CN103412289A - Radar high-resolution distance imaging method based on simulation sinc kernel sampling - Google Patents

Abstract

The invention discloses a radar high-resolution distance imaging method based on simulation sinc kernel sampling. The problems that the sampling rate of an existing broadband radar signal is too high, and the signal processing data size is too large are mainly solved. The implementation process comprises the steps that first, simulation sinc kernel sampling is conducted on radar echo signals; second, Fourier coefficients of the echo signals are estimated; third, a radar target complex envelope matrix is estimated; fourth, radar high-resolution distance imaging is carried out. By means of the radar high-resolution distance imaging method based on simulation sinc kernel sampling, the sampling rate of broadband radar signals can be effectively lowered, the difficult problem that the signal processing data size is too large is solved, and meanwhile the precision and the detection probability of radar target parameter extraction can also be guaranteed.

Description

Radar high-resolution Range Imaging method based on the sampling of simulation sinc core

Technical field

The invention belongs to the Radar Technology field, relate to a kind of utilization simulation sinc and check radar return and carry out low speed sampling and high-resolution Range Imaging method.

Background technology

Wideband radar with the traditional narrow radar, compare have High Range Resolution, target identification and imaging capability, extremely strong advantages such as four anti-abilities, but broadband signal has also been brought huge challenge to digitized sampling.Nyquist sampling theorem is told us, must with the speed that is not less than the twice bandwidth, sample to bandlimited signal, never in the frequency spectrum of aliasing, recovers original signal.The bandwidth of broadband signal is usually more than 500MHz, and this just requires to design the ADC of sampling rate for number GHz, and high like this sampling rate will cause system complexity to increase, and cost rises, and implementation procedure is very difficult.Therefore, the new radar signal method of sampling and the corresponding signal processing technology of exploration becomes the current research focus.

Vetterli has proposed the limited new fixed rate of interest (FRI) sampling theory in 2002, for the radar signal sampling provides new thinking.The FRI sampling theory differs from nyquist sampling theorem, it is pointed out: some non-band limit parameter signal, burst signal for example, can the number of free parameter in the unit interval be called to the new fixed rate of interest of signal by a limited number of free expressed as parameters, the signal that the new fixed rate of interest is limited is the FRI signal, to the FRI signal, as long as select suitable sampling kernel function to sample with the speed of the new fixed rate of interest higher than signal, just can utilize certain algorithm to estimate free parameter, thereby rebuild original signal.

Current research to the FRI sampling theory is mainly for Dirac impulse string signal, Dirac impulse string signal is the unlimited signal of a kind of bandwidth in theory, according to traditional nyquist sampling theorem, can't sample and rebuild sort signal, yet according to the FRI theory, Dirac impulse string signal is a kind of FRI signal, therefore can to it, sample by the FRI sampling theory, then utilize the reconstruction algorithm such as pulverised wave filter, subspace accurately to rebuild original signal.In reality, Dirac impulse string signal is non-existent, and more practical is the burst signal of known pulse shape, and this all belongs to this situation in communication and radar, yet, also immature for the FRI sampling theory of burst signal.

From sampling structure, the FRI sampling structure is divided into two kinds of single channel and hyperchannels, and the analysis of multi-channel sampling structural theory is very simple, reduction sampling rate that can be larger, but complicated structure, and single channel sampling structure theoretical analysis complexity, but simple in structure, be the emphasis of research.

As can be known by the FRI sampling theory, be the Nyquist speed that fully likely breaks through traditional radar signal sampling, realize the low rate observation of radar signal, and therefrom recover target information, reach radar high-resolution Range Imaging method.

Summary of the invention

The object of the invention is to overcome the deficiency of above-mentioned prior art, propose a kind of radar high-resolution Range Imaging method based on the sampling of simulation sinc core, to realize low speed sampling and the Range Imaging to radar echo signal.

