CN102788972B - Self-focusing method suitable for ultra-high-resolution SAR (synthetic aperture radar) imaging - Google Patents

Abstract

The invention relates to a self-focusing of method suitable for the ultra-high-resolution SAR (synthetic aperture radar) imaging, comprising the following steps: (1) subjecting the two-dimensional echo data to polar coordinate format conversion; (2) reducing the resolution of the distance vector; (3) carrying out coarse-resolution imaging to the distance vector; (4) estimating the azimuth phase error; (5) calculating the residual range migration; (6) compensating the residual error; and (7) carrying out two-dimensional Fourier transform for imaging.

Description

A kind of auto-focus method being applicable to ultrahigh resolution SAR imaging

Technical field

The present invention relates to a kind of synthetic-aperture radar (synthetic aperture radar is called for short SAR) imaging signal processing method, particularly relate to a kind of synthetic-aperture radar autofocus algorithm.

Background technology

Synthetic-aperture radar (Synthetic Aperture Radar is called for short SAR) is a kind of new system radar by signal processing technology, ground scenery being carried out to imaging.SAR is realized by distance and bearing bidimensional high-resolution the imaging of target, its middle distance obtains to high resolving power by carrying out process of pulse-compression to broadband signal, orientation high resolving power then carries out Coherent processing realization by Technologies Against Synthetic Aperture data, and this Coherent processing depends on the instantaneous relative position information accurately known between radar and target.In practical application, by radar site disturbance with electromagnetic wave propagation medium is uneven etc. that factor affects, this coherence is often difficult to directly be guaranteed.The major measure taked at present increases auxiliary movement measuring unit (typically as Inertial Measurement Unit and GPS) to measure acquisition radar site information, and ignore the uneven impact of propagation medium.But, along with the raising of imaging resolution, the positional information precision that motion measurement unit provides still may cannot meet coherence requirement, and, the uneven radar echo delay error effects caused of propagation medium also becomes can not ignore, and will become the important limiting factor of following ultrahigh resolution SAR system vernier focusing imaging gradually.Therefore be necessary to study from radar return extracting data and the way of compensating error, i.e. auto-focus method.

The delay error of echo has the impact of two aspects to SAR imaging: one is to produce extra range migration, cannot be compensated in SAR imaging process, and two is in orientation to introducing phase error, to cause image orientation to defocus.When delay error is less, when the additional distance migration produced is less than a Range resolution unit, this residual range migration effect is completely negligible, the phase error of orientation one dimension only need be estimated and compensate to now self-focusing, this is also that (typical algorithm is as Subaperture method MD for autofocus algorithm nearly all at present, phase difference algorithm PD, Phase-gradient autofocus algorithem PGA, eigenvalue method and the algorithm etc. based on image criterion) prerequisite supposed, as document 1 (Mancill, C.E., and J.M.Swiger.A Map DriftAutofocus Technique for Correcting High Order SAR Phase Errors.27 ^thannual Tri-Service Radar Symposium.Record, Monterey, CA, 1981, pp.391-400.), document 2 (G.N.Yoji.Phase Difference Auto Focusing for SyntheticAperture Radar Imaging.United States Patent No.4999635, 1991.), document 3 (Wahl, D.E., P.H.Eichel, D.C.Ghiglia, and C.V.Jakowatz, Jr.Phase Gradient Autofocus-A Robust Tool for High Resolution SAR PhaseCorrection.IEEE Transaction on Aerospace and Electronic Systems, 30 (3), 1994, and document 4 (C.V.Jakowatz pp.827-834.), Jr., D.E.Wahl.Eigenvector Method for Maximum-likelihood Estimation of Phase Errors in SyntheticAperture Radar Imagery.J.Opt.Soc.Am.A., 10 (12), 1993, pp.2539-2546.) technology disclosed in.But, along with the increase of error, when especially imaging resolution is high especially, residual range migration will become inevitable across range unit, therefore, effective autofocus algorithm compensates while must consider residual range migration and orientation phase error with this understanding.Document 5 (D.W.Warner, D.C.Ghiglia, A.FitzGerrel, J.Beaver.Two-dimensional Phase Gradient Autofocus.Proceedings of SPIE, Vol.4123, 2000, pp.162-173.) disclose in and traditional one dimension Phase-gradient autofocus algorithem (PGA) is expanded to bidimensional, propose bidimensional Phase-gradient autofocus algorithem (2-D PGA) to attempt to address this problem, but it is said in conclusion as author, the method will reach practical as one dimension PGA, still there is many problems to need to solve.

