A monopulse radar forward-looking imaging method based on Doppler domain super-resolution processing

Abstract

The present invention discloses a single pulse radar forward-looking imaging method based on Doppler domain super-resolution processing, which adopts sum and difference channels to receive data, performs range pulse compression on the signals of the two channels, and divides the frames; uses an iterative adaptive method based on minimum mean square error (MMSE) to perform super-resolution processing on each subframe of the sum channel data after pulse compression, and transforms the data into the range Doppler domain; uses the best weight coefficient of the sum channel data adaptive iterative estimation to perform the same processing on the difference channel data; performs single pulse angle measurement processing on the super-resolved sum and difference channel data by range-Doppler unit, and projects the sum beam energy to the corresponding range-azimuth unit to obtain a range-azimuth image; and then projects the image to the ground rectangular coordinate to obtain the final output. The present invention solves the problem of reduced resolution of single pulse imaging caused by angle measurement of multiple targets in the same beam, and improves the quality of radar forward-looking imaging.

Description

Technical Field

[0001] The invention belongs to the technical field of airborne monopulse radar adaptive filtering and array signal processing, and in particular relates to a monopulse radar forward-looking imaging method based on Doppler domain super-resolution processing. Background Art

[0002] In order to obtain two-dimensional high-resolution images, moving platform radars often use synthetic aperture radar (SAR) technology or Doppler beam sharpening (DBS) technology to achieve imaging. Synthetic aperture radar uses a single antenna with a smaller actual size to transmit pulse signals at a certain frequency during the movement of the carrier radar platform, and forms an equivalent long linear array by coherent accumulation of echo signals. DBS uses the different Doppler frequencies of targets in different azimuths to achieve the purpose of distinguishing targets in different azimuths. However, the existing SAR and DBS imaging cannot cover the front of the flight path. This is because the Doppler gradient of the echo received by SAR and DBS is almost zero in front of and below the moving path of the carrier radar, resulting in a sharp drop in azimuth resolution. In addition, the Doppler center frequency of the left-right symmetrical target echoes in front and below is the same, resulting in left-right blurring, thus forming the so-called imaging "blind area". The forward-looking area is an extremely important and sensitive area for the moving platform, which is of great significance to aircraft blind landing, terminal guidance, battlefield reconnaissance, etc.

[0003] In response to this situation, many scholars have conducted research on forward-looking imaging for airborne platforms. In the public technical solution, single-pulse imaging technology combines antenna scanning with single-pulse angle measurement, and uses angle measurement to achieve accurate target angle estimation, greatly improving the azimuth resolution. It has the advantages of low system structure complexity, strong real-time performance, and no special requirements for the flight track of the carrier platform. It is a practical forward-looking imaging technology. Single-pulse angle measurement technology has high measurement accuracy in theory, but when a single pulse measures the angle of multiple targets within the same beam width, there will be problems with false targets, resulting in deterioration of angle measurement performance and reduced resolution quality in the forward-looking area. In response to the above problems, how to achieve super-resolution in the Doppler domain, suppress the high side lobes caused by FFT and reduce the main lobe width, and then reduce the influence of Doppler side lobes on single-pulse angle measurement and energy projection, is still a key problem that needs to be solved in actual processing. Summary of the invention

[0004] Purpose of the invention: The present invention proposes a single pulse radar forward-looking imaging method based on Doppler domain super-resolution processing, which solves the problem of achieving super-resolution in the Doppler domain, overcomes the problem of high side lobes caused by Fourier transform by using an adaptive iterative method, and improves the quality of radar forward-looking imaging.

[0005] Technical solution: The present invention discloses a monopulse radar forward-looking imaging method based on Doppler domain super-resolution processing, comprising the following steps:

[0006] (1) Using the sum and difference channels to receive data, performing range pulse compression on the signals of the two channels and dividing them into frames;

[0007] (2) Super-resolution processing is performed on each subframe of the pulse compressed and channel data using an iterative adaptive method based on minimum mean square error (MMSE) to transform the data into the range Doppler domain;

[0008] (3) directly applying the optimal weight coefficients of the adaptive iterative estimation of the sum channel data in step (2) to the corresponding difference data, and also transforming it into the range Doppler domain;

[0009] (4) Performing single pulse angle measurement processing on the super-resolved sum and difference channel data in each range-Doppler unit, projecting the sum beam energy to the corresponding range-azimuth unit, and obtaining a range-azimuth image;

[0010] (5) The image is then projected to the ground rectangular coordinates to obtain the final output.