For achieving the above object, the scheme concrete steps of the present invention's proposition are as follows:

A kind of radar high-resolution Range Imaging method based on the sampling of simulation sinc core, comprise the steps:

1) radar echo signal is simulated to the sampling of sinc core, wherein the radar emission signal is linear FM signal LFM:

$s_T\left(t\right)=\mathrm{rect}\left(\frac{t}{T_0}\right)\exp \left(j2\mathrm{\π}(f_{c}t+\frac{1}{2}K{t}^2)\right)$

In formula, f _cThe signal carrier frequency information, T ₀The pulse width transmitted, K (K=B/T ₀) be the linear frequency modulation rate, B is the bandwidth of linear FM signal, and t is the fast time, and T is the repetition period of radar transmitted pulse, supposes that the scattering point number in scene is L; Low speed sampling obtain sampled value c[1], c[2], Kc[p], K, c[P], p indicating impulse sequence number wherein, P is pulse accumulation number, c[p]=[c _P1, c _P2, K, c _PM] ^TBe the sampled value of p pulse, M is the sampling number of each pulse;

2) Fourier coefficient of estimate echo signal; To step 1) described sampled value operates, obtain echoed signal Fourier coefficient x[1], x[2], K, x[P] }, x[p wherein]=[x _P1, x _P2, k, x _PM] be the Fourier coefficient of p pulse echo signal, M means required Fourier coefficient number;

3) estimate radar target complex envelope matrix; Adopt joint sparse Bayesian learning Algorithm for Solving radar target complex envelope matrix B _{N * P}=b[1], and b[2], K, b[P], wherein N means range unit number, b[p]=[b _P1, b _P2, K, b _PN] ^TTarget complex envelope corresponding to range unit while meaning p pulse of radar emission:

4) radar high-resolution Range Imaging method; According to step 3) in resulting target complex envelope matrix carry out the Fast Fourier Transform (FFT) operation, finally obtain high resolution target distance image.

Described method, wherein step 1) described simulation sinc core sampling structure comprises sampling core and low speed ADC two parts; Simulation sinc sampling core is:

$h(-t)=B_s\sin c\left({-B}_{s}t\right)=\frac{\sin (-\mathrm{\π}{B}_{s}t)}{-\mathrm{\πt}}$

The speed of low speed ADC is T _s=T/M, far below the nyquist sampling rate.

Described method, wherein step 2) described to being operating as that sampled value is carried out:

x[p]=G ^-1DFT{c[p]}，p=1，2，K，P

X[p in formula]=[x _P1, x _P2, K, x _PM] be the Fourier coefficient of p pulse echo signal, G is the diagonal matrix on M * M rank, its m+K+1 diagonal element is

G (ω) means the Fourier transform of g (t), and m ∈ I, DFT mean the discrete Fourier transformation operation.

Described method, wherein step 3) the described radar target complex envelope matrix method that solves is:

Radar range is divided into to N range unit, and the pass of the Fourier coefficient of echoed signal and echo time delay and target complex envelope is:

x[p]=SgV(t)b[p]，p=1，L，P

In formula, S is the diagonal matrix on M * M rank, and its m+K+1 diagonal element is (1/T) S (2 π m/T), and S (ω) means the Fourier transform of radar emission signal, and V (t) is M * N rank matrix, and its m+K+1 is capable, and the n column element is

T={t ₁, t ₂, L, t _N) echo time delay corresponding to expression range unit, n=1,2, K, N, m ∈ I; B[p]=[b _P1, b _P2, K, b _PN] ^TTarget complex envelope corresponding to scattering point on range unit while meaning p pulse of radar emission, if there is no target on range unit n, b _Pn=0; If target is arranged, b _Pn=a _Pn

Adopt the joint sparse Bayesian learning algorithm just can be from above-mentioned relation, solving target complex envelope b[p], p=1,2, K, P, composition target complex envelope matrix B _{N * P}=b[1], and b[2], K, b[P].

The present invention has the following advantages compared with prior art:

1) the present invention adopts the limited new fixed rate of interest method of sampling to sample to radar echo signal, when guaranteeing system performance, greatly reduce the sampling rate of echoed signal, and this method of sampling do not bring hard-wired complexity, this difficult problem of high-speed sampling for broadband and ultra-wideband radar signal provides solution route.