At present, Correction Problems while also not having proven technique can solve residual range migration and orientation phase error in prior art.

Summary of the invention

Correction Problems while the object of the invention is to remain range migration and orientation phase error in solution ultrahigh resolution SAR imaging.

To achieve these goals, the invention provides a kind of auto-focus method being applicable to ultrahigh resolution SAR imaging, comprise the steps:

(1) polar format conversion is carried out to bidimensional echo data;

(2) range resolution is reduced;

(3) distance is to coarse resolution imaging;

(4) orientation phase error is estimated;

(5) residual range migration is calculated;

(6) residue-error compensation;

(7) bidimensional Fourier transform imaging.

Wherein, step (1) is by distance and bearing two one dimension interpolation, realizes the conversion of data layout from polar coordinates to rectangular coordinates.

Step (2) be by step (1) process after data in distance to carrying out frequency spectrum intercepting, reduce range resolution, frequency spectrum intercepts that to reduce the object of range resolution be in order to target residual after preventing step (1) from processing to be walked about the follow-up orientation phase error estimation and phase error of impact across Range resolution unit, therefore the ratio that frequency spectrum intercepts depends on the number of residual range migration across resolution element, for example residual range migration crosses over N number of resolution element, then intercept ratio and should be greater than 1/N, after ensureing that resolution reduces, residual range migration is negligible.

Step (3) is that the data after step (2) being processed do bidimensional fast Flourier (Fourier) conversion (FFT), is embodied as picture.

Step (4) processes the conventional autofocus algorithm (as Phase-gradient autofocus algorithem) of the imagery exploitation obtained to step (3) to process, and estimates to obtain orientation phase error.

Step (5) is that the orientation phase error obtained directly calculates the estimated value of residual range migration to utilize step (4) to estimate according to residual analytic relationship between range migration and orientation phase error.

Step (6) utilizes the residual range migration and orientation phase error estimating to obtain, carries out residue-error compensation to the data after step (1) process.

Step (7) is that after processing to step (6) compensation obtained, data make bidimensional FFT, realizes vernier focusing imaging.

Compared with prior art, the invention has the beneficial effects as follows:

(1) only compensate orientation phase error from conventional autofocus algorithm different, compensate while the present invention realizes residual range migration and orientation phase error, compensation precision is higher.

(2) compared with remaining range migration method with sub-aperture (or pulse) correlation estimation, the present invention directly calculates residual range migration, counting yield is higher, and the estimated value of residual range migration derives from orientation phase error estimation and phase error simultaneously, and therefore estimated accuracy is also higher.

Accompanying drawing explanation

Fig. 1 is Spotlight SAR Imaging data acquisition geometric model figure.

Fig. 2 is signal processing flow figure of the present invention.

Fig. 3 is the Range compress images (locally) of data after PFA process.

Fig. 4 is the Range compress image (locally) after adopting the inventive method to compensate.

Fig. 5 is measured data image after polar format algorithm (PFA) process.

Fig. 6 is the result of PFA image after conventional self-focusing (employing Phase-gradient autofocus algorithem) process.

Fig. 7 is the final imaging results adopting the inventive method process to obtain.

Embodiment

Below in conjunction with the drawings and specific embodiments, the present invention will be further described in detail.

Spotlight SAR Imaging imaging geometry model as shown in Figure 1, without loss of generality, supposes that radar is operated in strabismus mode, and angle of squint is θ ₀, the corresponding θ of positive side-looking ₀=0.Texas tower is with the horizontal unaccelerated flight of speed v, and antenna phase center (APC) momentary position coordinates is (x ₀+ vt, y ₀, z ₀), wherein t represents the orientation time, at aperture center moment t=0, θ ₀+ θ and be respectively instantaneous azimuth and the angle of pitch of antenna phase center, equal θ respectively in the aperture center moment ₀with suppose again have a point target in scene, its position is (x _p, y _p).Antenna phase center is designated as R respectively to the instantaneous distance of this target and scene center _tand R _a.

Radar return, after demodulation, distance compensate to matched filtering and azimuth motion, can be expressed as

$S(t,f_{\mathrm{\τ}})=\exp \left\{j\frac{4\mathrm{\π}}{c}(f_0+f_{\mathrm{\τ}})R_{\mathrm{\Δ}}\right\}---\left(1\right)$

Wherein c is radio wave propagation speed, f ₀for the carrier frequency that transmits, f _τfor distance is to frequency, its effective range is b is transmitted signal bandwidth, and respective distances is to resolution r _Δ=R _a-R _tfor APC is to the difference distance of scene center and target.