[0011] Furthermore, the implementation process of step (1) is as follows: two array elements are placed along the azimuth direction to independently receive echo data, the obtained data are added and subtracted to form sum and difference channel data, forming a typical sum and difference receiving system, and then the two sets of data are pulse compressed in the range direction;

[0012] There are K point targets ahead in the flight direction, with azimuth angles of θ = [θ1, θ2, …, θ k ]; then the echo signal vectors of the sum and difference channels are:

[0013]

[0014] Among them, a(θ k ) is the amplitude parameter of targets at different azimuths, which is directly proportional to the radar cross section (RCS). k ) and △(θ k ) are the sum and difference beam antenna patterns of the radar, s(t) is the M×1 transmission signal vector; the motion platform moves forward along the y-axis, and obtains two-dimensional echo sampling by scanning, performs range migration correction, and uses pulse compression technology to improve the range resolution, and finally obtains the two-dimensional data Y of the sum and difference channels Σ , Y △ ∈C M×P ;

[0015] The traditional matched filtering method is used to perform pulse compression on the echo signal in the frequency domain. The matched filter is the complex conjugate of the transmit signal spectrum, which is:

[0016] H(f)=S(f) * (2)

[0017] Among them, S(f) represents the spectrum of the transmitted signal, * represents the complex conjugate of the vector, and after the range pulse compression, N consecutive pulse echoes are used as a frame sub-image. Σ , Y △ Perform framing.

[0018] Furthermore, the implementation process of step (2) is as follows:

[0019] Taking N consecutive pulse echoes as a frame sub-image, when performing azimuth Q-point FFT on it, the Fourier transform basis vector of the qth Doppler frequency point is:

[0020]

[0021] Where △fd = PRF / Q is the uniform Doppler domain sampling interval determined by the pulse repetition frequency, and t = [t1, t2, …, t N ] is the azimuth slow time corresponding to the subframe, [·] T represents the matrix transpose; when the Doppler frequency difference between different targets is greater than △fd, Doppler processing can distinguish the targets and achieve the ability of multi-target resolution;

[0022] The discrete signal model of the matched filter output is:

[0023]

[0024] in, is the estimated value of the lth Doppler unit output by the matched filter, is the N consecutive samples in the same range gate as the channel echo, l = 1, 2, ..., Q, [i] H represents the complex conjugate transpose of a matrix;

[0025] The iterative algorithm based on the minimum mean square error (MMSE) estimates the optimal filter coefficient ω(l) for each Doppler unit. The estimated value of the lth Doppler unit is:

[0026]

[0027] Based on the MMSE criterion, the cost function is constructed:

[0028]

[0029] Among them, E[i] represents the expectation, x(l) represents the azimuthal impulse response; in order to minimize the estimated error, J(l) calculates the gradient of ω(l) and sets it to zero, that is:

[0030]

[0031] Assuming that the impulse responses of each Doppler unit are uncorrelated, that is, E[x(m)x(n)]=0, m≠n, and are statistically independent of the noise, equation (7) is transformed into:

[0032]

[0033]

[0034] where |x(l)| 2 represents the expected power, Represents the power of Gaussian white noise in the echo signal, I N Represents an N×N dimensional identity matrix; at the same time, an iterative algorithm is used, and the matching output of the previous iteration is used as the input of the next iteration.

[0035] Furthermore, the implementation process of step (4) is as follows:

[0036] Construct the directional characteristic relationship:

[0037]

[0038] Among them, S ∑ (n,m),S △ (n,m) are the sum and difference signals of the nth range unit and the mth Doppler unit, respectively. θ represents the azimuth relative to the beam centerline direction. d represents the phase center interval of the dual channels of the monopulse antenna. Each range-Doppler unit performs a monopulse angle measurement process on each storage unit signal. Since the variation range of θ is very small within the main lobe range, the approximate relationship of sin(θ)≈θ holds true. Therefore:

[0039]

[0040] According to formula (11), the azimuth angle θ of the ground target corresponding to each storage unit signal in the real beam is obtained, the signal is relocated, and the channel signal is accumulated according to the beam scanning speed to complete the energy projection from the range Doppler domain to the range-azimuth domain; the above operation is performed on each frame sub-image in turn to obtain the image of the scanned area.