2) the present invention directly reconstructs the range information of target according to sampled value, has saved the pulse compression process in traditional Radar Signal Processing, has simplified signal processing flow.Adopt simultaneously the thought of sparse reconstruct to be reconstructed the radar target range information, without this prior imformation of target scattering point number, also improved simultaneously the estimated accuracy of target range.

3) there are the shortcomings such as secondary lobe leakage in the distance by radar formation method based on pulse compression, and the sparse Bayesian learning algorithm adopted in the present invention can finely overcome these shortcomings.

The accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention;

Fig. 2 is sampling structure figure of the present invention;

Fig. 3 is the present invention and classic method target range imaging effect comparison diagram;

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.

Step 1, simulate the sampling of sinc core to radar echo signal.

1.1) transmit in the present invention and select linear FM signal (LFM):

$S_T\left(t\right)=\mathrm{rect}\left(\frac{t}{T_0}\right)\exp \left(j2\mathrm{\π}(f_{c}t+\frac{1}{2}{K}_0{t}^2)\right)$

Wherein, f _cThe signal carrier frequency, T ₀The pulse width transmitted, K ₀(K ₀=B/T ₀) be the linear frequency modulation rate, B is the bandwidth of linear FM signal, and t is the fast time, and T is the repetition period of radar transmitted pulse.Suppose that the scattering point number in scene is L, the envelope that note transmits is s ₀(t)=rect (t/T ₀) exp (j π K ₀t ²), in radar usually the emission multiple pulses do coherent accumulation and detect performance to improve, suppose that radar launches P pulse altogether, the echo fundamental frequency signal of radar can be expressed as:

$s_B\left(t\right)=\underset{p=1}{\overset{P}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\σ}}_l{s}_0(t-\frac{{2r}_{\mathrm{pl}}}{c}-\mathrm{pT})\exp (-j4\mathrm{\π}{f}_c\frac{r_{\mathrm{pl}}}{c})+n_B\left(t\right)$

$=\underset{p=1}{\overset{P}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{a}_{\mathrm{pl}}{s}_0(t-\frac{{2r}_l}{c}-\mathrm{pT})+n_B\left(t\right)$

$=\underset{p=1}{\overset{P}{\mathrm{\Σ}}}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{a}_{\mathrm{pl}}{s}_0(t-t_l-\mathrm{pT})+n_B\left(t\right)$

In formula, n _B(t) mean the summation of Noise and Interference, σ _lThe backscatter intensity that means scattering point l, a _Pl=σ _PlExp (j4 π f _cr _Pl/ c), for comprising the echo complex coefficient of target scattering intensity and target doppler information, r _Pl=r _l+ (p-1) Tv _lThe distance of scattering point l and radar receiving antenna while meaning p pulse of radar emission, v _lThe relative radial rate that means scattering point l and radar.The coherent accumulation time will guarantee that range walk does not occur target, so during usually ignoring pulse accumulation, target range changes the impact on echo envelope, namely thinks s ₀(t-2r _Pl/ c-pT)=s ₀(t-2r _l/ c-pT), and the target range variation can not be ignored the impact of phase place, has wherein comprised the doppler information of target.T _l=2r _l/ c means the time delay that scattering point l is corresponding.

1.2) by simulation sinc core sampling structure shown in Figure 2, radar echo signal is sampled.

In Fig. 2, simulation sinc sampling core is:

$h(-t)=B_s\sin c(-B_{s}t)=\frac{\sin (-\mathrm{\π}{B}_{s}t)}{-\mathrm{\πt}}$

B wherein _sBandwidth for sinc sampling core.

In Fig. 2, the sampling rate of ADC is f _s=2B _s.

Radar fundamental frequency echoed signal through sampling core sampling obtain sampled value c[1], c[2], K c[p], K, c[P], p indicating impulse sequence number wherein, c[p]=[c _P1, c _P2, K, c _PM] ^TIt is the sampled value of p pulse.The sampling number of M for each impulse sampling is obtained, must meet M > 2L+1, wherein L is the scattering point number in the radar scene.The wide pass of sampling number and sinc sampling nucleus band is: M=2B _sT, the bandwidth of the core of therefore sampling must meet B _s(2L+1)/2T.In noise circumstance, in the situation that the ADC permission, B _sLarger with M, system performance is better.