Method provided by the invention processes for echo data formula (1) Suo Shi, and final goal is the vernier focusing image obtaining point target, and as shown in Figure 2, specific implementation step comprises its treatment scheme:

(1) distance of polar format conversion is to interpolation

Polar format algorithm (PFA) distance is mathematically equivalent to interpolation frequency of adjusting the distance and makes a following change of scale

$f_{\mathrm{\τ}}=\mathrm{\δ}{\stackrel{\·}{f}}_{\mathrm{\τ}}+f_0(\mathrm{\δ}-1)---\left(2\right)$

Wherein for the frequency of distance after conversion, for scale factor.Therefore, by distance after interpolation, signal can be expressed as

Wherein for the function of orientation time t, following Taylor expansion can be done to it about t

r(t)＝y′ _p+x′ _pt+ε(t) (4)

Wherein y ' _p+ x ' _pt is constant term and once item, the position of reflection target in the end in image, by visual angle effect, inclination of wave front and other error effect, and x ' _p, y ' _px might not be equaled _p, y _p, but this not effect diagram image focu, can be realized by the resampling of image area its correction, the present invention does not discuss; ε (t) is secondary and the above higher order term of secondary, is the key factor causing image defocus.

(2) orientation of polar format conversion is to interpolation

Orientation interpolation is in fact that data after interpolation of adjusting the distance do nonlinear transformation in the orientation time wherein

In formula for the orientation time variable after conversion, Ω=v/y ₀.

Therefore, the signal that can obtain after orientation interpolation is

Wherein for constant term, for bidimensional phase error.

For analytical error effect, can by error term about frequency of distance make following Taylor expansion

$g(\stackrel{\·}{t},{\stackrel{\·}{f}}_{\mathrm{\τ}})={\mathrm{\φ}}_0\left(\stackrel{\·}{t}\right)+{\mathrm{\φ}}_1\left(\stackrel{\·}{t}\right){\stackrel{\·}{f}}_{\mathrm{\τ}}+{\mathrm{\φ}}_2\left(\stackrel{\·}{t}\right){\stackrel{\·}{f}}_{\mathrm{\τ}}^2+\·\·\·---\left(7\right)$

Can solve

Wherein for residual orientation phase error, corresponding residual range migration (residual range migration amount

According to formula (8) and formula (9), residual range migration item can be obtained with orientation phase error there is following relation

Wherein for inverse mapping, its expression formula can obtain as follows according to formula (5):

(3) distance frequency spectrum intercepts and reduces range resolution

By polar format change after data in distance to carrying out frequency spectrum intercepting, reduce range resolution, frequency spectrum intercepts that to reduce the object of range resolution be to prevent the target that remains to walk about the follow-up orientation phase error estimation and phase error of impact across resolution element, therefore the ratio that frequency spectrum intercepts depends on the number of residual range migration across resolution element, for example residual range migration crosses over N number of resolution element, then intercept ratio and should be greater than 1/N, after ensureing that resolution reduces, residual range migration is negligible.Keep centre frequency constant during intercepting, if intercepting ratio is 1/N, then frequency of distance after intercepting scope be

(4) bidimensional FFT imaging

After the polar format conversion of step (1) and step (2), obtain the orthogonal bidimensional sampling of target scene frequency spectrum, after the distance frequency spectrum of step (3) intercepts reduction resolution, then by the full resolution distance coarse resolution image in orientation of target scene just can be obtained after distance and bearing bidimensional FFT.

(5) orientation phase error is estimated

The image that step (4) obtains, range resolution is lower, and residual range migration is no longer across resolution element, therefore negligible.Conventional autofocus algorithm (as Phase-gradient autofocus algorithem) now can be utilized to carry out orientation phase error estimation and phase error, suppose to estimate that the phase error obtained is

(6) residual range migration is calculated

According to relation formula (10) Suo Shi, step (5) can be utilized to estimate, and the orientation phase error obtained directly calculates residual range migration item

${\widehat{\mathrm{\φ}}}_1\left(\stackrel{\·}{t}\right)=\frac{1}{f_0}\{{\widehat{\mathrm{\φ}}}_0\left(\stackrel{\·}{t}\right)-\frac{\stackrel{\·}{t}}{{[1-0.5\mathrm{\Ω}\sin \left(2{\mathrm{\θ}}_0\right)\stackrel{\·}{t}]}^2}\frac{d{\widehat{\mathrm{\φ}}}_0\left[{\mathrm{\θ}}^{-1}\left(t\right)\right]}{\mathrm{dt}}{|}_{t=\mathrm{\θ}(\stackrel{\·}{t},0)}\}---\left(12\right)$