[0041] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects: The present invention combines antenna scanning with single-pulse angle measurement, and uses sum and difference channels to independently receive echo signals. Only one pulse echo is needed to provide all information about the target's azimuth position; sufficiently high range resolution is obtained through range pulse compression technology, and the adaptive iterative algorithm based on the minimum mean square error criterion estimates the optimal weight coefficient for each range-Doppler unit, thereby overcoming the high side lobes caused by the FFT method and reducing the main lobe width, improving the resolution of adjacent targets and reducing the impact of false targets on single-pulse angle measurement. For multiple targets that are difficult to distinguish in the same beam, the super-resolution method can be used to achieve primary resolution in the Doppler domain, and secondary resolution can be achieved through single-pulse angle measurement, which is expected to improve the resolution of multiple targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flow chart of the present invention;

[0043] Figure 2 It is a simplified diagram of the spatial geometric relationship of the data received in the forward oblique direction of the scanning radar;

[0044] Figure 3 It is the data of point target simulation experiment and channel distance pulse pressure;

[0045] Figure 4 It is the data after the difference channel distance pulse pressure of the point target simulation experiment;

[0046] Figure 5 This is the imaging result of the single pulse technology in the point target simulation experiment;

[0047] Figure 6 This is the imaging result of the point target simulation experiment based on DBS single pulse technology;

[0048] Figure 7 This is the imaging result obtained after the point target simulation experiment is processed by this method;

[0049] Figure 8 It is the first set of original SAR background images;

[0050] Fig. 9 This is the single pulse imaging result diagram of the first set of data;

[0051] Fig.10 This is the imaging result obtained after the first set of data is processed by this method;

[0052] Fig.11 is the second set of original SAR background images;

[0053] Fig.12 This is the result of single pulse imaging of the second set of data;

[0054] Fig.13This is the imaging result obtained after the second set of data is processed by this method. DETAILED DESCRIPTION

[0055] The present invention is further described in detail below with reference to the accompanying drawings.

[0056] The present invention proposes a single pulse radar forward imaging method based on Doppler domain super-resolution processing, such as Figure 1 As shown, the specific steps include:

[0057] Step 1: Use the sum and difference channels to receive echo data, perform range pulse compression processing on the signals of the two channels, and divide them into frames.

[0058] The single pulse technology uses overlapping antenna beams to receive multiple signals. Only one pulse echo is needed to provide all the information about the target's position. The geometric model of the scanning radar's front squint direction is as follows: Figure 2 As shown. The single pulse radar mounted on the moving platform moves at an angular velocity v ω The scene in a smaller forward-looking range is scanned and a linear frequency modulation signal (LFM) is transmitted at a fixed pulse repetition frequency (PRF).

[0059] Assume that there are K point targets ahead in the flight direction, with azimuth angles θ=[θ1,θ2,…,θ k ], then the echo signal vectors of the sum and difference channels can be expressed as:

[0060]

[0061] Among them, a(θ k ) is the amplitude parameter of the target at different azimuth points, which is directly proportional to the RCS. k ) and △(θ k ) are the sum and difference beam antenna patterns of the radar, and s(t) is the M×1 transmission signal vector. The motion platform moves forward along the y-axis and obtains two-dimensional echo sampling by scanning, performs range migration correction, and uses pulse compression technology to improve the range resolution. Finally, the two-dimensional data Y of the sum and difference channels is obtained. Σ , Y △ ∈C M×P .

[0062] The traditional matched filtering (MF) method is used to perform pulse compression on the echo signal in the frequency domain. The matched filter is the complex conjugate of the transmit signal spectrum, which can be expressed as:

[0063] H(f)=S(f) * (2)

[0064] Where S(f) represents the spectrum of the transmitted signal, and * represents the complex conjugate of the vector. After the range pulse compression, take N consecutive pulse echoes as a frame sub-image, and Σ , Y △ Perform framing.

