A method for suppressing strong ground clutter based on three-level windowing filtering

Abstract

The present invention discloses a method for suppressing strong ground clutter based on three-level windowing filtering, comprising step S1: obtaining coherently accumulated radar echo data, performing data rearrangement and pulse compression, and obtaining a signal matrix to be measured in the range-slow time domain; step S2: performing windowed mean filtering on the data to be measured in the slow time dimension; step S3: performing windowed Doppler filtering on the data after the windowed mean filtering in the slow time dimension to obtain data in the range-Doppler domain; step S4: convolving the data after the windowed Doppler filtering with a two-dimensional Gaussian smoothing filter to obtain the data after the Gaussian filtering, and completing the ground clutter suppression based on three-level windowing filtering. The present invention suppresses strong ground clutter in both the time domain and the frequency domain through windowed mean filtering, windowed Doppler filtering and two-dimensional Gaussian filtering, effectively controls the detection probability and false alarm probability, and improves the detection probability of targets with low speed and small RCS under the background of strong ground clutter.

Description

Technical Field

[0001] The invention relates to the technical field of radar signal processing, and in particular to a method for suppressing strong ground clutter based on three-level windowing filtering. Background Art

[0002] Radar faces many problems when detecting targets in a strong ground clutter background:

[0003] 1. The targets detected by radar have the characteristics of small RCS, slow speed and low altitude;

[0004] 2. The ground environment is complex. Scattering bodies such as mountains, trees, grass, and buildings will generate ground clutter. Small targets will be submerged in strong ground clutter, resulting in poor radar detection performance for small targets.

[0005] 3. Affected by wind speed, radar beam scanning, etc., the ground clutter spectrum will be broadened and the signal-to-clutter ratio of low-speed targets will be reduced, resulting in poor detection performance of radar for low-speed targets.

[0006] In traditional radar signal processing, the method of moving target display MTI + moving target detection MTD is usually used to achieve target detection in a strong ground clutter background. However, the notch produced by the pulse cancellation filter is relatively wide. As the cancellation level increases, the zero-frequency suppression capability is stronger, but the notch is also wider, which will reduce the signal-to-noise ratio of low-speed targets while suppressing ground clutter. At the same time, the side lobes of MTD are high without windowing, and the side lobes of strong ground clutter may also submerge low-speed targets, affecting target detection performance. In addition, the traditional method only considers the suppression of ground clutter at zero frequency, and lacks the ability to suppress moving ground clutter with widened clutter spectrum. Summary of the invention

[0007] In order to overcome the shortcomings in the prior art, the present invention provides a strong ground clutter suppression method based on three-level windowing filtering, which suppresses strong ground clutter in both time domain and frequency domain through windowing mean filtering, windowing Doppler filtering and two-dimensional Gaussian filtering, effectively controls the detection probability and false alarm probability, and improves the detection probability of targets with low speed and small RCS under the background of strong ground clutter.

[0008] In order to achieve the above-mentioned invention object, the technical solution adopted to solve the technical problem is as follows:

[0009] A method for suppressing strong ground clutter based on three-level windowing filtering comprises the following steps:

[0010] Step S1: obtaining coherently accumulated radar echo data, performing data rearrangement and pulse compression, and obtaining a signal matrix to be measured in the range-slow time domain;

[0011] Step S2: performing windowed mean filtering on the data to be measured in the slow time dimension;

[0012] Step S3: performing windowed Doppler filtering on the data after windowed mean filtering in the slow time dimension to obtain data in the range-Doppler domain;

[0013] Step S4: convolve the windowed Doppler filtered data with a two-dimensional Gaussian smoothing filter to obtain Gaussian filtered data, thereby completing ground clutter suppression based on three-level windowing filtering.

[0014] Furthermore, the step S1 specifically includes the following steps:

[0015] Step S11: Obtain radar echo data s after coherent accumulation under strong ground clutter background r (n t ), n t is fast time dimension sampling, n t =1, 2, ..., N*M, N is a pulse repetition period T r The number of sampling points in, M is the number of pulses of coherent accumulation;

[0016] Step S12: processing the radar echo data s r (n t ) to rearrange the data and obtain the data matrix s with M rows and N columns r (n a(m) , n r(n) ), where n a is the slow time dimension, m=1,2,...,M,n r is the fast time dimension, n=1, 2, ..., N;

[0017] Step S13: The radar echo data matrix s r (n a(m) , n r(n) ) is pulse compressed to obtain the signal matrix s in the range-slow time domain. pc (n a(m) , r (n) ), where r is the distance value corresponding to each sampling point after pulse compression.