(7) residue-error compensation

Utilize step (5) and step (6) to estimate to obtain orientation phase error and residual range migration compensates the data after step (2) process, namely following compensation rate is multiplied by formula (6)

$S_E(\stackrel{\·}{t},{\stackrel{\·}{f}}_{\mathrm{\τ}})=\exp \{-j[{\widehat{\mathrm{\φ}}}_0\left(\stackrel{\·}{t}\right)+{\widehat{\mathrm{\φ}}}_1\left(\stackrel{\·}{t}\right){\stackrel{\·}{f}}_{\mathrm{\τ}}\left]\right\}---\left(13\right)$

Therefore the signal after can being compensated is

(8) bidimensional FFT imaging

If the orientation phase error estimated and residual range migration are enough accurate, then residual phase error term usually negligible, therefore formula (14) can be approximated to be

Now, a bidimensional FFT is done to the signal after compensating, just can realize the vernier focusing imaging to target

In formula, F represents that bidimensional Fourier converts, sinc _a(u)=sin (π B _du)/(π B _du) for orientation is to Sinc function, sinc _r(x)=sin (π B _τx)/(π B _τx) for distance is to Sinc function, B _dand B _τbe respectively azran descriscent signal bandwidth.Above formula also can be write as the form of spatial domain coordinate

Wherein with represent spatial domain azimuth-range coordinate respectively.

Measured data process is utilized to verify the autofocus algorithm that the present invention proposes, experimental result sufficient proof validity of the present invention.

Measured data is that certain high-resolution air-borne test SAR enrolls.This experimental radar works in X-band, and transmitted signal bandwidth is 1.18GHz, and respective distances is better than 0.15m to theoretical resolution, and adopt beam bunching mode admission data, during process, effective length of synthetic aperture is 2300m, and counterparty's potential theory resolution is about 0.067m.This radar instantaneous position information provided owing to not having movement measuring unit, therefore supposes during PFA process that radar does linear uniform motion.Fig. 3 gives the Range compress image (locally) of data after PFA process, can clearly see from figure, although have passed through the range migration correction of PFA, target still remains very large across Range cell migration, must consider to compensate it.Fig. 4 is the Range compress image after adopting the inventive method to compensate, and can see, the range migration of object residue obtains effective compensation all.Further, Fig. 7 gives the final imaging results adopting the inventive method process to obtain, in order to contrast, give PFA image and the result of PFA image after conventional self-focusing (adopting Phase-gradient autofocus algorithem here) process, respectively as shown in Figure 5 and Figure 6 simultaneously.Because residual range migration and phase error are all very large, therefore PFA image has occurred that two dimension clearly defocuses, conventional autofocus algorithm is only applicable to the phase error compensation in the negligible situation of residual range migration, therefore also unsatisfactory to the result of these data, and after adopting the inventive method process, imaging effect is obviously improved.

Claims (1)

1. be applicable to an auto-focus method for ultrahigh resolution SAR imaging, comprise the steps:

(1) polar format conversion is carried out to bidimensional echo data;

(2) range resolution is reduced;

(3) distance is to coarse resolution imaging;

(4) orientation phase error is estimated;

(5) residual range migration is calculated;

(6) residue-error compensation;

(7) bidimensional Fourier transform imaging;

Wherein, step (1) is by distance and bearing two one dimension interpolation, realizes the conversion of data layout from polar coordinates to rectangular coordinates; Step (2) be by step (1) process after data in distance to carrying out frequency spectrum intercepting, reduce range resolution; Step (3) is that the data after step (2) being processed do bidimensional fast fourier transform, is embodied as picture; Step (4) processes the conventional autofocus algorithm of the imagery exploitation obtained to step (3) to process, and estimates to obtain orientation phase error; Step (5) is that the orientation phase error obtained directly calculates the estimated value of residual range migration to utilize step (4) to estimate according to residual analytic relationship between range migration and orientation phase error; Step (6) utilizes the residual range migration and orientation phase error estimating to obtain, carries out residue-error compensation to the data after step (1) process; Step (7) is that after processing to step (6) compensation obtained, data make bidimensional Fourier transform, realizes vernier focusing imaging;

The computing method of described step (5) are as follows:

Wherein represent orientation phase error, corresponding residual range migration, t and represent the orientation time before and after polar format conversion process respectively, f ₀for the carrier frequency that transmits, θ ₀for the position angle of aperture center moment radar antenna phase center, Ω=v/y ₀for by radar speed v and radar to target area distance y ₀the amount determined, for the function determined by the resampling of PFA orientation, for inverse function.