[0065] Step 2: Super-resolution processing is performed on each subframe of the pulse compressed and channel data using an iterative adaptive method based on minimum mean square error (MMSE) to transform the data into the range Doppler domain.

[0066] Taking N consecutive pulse echoes as a frame sub-image, when performing azimuth Q-point FFT on it, the Fourier transform basis vector of the qth Doppler frequency point is:

[0067]

[0068] Where △fd = PRF / Q is the uniform Doppler domain sampling interval determined by the pulse repetition frequency, and t = [t1, t2, …, t N ] is the azimuth slow time corresponding to the subframe, [i] T When the Doppler frequency difference between different targets is greater than △fd, Doppler processing can distinguish the targets and achieve the ability to distinguish multiple targets.

[0069] The basis vector of Fourier transform is used as the matched filter for each frequency point, and the output of the matched filter can be expressed as the convolution of the echo signal and the basis vector. Therefore, the discrete signal model of the matched filter output can be expressed as:

[0070]

[0071] in, is the estimated value of the lth Doppler unit output by the matched filter, is the N consecutive samples in the same range gate as the channel echo, l = 1, 2, ..., Q, [i] H Represents the complex conjugate transpose of a matrix.

[0072] The MMSE-based iterative algorithm adaptively estimates the optimal filter coefficient ω(l) for each Doppler unit to replace the s in the matched filter output formula (4). l , so that the side lobe interference can be reduced as much as possible, making The estimated value of the lth Doppler unit is:

[0073]

[0074] Based on the MMSE criterion, the cost function is constructed:

[0075]

[0076] Among them, E[i] represents the expectation, and x(l) represents the azimuthal impulse response. In order to minimize the estimated error, J(l) calculates the gradient of ω(l) and sets it to zero, that is:

[0077]

[0078] Assuming that the impulse responses of each Doppler unit are uncorrelated, that is, E[x(m)x(n)]=0, m≠n, and are statistically independent of the noise, equation (7) can be transformed into:

[0079]

[0080]

[0081] where |x(l)| 2 represents the expected power, Represents the power of Gaussian white noise in the echo signal, I N Represents the N×N dimensional unit matrix. At the same time, an iterative algorithm is used, and the matching output of the previous iteration is used as the input of the next iteration. Experiments show that generally, 2 to 3 iterations can well reduce the azimuth sidelobes, thereby achieving the best estimate of x(l). The estimated value after FFT is usually used as the initial value of the first iteration of the adaptive algorithm.

[0082] Step 3: Perform the same processing on the difference channel data using the optimal weight coefficient estimated by adaptive iteration of the channel data, and also transform it into the range Doppler domain.

[0083] In order not to affect the accuracy of the single pulse angle measurement in the next step, the optimal weight coefficients adaptively iteratively estimated for the sum channel data in step 2 are directly applied to the corresponding difference data, and they are also transformed into the range Doppler domain.

[0084] Step 4: Perform single pulse angle measurement processing on the super-resolved sum and difference channel data for each range-Doppler unit, project the sum beam energy to the corresponding range-azimuth unit, and obtain a range-azimuth image.

[0085] The single pulse technology is used to accurately reposition the echo signal of each storage unit corresponding to the sum and difference channel data in the azimuth direction, further improving the azimuth accuracy of the image.

[0086] Construct the directional characteristic relationship:

[0087]

[0088] In the formula, S ∑ (n,m),S △(n,m) are the sum and difference signals of the nth range unit and the mth Doppler unit, respectively. θ represents the azimuth relative to the beam centerline direction, and d represents the phase center interval of the two channels of the monopulse antenna. The range-Doppler unit performs monopulse angle measurement processing on each storage unit signal.

[0089] Since the variation range of θ is very small within the main lobe range, the approximate relationship of sin(θ)≈θ holds true, so:

[0090]

[0091] According to the above formula, the azimuth angle θ of the ground target corresponding to each storage unit signal in the real beam is obtained. The signal is relocated according to the obtained azimuth angle value, and then the signal after single pulse processing is accumulated according to the beam scanning speed to complete the energy projection from the range Doppler domain to the range-azimuth domain. The above operation is performed on each frame sub-image in turn to obtain the image of the scanned area.