Step 2, the Fourier series coefficient of estimate echo signal.

Press following formula to the described sampled value of step 1 c[1], c[2], Kc[p], K, c[P] operate, obtain the Fourier series coefficient of echoed signal:

x[p]=DFT{c[p]}／Tf _s，p=1，2,K，P

X[p in formula]=[x _P1, x _P2, K, x _PM] be the Fourier coefficient of p pulse echo signal, M means required Fourier coefficient number.DFT means the discrete Fourier transformation operation.

Step 3, estimate radar target complex envelope matrix.

In the present invention, adopt joint sparse Bayesian learning Algorithm for Solving to comprise the radar target complex envelope matrix of target scattering intensity and doppler information.Radar range is divided into to N range unit, obtains the Fourier series coefficient of echoed signal and the pass of echo time delay and echo coefficient and be:

x[p]=SV(t)b[p]，p=1,L，P

In formula, S is the diagonal matrix on M * M rank, and its m+K+1 diagonal element is (1/T) S (2 π m/T), and S (ω) means the Fourier transform of radar emission signal, and V (t) is M * N rank matrix, and its m+K+1 is capable, and the n column element is

T={t ₁, t ₂, L, t _NEcho time delay corresponding to expression range unit, n=1,2, K, N, m ∈ I.B[p]=[b _P1, b _P2, K, b _PN] ^TComplex envelope vector corresponding to scattering point on range unit while meaning p pulse of radar emission, if there is no target on range unit n, b _Pn=0; If target is arranged, b _Pn=a _Pn.

Adopt the joint sparse Bayesian learning algorithm just can be from above-mentioned relation, solving echo complex envelope vector b[p], p=1,2, K, P, composition target complex envelope matrix B _{N * P}=b[1], and b[2], K, b[P].

Step 4, radar high-resolution Range Imaging method; According to step 3) in resulting target complex envelope matrix carry out the Fast Fourier Transform (FFT) operation, finally obtain high resolution target distance image.

Effect of the present invention further illustrates by following l-G simulation test:

1. simulated conditions

Radar parameter is as follows: radar emission linear FM signal, carrier frequency f ₀=3GHz, repetition frequency PRF=10KHz, pulse width T=5 μ s, bandwidth B=15MHz, Nyquist sampling rate 30MHz, the coherent accumulation umber of pulse is 50, signal to noise ratio snr is defined as the scattering point signal to noise ratio (S/N ratio) at the minimum place of backscattering coefficient.

In Fig. 3,3 targets lay respectively at [6000,6900,7500] m, and backscattering coefficient is respectively [1,0.8,0.8], and target velocity is respectively [80,100,60] m/s, and signal to noise ratio (S/N ratio) is 0dB.The inventive method sampling number is 200 points, and namely sampling rate is 2MHz, and by Nyquist's theorem, the classic method sampling rate is 15MHz.

2. emulation content

The correctness of checking the inventive method, compare the inventive method and the performance of traditional impulse compression method aspect the radar target Range Imaging simultaneously.

3. analysis of simulation result

Fig. 3 (a) target distance image that traditional impulse compression method obtains of serving as reasons, sampling rate is 15MHz.Fig. 3 (b) is target distance image obtained by the method for the present invention, and the systematic sampling rate is 2MHz.Fig. 3 (b) proves, the present invention is based on the radar high-resolution Range Imaging method of simulating the sampling of sinc core and can effectively rebuild Radar Target Using Range Profiles under the condition that greatly reduces sampling rate.Comparison diagram 3 (a) and (b) as can be known, the Range Profile that traditional impulse compression method obtains under aforementioned simulated conditions has higher secondary lobe, and there is not secondary lobe basically in method imaging results of the present invention, can obtain than the better Range Profile of traditional impulse compression method.This has also proved correctness and the validity of the inventive method.

Should be understood that, for those of ordinary skills, can be improved according to the above description or conversion, and all these improve and conversion all should belong to the protection domain of claims of the present invention.