[0018] Furthermore, the step S2 specifically includes the following steps:

[0019] Step S21: Generate a set of window function coefficients win1(m), where m=1, 2, ..., M, and the length of the window function is the same as the measured signal matrix s pc (n a(m) , r (n) ) have the same length of slow time dimension;

[0020] Step S22: Normalize the window function coefficient win1(m) to obtain the normalized window function coefficient winNOR (m), the formula is:

[0021]

[0022] Step S23: The signal matrix s to be tested pc (n a(m) , r (n) ) each column of the same distance unit signal sequence respectively uses the normalization coefficient win NOR (m) weighted sum to get the filtered value The formula is:

[0023]

[0024] Step S24: The signal matrix s to be tested pc (n a(m) , r (n) ) is subtracted from the corresponding filter value of the signal sequence of each column of the same distance unit to obtain the primary filtered data s after the windowed mean filter suppresses the ground clutter. 1fil (n a(m) , r (n) ).

[0025] Preferably, the types of the window function coefficient win1(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window.

[0026] Furthermore, the step S3 specifically includes the following steps:

[0027] Step S31: Generate a set of window function coefficients win2(m), where m=1, 2, ..., M, and the window function length is the same as the primary filtering data matrix s 1fil (n a(m) , r (n) ) have the same length of slow time dimension;

[0028] Step S32: filtering the primary filtered data s 1fil (n a(m) , r (n) ) uses the window function coefficient win2(m) to perform slow time dimension time domain windowing on the signal sequence of each column of the same distance unit, and obtains the windowed data matrix s′ 1fil (n a(m) , r (n) ), the formula is:

[0029] s′ 1fil (n a(m) , r (n) )=s 1fil (na(m) , r (n) )win2(m);

[0030] Step S33: Windowing the slow-time dimension data s′ 1fil (n a(m) , r (n) ) Perform FFT operation in the slow time dimension to obtain the secondary filtering data s after Doppler filtering 2fil (n a(m) , r (n) ).

[0031] Preferably, the types of the window function coefficient win2(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window.

[0032] Furthermore, the step S4 specifically includes the following steps:

[0033] Step S41: Generate a set of two-dimensional Gaussian function matrices G(x, y), which are square matrices with odd rows and odd columns, and the formula is:

[0034]

[0035] Among them, σ is the smoothness parameter. The larger σ is, the wider the frequency band of the Gaussian filter is, and the better the filtering smoothness is. gs -1) / 2∶1∶(M gs -1) / 2,M gs is the number of rows, y=-(N gs -1) / 2∶1∶(N gs -1) / 2,N gs is the number of columns;

[0036] Step S42: Normalize the Gaussian function matrix G(x, y) to obtain the Gaussian kernel G core (x, y), the formula is:

[0037]

[0038] Step S43: Move the Gaussian kernel G core (x, y), so that its central element is located in the secondary filtering data matrix s 2fil (n a(m) , r (n) ) is placed directly above the unit to be processed, and the overlapping units of the two two-dimensional matrices are multiplied and added in sequence to obtain the filtering value of the unit to be processed. For all the secondary filtering data matrices s 2fil (n a(n) , r (n)) units repeat this step to obtain the ground clutter suppression result s after three-level windowing filtering 3fil (n a(m) , r (n) ).

[0039] Due to the adoption of the above technical solution, the present invention has the following advantages and positive effects compared with the prior art:

[0040] 1. The present invention forms a narrower and more jittery notch at the zero frequency by combining windowed mean filtering with windowed Doppler filtering, compared with the traditional MTI combined with Doppler filtering, reducing the spectrum broadening of strong ground clutter and avoiding the amplitude of small targets being submerged. It not only suppresses the ground clutter near the zero frequency, but also avoids the loss of signal-to-noise ratio of low-speed targets, thereby improving the detection probability of low-altitude and low-speed small targets;

[0041] 2. The present invention uses two-dimensional Gaussian filtering to further filter out ground clutter in the range-Doppler domain, compared with the traditional ground clutter suppression algorithm performed in the time domain, using the power spectrum distribution characteristics of strong ground clutter. The method of the present invention can effectively reduce false alarms caused by strong ground clutter. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. 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 work. In the drawings:

[0043] Figure 1 The present invention is a flowchart of a method for suppressing strong ground clutter based on three-level windowing filtering. DETAILED DESCRIPTION

[0044] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] like Figure 1 As shown, this embodiment discloses a method for suppressing strong ground clutter based on three-level windowing filtering, comprising the following steps:

[0046] Step S1: obtaining coherently accumulated radar echo data, performing data rearrangement and pulse compression, and obtaining a signal matrix to be measured in the range-slow time domain;

[0047] Specifically, step S1 includes the following steps:

[0048] Step S11: Obtain radar echo data s after coherent accumulation under strong ground clutter background r (n t ), n t is fast time dimension sampling, n t =1, 2, ..., N*M, N is a pulse repetition period T r The number of sampling points in, M is the number of pulses of coherent accumulation;

[0049] Step S12: processing the radar echo data s r (n t ) to rearrange the data and obtain the data matrix s with M rows and N columns r (n a(m) , n r(n) ), where n a is the slow time dimension, m=1,2,...,M,n r is the fast time dimension, n=1, 2, ..., N;

[0050] Step S13: The radar echo data matrix s r (n a(m) , n r(n) ) is pulse compressed to obtain the signal matrix s in the range-slow time domain. pc (n a(m) , r (n) ), where r is the distance value corresponding to each sampling point after pulse compression.