[0092] Step 5: Project the image to the ground rectangular coordinates to obtain the final output result.

[0093] According to the range-azimuth image obtained in step 4 and the corresponding geometric relationship, it is projected into the ground rectangular coordinates and converted into the field of view image that people are accustomed to, and the final output can be obtained.

[0094] The single pulse radar forward-looking imaging method based on Doppler domain super-resolution processing proposed in the present invention is verified by simulation experiments, and the experimental results fully demonstrate the effectiveness of the method of the present invention.

[0095] Two point targets are set in the forward-looking area of ​​the radar and are located in the same beam width of the same range gate. The radar transmits a linear frequency modulation signal and obtains high-resolution in the range direction through pulse compression processing at the receiving end. The system parameters of the radar during simulation are: system bandwidth is 7.5MHz, pulse repetition frequency (PRF) is 1000Hz, pulse width is 5us, center frequency is 9.6GHz, beam width is 3°, aircraft flight altitude is 4Km, aircraft flight speed is 100m / s, sampling frequency is 10MHz, and the distance between the two point targets is 180Km, centered on the y-axis, and distributed at azimuths of 3° and 5°. The radar sum and difference channel receiving system is simulated, with a beam scanning range of -15° to 15°, 16 dwell pulses, and each step is 0.3°. The echo data generated by the simulation is processed using the method of the present invention, and the data after distance pulse compression of the sum and difference channels are as follows Figure 3 , Figure 4 The results after processing are shown as Figure 7As shown in the figure, single pulse imaging and DBS-based single pulse imaging are used as a comparison. It can be seen that the method of the present invention can effectively improve the imaging quality of single pulse and reduce the influence of false targets. Figure 5 , 6 It can be seen that both the single pulse imaging and DBS-based single pulse imaging results have false targets. This is because the Doppler sidelobe of the point target is high, causing the energy of the point target to be dispersed into adjacent azimuth units. Although the single pulse imaging based on Doppler estimation has some improvements compared to the single pulse technology, the phenomenon of false targets still exists. The method of the present invention can suppress the high sidelobes caused by FFT and reduce the main lobe width in the first resolution of the signal from the time domain to the frequency domain, and then use the single pulse angle measurement for secondary resolution, weakening the influence of false targets on energy projection.

[0096] At the same time, two SAR images were selected as the initial scenes for verification. The system bandwidth is 40MHz, the pulse repetition frequency (PRF) is 1000Hz, the pulse width is 6us, the sampling frequency is 50MHz, the wavelength is 0.03 meters, the aircraft flight altitude is set to 0, the aircraft flight speed is 80m / s, and the imaging scene center is 5Km. Centered on the y-axis, the beam scanning range is -12° to 12°, the dwell pulse is 16, and each step is 0.5°. The first set of original SAR background images is shown in Figure 1. Figure 8 As shown, the single pulse imaging results are as follows Fig. 9 As shown, the results obtained after processing by this inventive method are as follows Fig.10 The second set of original SAR background images is shown in Fig.11 As shown, the single pulse imaging results are as follows Fig.12 The processing results are shown as Fig.13 As shown in the figure, it can be seen from the processing results that this method can achieve a relatively ideal effect in suppressing the Doppler sidelobe of the target, reduce its influence on the energy projection, and thus further improve the azimuth resolution and obtain a clearer image of the terrain features, which verifies the effectiveness of this method.