Claims (4)

1. the radar high-resolution Range Imaging method based on the sampling of simulation sinc core, is characterized in that, comprises the steps:

1) radar echo signal is simulated to the sampling of sinc core, wherein the radar emission signal is linear FM signal LFM:

$s_T\left(t\right)=\mathrm{rect}\left(\frac{t}{T_0}\right)\exp \left(j2\mathrm{\π}(f_{c}t+\frac{1}{2}{\mathrm{Kt}}^2)\right)$

In formula, f _cThe signal carrier frequency information, T ₀The pulse width transmitted, K (K=B/T ₀) be the linear frequency modulation rate, B is the bandwidth of linear FM signal, and t is the fast time, and T is the repetition period of radar transmitted pulse, supposes that the scattering point number in scene is L; Low speed sampling obtain sampled value c[1], c[2], Kc[p], K, c[P], p indicating impulse sequence number wherein, P is pulse accumulation number, c[p]=[c _P1, c _P2, K, c _PM] ^TBe the sampled value of p pulse, M is the sampling number of each pulse;

2) Fourier coefficient of estimate echo signal; To step 1) described sampled value operates, obtain echoed signal Fourier coefficient x[1], x[2], K, x[P] }, x[p wherein]=[x _P1, x _P2, K, x _PM] be the Fourier coefficient of p pulse echo signal, M means required Fourier coefficient number;

3) estimate radar target complex envelope matrix; Adopt joint sparse Bayesian learning Algorithm for Solving radar target complex envelope matrix B _{N * P}=b[1], and b[2], K, b[P], wherein N means range unit number, b[p]=[b _P1, b _P2, K, b _PN] ^TTarget complex envelope corresponding to range unit while meaning p pulse of radar emission;

4) radar high-resolution Range Imaging method; According to step 3) in resulting target complex envelope matrix carry out the Fast Fourier Transform (FFT) operation, finally obtain high resolution target distance image.

2. method according to claim 1, is characterized in that, wherein step 1) described simulation sinc core sampling structure comprises sampling core and low speed ADC two parts;

Simulation sinc sampling core is:

$h(-t)=B_s\sin c\left({-B}_{s}t\right)=\frac{\sin \left({-\mathrm{\πB}}_{s}t\right)}{-\mathrm{\πt}}$

The speed of low speed ADC is T _s=T/M, far below the nyquist sampling rate.

3. method according to claim 1, is characterized in that, wherein step 2) described to being operating as that sampled value is carried out:

x[p]=G ^-1DFT{c[p]}，p=1，2，K，P

X[p in formula]=[x _P1, x _P2, K, x _PM] be the Fourier coefficient of p pulse echo signal, G is the diagonal matrix on M * M rank, its m+K+1 diagonal element is

G (ω) means the Fourier transform of g (t), and m ∈ I, DFT mean the discrete Fourier transformation operation.

4. method according to claim 1, is characterized in that, wherein step 3) the described radar target complex envelope matrix method that solves is:

Radar range is divided into to N range unit, and the pass of the Fourier coefficient of echoed signal and echo time delay and target complex envelope is:

x[p]=SgV(t)b[p]，p=1,L，P

In formula, S is the diagonal matrix on M * M rank, and its m+K+1 diagonal element is (1/T) S (2 π m/T), and S (ω) means the Fourier transform of radar emission signal, and V (t) is M * N rank matrix, and its m+K+1 is capable, and the n column element is

T={t ₁, t ₂, L, t _NEcho time delay corresponding to expression range unit, n=1,2, K, N, m ∈ I; B[p]=[b _P1, b _P2, K, b _PN] ^TTarget complex envelope corresponding to scattering point on range unit while meaning p pulse of radar emission, if there is no target on range unit n, b _Pn=0; If target is arranged, b _Pn=a _Pn

Adopt the joint sparse Bayesian learning algorithm just can be from above-mentioned relation, solving target complex envelope b[p], p=1,2, K, P, composition target complex envelope matrix B _{N * P}=b[1], and b[2], K, b[P].

Images