method, which can complete the mining of the difference characteristics of interference and target, and proposes an adaptive dynamic threshold search strategy for the most critical binary threshold determination problem, finally realizing interference suppression and improving the anti-interference ability of the radar system.

[0005] The technical solution adopted by the present invention is as follows: Figure 1 As shown, the following steps are included:

[0006] S1. Get radar receiving data:

[0007] The radar transmission signal is simulated by MATLAB as a linear frequency modulation signal. Based on the linear frequency modulation signal, the target echo signal and intermittent sampling forwarding interference are simulated. The optional styles of intermittent sampling forwarding interference include direct forwarding interference (ISDRJ), repeated forwarding interference (ISPRJ) and cyclic forwarding interference (ISCRJ). The RF simulated radar receiving signal is generated by superposition of target echo, interference signal and Gaussian white noise.

[0008] S2. Perform orthogonal reception and down-sampling of the RF received signal:

[0009] For the simulated RF signal, mixing processing is required to convert it into a baseband signal. The receiving module adopts an orthogonal dual-channel processing method to obtain the IQ two-way baseband signal and synthesize it into a baseband complex signal. Then, the baseband signal is processed by combining an anti-aliasing filter and downsampling to reduce its data rate.

[0010] S3. De-skewing and time-frequency analysis of the baseband complex signal:

[0011] For the baseband complex signal after downsampling, the frequency ramp effect is first eliminated by de-skewing. Then, the de-skewing signal is converted into a two-dimensional time-frequency matrix S(t,f) using short-time Fourier transform, and a normalized grayscale image is generated.

[0012] S4. Extract the distribution area of ​​the target echo using the image processing module and the dynamic threshold search algorithm:

[0013] The image processing module contains three parts, namely binarization, morphological filtering and connected area analysis. First, the grayscale image is binarized based on the current dynamic threshold; then, a structural element with a shape similar to the linear frequency modulation signal of the transmitted wave is constructed in the time-frequency graph, and the binary image is subjected to morphological filtering with an opening operation followed by a closing operation; finally, the processed image is subjected to a connected area analysis to find areas with horizontal lengths close to the linear frequency modulation signal of the transmitted wave. These areas are preliminarily determined to be the areas corresponding to the target echoes, and the target area distribution map is converted into a target area matrix:

[0014]

[0015] The dynamic threshold search algorithm refers to: traversing the binary threshold from high to low, each image processing can generate a target area matrix. In the time-frequency two-dimensional matrix S(t,f), index the area corresponding to the target echo according to T(t,f), and calculate the coefficient of variation of each area:

[0016]

[0017] Among them, x is the set of all values ​​contained in a target area in S(t,f), σ(x) is its standard deviation, and μ(x) is its mean. The coefficient of variation threshold that distinguishes the target echo and the background noise in the time-frequency two-dimensional matrix is ​​used to find the difference in the discrete degree of the numerical distribution of the two. The coefficient of variation of each target area is counted, and the maximum value is compared with the coefficient of variation threshold, and the highest round T(t,f) that is less than the threshold is found as the final target area matrix.

[0018] S5. Reconstruct echo signals based on the target area matrix:

[0019] The echo signal reconstruction includes the following steps: first, multiply the time-frequency two-dimensional matrix S(t, f) by the target area matrix T(t, f) to obtain a time-frequency two-dimensional matrix containing only the target echo; second, use the inverse short-time Fourier transform to restore the time-frequency domain signal to the time domain signal; finally, perform inverse de-skewing on the time domain signal to achieve the reconstruction of the target echo.

[0020] The beneficial effect of the invention is that the present invention adopts the method of time-frequency analysis and image processing to perform radio frequency reception, de-skewing and short-time Fourier transformation on the echo signal containing intermittent sampling forwarding interference, so as to convert it into a two-dimensional matrix in the time-frequency domain, which is convenient for analyzing and extracting the target. The subsequent use of the image processing module and the dynamic threshold search algorithm can more accurately locate the target area in the two-dimensional matrix of time-frequency. Combined with the target reconstruction method, the target echo signal is further restored, thereby better achieving the effect of interference filtering. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Flowchart of radar interference suppression method based on time-frequency analysis and image processing.