[0051] Step S2: performing windowed mean filtering on the data to be measured in the slow time dimension;

[0052] Specifically, step S2 includes the following steps:

[0053] Step S21: Generate a set of window function coefficients win1(m), where m=1, 2, ..., M, and the window function length is the same as the signal matrix s to be measured. pc (n a(m) , r (n) ) have the same length of slow time dimension;

[0054] Preferably, the types of the window function coefficient win1(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window. In this embodiment, win1(m) is set to a Kaiser window, which is used to control the parameter β=10 of the relative sidelobe attenuation. For a given β, the relative sidelobe attenuation is fixed relative to the window length.

[0055] Step S22: Normalize the window function coefficient win1(m) to obtain the normalized window function coefficient win NOR (m), the formula is:

[0056]

[0057] Step S23: The signal matrix s to be tested pc (n a(m) , r (n) ) each column of the same distance unit signal sequence respectively uses the normalization coefficient win NOR (m) weighted sum to get the filtered value The formula is:

[0058]

[0059] Step S24: The signal matrix s to be tested pc (n a(m) , r (n) ) is subtracted from the corresponding filter value of the signal sequence of each column of the same distance unit to obtain the primary filtered data s after the windowed mean filter suppresses the ground clutter. 1fil (n a(m) , r (n) ).

[0060] Step S3: performing windowed Doppler filtering on the data after windowed mean filtering in the slow time dimension to obtain data in the range-Doppler domain;

[0061] Specifically, step S3 includes the following steps:

[0062] Step S31: Generate a set of window function coefficients win2(m), where m=1, 2, ..., M, and the window function length is the same as the primary filtering data matrix s 1fil (n a(m) , r (n) ) have the same length of slow time dimension;

[0063] Preferably, the types of the window function coefficient win2(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window. In this embodiment, win2(m) is set to be a Hamming window.

[0064] Step S32: filtering the primary filtered data s 1fil (n a(m) , r (n) ) uses the window function coefficient win2(m) to perform slow time dimension time domain windowing on the signal sequence of each column of the same distance unit, and obtains the windowed data matrix s′1fil (na (m) , r (n) ), the formula is:

[0065] s′ 1fil (n a(m) , r (n) )=s 1fil (n a(m) , r (n) )win2(m);

[0066] Step S33: Windowing the slow-time dimension data s′ 1fil (n a(m) , r (n) ) Perform FFT operation in the slow time dimension to obtain the secondary filtering data s after Doppler filtering 2fil (n a(m) , r (n) ).

[0067] Step S4: convolve the windowed Doppler filtered data with a two-dimensional Gaussian smoothing filter to obtain Gaussian filtered data, thereby completing ground clutter suppression based on three-level windowing filtering.

[0068] Specifically, step S4 includes the following steps:

[0069] Step S41: Generate a set of two-dimensional Gaussian function matrices G(x, y), which are square matrices with odd rows and odd columns, and the formula is:

[0070]

[0071] Among them, σ is the smoothness parameter. The larger σ is, the wider the frequency band of the Gaussian filter is, and the better the filtering smoothness is. gs -1) / 2∶1∶(M gs -1) / 2,M gs is the number of rows, y=-(N gs -1) / 2∶1∶(N gs -1) / 2,N gs is the number of columns;

[0072] Step S42: Normalize the Gaussian function matrix G(x, y) to obtain the Gaussian kernel G core (x, y), the formula is:

[0073]

[0074] Step S43: Move the Gaussian kernel G core (x, y), so that its central element is located in the secondary filtering data matrix s 2fil (n a(m) , r(n) ) is placed directly above the unit to be processed, and the overlapping units of the two two-dimensional matrices are multiplied and added in sequence to obtain the filtering value of the unit to be processed. For all the secondary filtering data matrices s 2fil (n a(m) , r (n) ) units repeat this step to obtain the ground clutter suppression result s after three-level windowing filtering 3fil (n a(m) , r (n) ).