Claims

1. A single pulse radar forward imaging method based on Doppler domain super-resolution processing, characterized in that: The following steps are involved: (1) Using the sum and difference channels to receive echo data, performing range pulse compression processing on the signals of the two channels and dividing them into frames; (2) Super-resolution processing is performed on each subframe of the pulse compressed and channel data using an iterative adaptive method based on minimum mean square error to transform the data into the range Doppler domain; (3) directly applying the optimal weight coefficients of the adaptive iterative estimation of the sum channel data in step (2) to the corresponding difference data, and also transforming it into the range Doppler domain; (4) Performing single pulse angle measurement processing on the super-resolved sum and difference channel data in each range-Doppler unit, projecting the sum beam energy to the corresponding range-azimuth unit, and obtaining a range-azimuth image; (5) Project the image to the ground rectangular coordinates to obtain the final output result; The implementation process of step (1) is as follows: two array elements are placed along the azimuth direction to independently receive echo data, and the obtained data are added and subtracted to form sum and difference channel data, forming a typical sum and difference receiving system, and then the two sets of data are pulse compressed in the range direction and framed; There are K point targets ahead in the flight direction, with azimuth angles of θ = [θ1, θ2, …, θ k ]; then the echo signal vectors of the sum and difference channels are: Among them, a(θ k ) is the amplitude parameter of targets at different azimuths, which is directly proportional to the radar cross section (RCS). k ) and Δ(θ k ) are the sum and difference beam antenna patterns of the radar, s(t) is the M×1 transmission signal vector; the motion platform moves forward along the y-axis, and obtains two-dimensional echo sampling by scanning, performs range migration correction, and uses pulse compression technology to improve the range resolution, and finally obtains the two-dimensional data Y of the sum and difference channels Σ , Y Δ ∈C M×P ; The traditional matched filtering method is used to perform pulse compression on the echo signal in the frequency domain. The matched filter is the complex conjugate of the transmit signal spectrum, which is: H(f)=S(f) * (2) Among them, S(f) represents the spectrum of the transmitted signal, * represents the complex conjugate of the vector, and after the range pulse compression, N consecutive pulse echoes are used as a frame sub-image. Σ , Y Δ Perform framing; The implementation process of step (2) is as follows: Taking N consecutive pulse echoes as a frame sub-image, when performing azimuth Q-point FFT on it, the Fourier transform basis vector of the qth Doppler frequency point is: Where Δfd = PRF / Q is the uniform Doppler domain sampling interval determined by the pulse repetition frequency, and t = [t1, t2, ..., t N ] is the azimuth slow time corresponding to the subframe, [·] T represents the matrix transpose; when the Doppler frequency difference between different targets is greater than Δfd, Doppler processing can distinguish the targets and achieve the ability to distinguish multiple targets; The discrete signal model of the matched filter output is: in, is the estimated value of the lth Doppler unit output by the matched filter, is N consecutive samples within the same range gate as the channel echo, l = 1, 2, ..., Q, [·] H represents the complex conjugate transpose of a matrix; The iterative algorithm based on the minimum mean square error estimates the optimal filter coefficient ω(l) for each Doppler unit. The estimated value of the lth Doppler unit is: Based on the MMSE criterion, the cost function is constructed: Among them, E[·] represents the expectation, x(l) represents the azimuthal impulse response; in order to minimize the estimated error, J(l) calculates the gradient of ω(l) and sets it to zero, that is: Assuming that the impulse responses of each Doppler unit are uncorrelated, that is, E[x(m)x(n)]=0, m≠n, and are statistically independent of the noise, equation (7) is transformed into: where |x(l)| 2 represents the expected power, Represents the power of Gaussian white noise in the echo signal, I N represents the N×N dimensional identity matrix; at the same time, an iterative algorithm is used, and the matching output of the previous iteration is used as the input of the next iteration; The implementation process of step (4) is as follows: Construct the directional characteristic relationship: Among them, S Σ (n,m),S Δ (n,m) are the sum and difference signals of the nth range unit and the mth Doppler unit, respectively. θ represents the azimuth relative to the beam centerline direction. d represents the phase center interval of the dual channels of the monopulse antenna. Each range-Doppler unit performs a monopulse angle measurement process on each storage unit signal. Since the variation range of θ is very small within the main lobe range, the approximate relationship of sin(θ)≈θ holds true. Therefore: According to formula (11), the azimuth angle θ of the ground target corresponding to each storage unit signal in the real beam is obtained, the signal is repositioned, and the signal after single pulse processing is accumulated according to the beam scanning speed to complete the energy projection from the range Doppler domain to the range-azimuth domain; the above operation is performed on each frame sub-image in turn to obtain the image of the scanned area.

Images