[0022] Figure 2 Comparison of pulse compression effects before and after interference suppression. DETAILED DESCRIPTION

[0023] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and embodiments:

[0024] Example

[0025] Step 1: Use MATLAB to generate linear frequency modulation target echo signal and intermittent sampling forwarding interference data. The radar transmission signal bandwidth is 40*10 6 Hz, pulse width is 20*10 -6 s, the modulation slope K is 2*10 12 The signal RF frequency is 1*10 9 Hz, sampling rate is 2.4*10 9 Hz. The sampling duration of intermittent sampling forwarding interference is 2*10 -6s, the forwarding interval is 8*10 -6 s. To simulate the situation where targets and interferences of different strengths appear at different distances, the interference-to-signal ratio of the RF receiving signal is set to 5-30dB, with a step distribution of 5dB intervals, and the signal-to-noise ratio of the additive white Gaussian noise is 10dB.

[0026] Step 2: Use orthogonal dual channels to receive the RF signal and superimpose the IQ channels to obtain the baseband complex signal. Assume that the RF signal is r(t), and the baseband complex signal r obtained by orthogonal reception is baseband (t) can be expressed as:

[0027] r baseband (t)=r(t)cos(2πf0t)-j*r(t)sin(2πf0t)

[0028] Afterwards, baseband (t) Anti-aliasing filtering and downsampling: The passband width of the anti-aliasing low-pass filter is 40*10 6 Hz, the stopband width is 50*10 6 Hz, the order is 16; the downsampling multiple is 48, after downsampling r baseband (t) The sampling rate becomes 50*10 6 Hz.

[0029] Step 3: De-slope the downsampled signal, i.e. baseband (t) is mixed with the complex conjugate reference signal of the baseband linear frequency modulation transmission wave. The reference signal can be expressed as:

[0030]

[0031] The signal after de-skewing is:

[0032] s de-chirp (t) = r baseband (t)*s LO (t)

[0033] After the baseband signal is de-skewed, the short-time Fourier transform is used to obtain the time-frequency two-dimensional feature matrix, where the STFT window function uses a hamming window with a length of 200, a sliding step of 1, and the number of FFT points is set to 400. The time-frequency two-dimensional feature matrix can be expressed as:

[0034] S(t,f)=STFT[s de-chirp (t)]

[0035] Step 4: Convert the time-frequency two-dimensional feature matrix obtained in the previous step into dB format according to the amplitude, then convert it into a normalized grayscale image and send it to the image processing module. Use the dynamic threshold search algorithm to process it and obtain the target area matrix with the same size as the time-frequency two-dimensional feature matrix. The specific process of the algorithm is as follows:

[0036] ① Set the initial threshold of binarization to 0.90, and decrease it by 0.02 each time during the traversal process. Each round of traversal completes the following tasks.

[0037] ② Use the threshold at this time to binarize the normalized grayscale image based on STFT to obtain the result T1(t,f).

[0038] ③ Define a rectangular structure similar to the linear frequency modulation signal of the transmitted wave, and perform morphological filtering on T1(t,f) by first opening and then closing operations to obtain the result T2(t,f). The horizontal length of the rectangular structure is set to 0.8 times the length of the linear frequency modulation signal after de-skewing, and the vertical length is set to 3.

[0039] ④ Perform connected region analysis on T2(t,f). Find the connected regions whose length is within the range of 0.8-1.2 times the length of the de-skewing linear frequency modulation signal, and preliminarily determine them as the target echo area. Set these areas to 1 and the rest to 0 to obtain T3(t,f).

[0040] ⑤In the original time-frequency matrix S(t,f), index the target area according to T3(t,f), calculate the coefficient of variation of each area, and take the maximum value to compare with the coefficient of variation threshold.

[0041] ⑥ Set the threshold to 0.6. If the maximum value of the coefficient of variation is less than the threshold, record T3(t,f) and continue the next round of traversal; if the maximum value of the coefficient of variation is greater than or equal to the threshold, interrupt the traversal and take the T3(t,f) of the previous round as the final target area matrix T(t,f).