[0075] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for suppressing strong ground clutter based on three-level windowing filtering, characterized in that: The steps include: Step S1: obtaining coherently accumulated radar echo data, performing data rearrangement and pulse compression, and obtaining a signal matrix to be measured in the range-slow time domain; The step S1 specifically includes the following steps: Step S11: Obtain radar echo data s after coherent accumulation under strong ground clutter background r (n t ), n t is fast time dimension sampling, n t =1,2,…,N*M, N is a pulse repetition period T r The number of sampling points in, M is the number of pulses of coherent accumulation; Step S12: processing the radar echo data s r (n t ) to rearrange the data and obtain the data matrix s with M rows and N columns r (n a(m) ,n r(n) ), where n a is the slow time dimension, m=1,2,…,M,n r is the fast time dimension, n=1,2,…,N; Step S13: Radar echo data matrix s r (n a(m) ,n r(n) ) is pulse compressed to obtain the signal matrix s in the range-slow time domain. pc (n a(m) ,r (n) ), where r is the distance value corresponding to each sampling point after pulse compression; Step S2: performing windowed mean filtering on the data to be measured in the slow time dimension; The step S2 specifically includes the following steps: Step S21: Generate a set of window function coefficients win1(m), where m=1, 2, ..., M, and the length of the window function is the same as the measured signal matrix s pc (n a(m) ,r (n) ) have the same length of slow time dimension; Step S22: Normalize the window function coefficient win1(m) to obtain the normalized window function coefficient win NOR (m), the formula is: Step S23: The signal matrix s to be tested pc (n a(m) ,r (n) ) uses the normalized window function coefficient win NOR (m) weighted sum to get the filtered value The formula is: Step S24: The signal matrix s to be tested pc (n a(m) ,r (n) ) is subtracted from the corresponding filter value of the signal sequence of each column of the same distance unit to obtain the primary filter data matrix s after the windowed mean filter suppresses the ground clutter. 1fil (n a(m) ,r (n) ); Step S3: performing windowed Doppler filtering on the data after windowed mean filtering in the slow time dimension to obtain data in the range-Doppler domain; Step S4: convolve the windowed Doppler filtered data with a two-dimensional Gaussian smoothing filter to obtain Gaussian filtered data, thereby completing ground clutter suppression based on three-level windowing filtering.

2. The method for suppressing strong ground clutter based on three-level windowing filtering according to claim 1 is characterized in that: The types of the window function coefficient win1(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window.

3. The method for suppressing strong ground clutter based on three-level windowing filtering according to claim 1 is characterized in that: The step S3 specifically includes the following steps: Step S31: Generate a set of window function coefficients win2(m), where m = 1, 2, ..., M, the window function length and the primary filtering data matrix s 1fil (n a(m) ,r (n) ) have the same length of slow time dimension; Step S32: filter the data matrix s 1fil (n a(m) ,r (n) ) are respectively windowed in the slow time dimension using the window function coefficient win2(m) to obtain the windowed data matrix s' 1fil (n a(m) ,r (n) ), the formula is: s’ 1fil (n a(m) ,r (n) )=s 1fil (n α(m) ,r (n) )win2(m); Step S33: Calculate the windowed data matrix s' in the slow time dimension 1fil (n α(m) ,r (n) ) Perform FFT operation in the slow time dimension to obtain the secondary filtering data matrix s after Doppler filtering 2fil (n α(m) ,r (n) ).

4. The method for suppressing strong ground clutter based on three-level windowing filtering according to claim 3 is characterized in that: The types of the window function coefficient win2(m) include: Kaiser window, Hamming window, Hanning window, rectangular window, flat-top window, Blackman window, Gaussian window, triangular window or Chebyshev window.

5. The method for suppressing strong ground clutter based on three-level windowing filtering according to claim 3 is characterized in that: The step S4 specifically includes the following steps: Step S41: Generate a set of two-dimensional Gaussian function matrices G(x,y), which are square matrices with odd rows and odd columns, and the formula is: Among them, σ is the smoothness parameter, x = -(M gs -1) / 2:1:(M gs -1) / 2,M gs is the number of rows, y=-(N gs -1) / 2:1:(N gs -1) / 2,N gs is the number of columns; Step S42: Normalize the Gaussian function matrix G(x, y) to obtain the Gaussian kernel G core (x,y), the formula is: Step S43: Move the Gaussian kernel G core (x, y), so that its central element is located in the secondary filtering data matrix s 2fil (n a(m) ,r (n) ) is placed directly above the unit to be processed, and the overlapping units of the two two-dimensional matrices are multiplied and added in sequence to obtain the filtering value of the unit to be processed. For all the secondary filtering data matrices s 2fil (n a(m) ,r (n) ) units repeat this step to obtain the ground clutter suppression result s after three-level windowing filtering 3fil (n a(m) ,r (n) ).

Images