[0042] Step 5: Reconstruct the target based on the target area matrix. First, multiply the corresponding positions of the two matrices to obtain the time-frequency two-dimensional feature matrix S containing only the echo signal area. tar (t,f):

[0043] S tar (t,f)=S(t,f)*T(t,f)

[0044] Afterwards, the inverse short-time Fourier transform (ISTFT) is used to restore the time-frequency two-dimensional feature matrix to the time domain signal:

[0045] s tar (t) = ISTFT[S tar (t,f)]

[0046] Among them, the parameters of the inverse short-time Fourier transform need to be the same as those of the STFT, that is, the window function is selected as a hamming window with a length of 200, the sliding step is 1, the number of FFT points is set to 400, and the OLA method is used for signal recovery. Finally, the recovered signal is directly mixed with the baseband linear frequency modulation transmission wave, that is, the inverse de-skewing process is performed to achieve the target reconstruction. The baseband linear frequency modulation transmission wave can be expressed as:

[0047]

[0048] The final reconstructed signal is:

[0049] s rec (t) = s tar (t)*s LFM (t)

[0050] The above method is used to verify the effect of the intermittent sampling forwarding interference suppression method based on the time-frequency image dynamic threshold search strategy of the present invention. The comparison results of the pulse pressure diagram before and after interference suppression are as follows: Figure 2 As shown in the figure, in the pulse pressure diagram before interference suppression, the target signal is masked by strong interference; in the pulse pressure diagram after interference suppression, the interference signal almost completely disappears, and the target strength after reconstruction is about 50dB. It can be seen that the method used in the present invention has a significant suppression effect on intermittent sampling forwarding interference.

Claims

1. A radar interference suppression method based on a dynamic threshold search strategy for time-frequency images, characterized in that: The following steps are involved: S1. Get radar receiving data: The radar transmission signal is simulated as a linear frequency modulation signal using MATLAB, and the target echo signal and intermittent sampling forwarding interference are generated based on the linear frequency modulation signal simulation; The intermittent sampling forwarding interference includes direct forwarding interference, repeated forwarding interference and cyclic forwarding interference; The RF simulated radar receiving signal is generated by superposition of target echo, interference signal and Gaussian white noise; S2. Perform orthogonal reception and down-sampling of the RF received signal: For the simulated RF signal, mixing processing is performed to convert it into a baseband signal. The orthogonal dual-channel processing method is adopted to obtain the IQ two-way baseband signal and synthesize it into a baseband complex signal. The baseband signal is then processed by combining an anti-aliasing filter and downsampling to reduce its data rate. S3. De-skewing and time-frequency analysis of the baseband complex signal: For the baseband complex signal after downsampling, the frequency ramp effect is eliminated by de-skewing, and then the de-skewing signal is converted into a two-dimensional time-frequency matrix S(t,f) by short-time Fourier transform, and a normalized grayscale image is generated; S4. Extract the distribution area of ​​the target echo using the image processing module and the dynamic threshold search algorithm: The image processing module includes binarization, morphological filtering and connected area analysis. First, the grayscale image is binarized based on the current dynamic threshold. Then, a structural element with a shape similar to the linear frequency modulation signal of the transmitted wave is constructed in the time-frequency graph, and the binary image is subjected to morphological filtering with an opening operation followed by a closing operation. Finally, the processed image is subjected to connected area analysis to find areas with horizontal lengths close to the linear frequency modulation signal of the transmitted wave. These areas are preliminarily determined as the areas corresponding to the target echo, and the target area distribution map is converted into a target area matrix: The processing method of the dynamic threshold search algorithm is: traverse the binary threshold from high to low, each image processing can generate a target area matrix, index the area corresponding to the target echo according to T(t,f) in the time-frequency two-dimensional matrix S(t,f), and calculate the coefficient of variation of each area: Where x is the set of all values ​​contained in a target area in S(t,f), σ(x) is its standard deviation, and μ(x) is its mean. The coefficient of variation threshold that distinguishes the target echo and the background noise in the time-frequency two-dimensional matrix is ​​used to find the coefficient of variation threshold. The coefficient of variation of each target area is counted, and the maximum value is compared with the coefficient of variation threshold, and the highest round T(t,f) less than the threshold is found as the final target area matrix. S5. Reconstruct echo signals based on the target area matrix: The time-frequency two-dimensional matrix S(t,f) is multiplied with the target area matrix T(t,f) to obtain a time-frequency two-dimensional matrix containing only the target echo; the time-frequency domain signal is restored to the time domain signal using the inverse short-time Fourier transform; the time domain signal is inverse de-skewed to achieve the reconstruction of the target echo.

Images