CN103226191A - High-resolution P waveband SAR space-variant kinematic error compensation algorithm based on BP algorithm - Google Patents

Abstract

The invention discloses a high-resolution P waveband SAR (Synthetic Aperture Radar) space-variant kinematic error compensation algorithm based on a BP (Backpropagation) algorithm. According to the compensation algorithm, azimuth space-invariant error compensation is conducted by a conventional method, and then residual space-variant kinematic error compensation is conducted according to the BP algorithm. During the residual space-variant kinematic error compensation, a secondary phase perturbation technique is utilized, so that the operation quantity is reduced, and the time is saved; in addition, two-dimensional coupling of SAR data is avoided during independent compensation on a single target in a two-dimensional time domain due to the advantages of the BP algorithm; distance migration error correction can be conducted; and therefore, the compensation algorithm can adapt to a greater airborne kinematic error, and has higher compensation accuracy.

Description

High resolving power pattern-band SAR space-variant kinematic error backoff algorithm based on the BP algorithm

Technical field

The present invention relates to a kind of high resolving power pattern-band SAR space-variant kinematic error backoff algorithm, relate to carried SAR imaging processing and motion compensation field based on the BP algorithm.

Background technology

Airborne synthetic aperture radar (SAR) need utilize the imaging processing algorithm that SAR original echo data are carried out matched filter processing, obtains the SAR image.Because the time domain Processing Algorithm is very consuming time, general using frequency domain Processing Algorithm (as WK algorithm, Chirp Scaling algorithm, apart from range and Doppler) is carried out imaging processing.But the frequency domain Processing Algorithm can only adapt to the situation that carrier aircraft is done the desirable level linear uniform motion, need compensate the imperfect motion of carrier aircraft by movement compensating algorithm.Because the two-dimensional coupled of SAR signal is considered for operation efficiency, the empty direct component that movement compensating algorithm generally can only the compensating motion error and apart from the space-variant component., the synthetic aperture time lower at the SAR systemic resolution lacks under the less situation of (as X-band), carrier aircraft kinematic error, and above-mentioned processing mode can satisfy various application requirements.But for high resolving power pattern-band SAR system, because its range unit is less, the synthetic aperture time is longer, when the carrier aircraft kinematic error is big, residue orientation space-variant kinematic error after the conventional motion compensation not only causes the orientation to phase error, cause the distance translation correction error simultaneously, therefore above-mentioned orientation space-variant phase error compensation can not accumulate fully to target echo signal, thereby causes the SAR image resolution ratio to reduce, and image quality descends.Therefore need provide a kind of space-variant kinematic error compensation method will, can solve the orientation space-variant kinematic error compensation of high resolving power pattern-band SAR system when kinematic error is big, especially the distance translation correction error that is caused, thus even make high resolving power pattern-band SAR system when bigger kinematic error, also can obtain the SAR image product of well focussed.

(Back Projection, BP) algorithm is a kind of time domain pointwise Processing Algorithm of the SAR of being used for imaging processing in back projection.With the frequency domain Processing Algorithm the relatively poor adaptability of carrier aircraft kinematic error is compared, the BP algorithm is because its time domain pointwise treatment characteristic can accurately recover each gray values of pixel points of SAR image according to the carrier aircraft movement locus under any carrier aircraft motion state.Comparing with the frequency domain Processing Algorithm, though the BP algorithm can adapt to big kinematic error, need hundreds of times operand, is intolerable in actual applications, has limited the application of BP algorithm greatly.

Summary of the invention

The present invention proposes a kind of high resolving power pattern-band SAR space-variant kinematic error backoff algorithm based on the BP algorithm, by phase perturbation BP space-variant error compensation, the distance translation correction error that orientation space-variant error causes can be compensated, therefore bigger orientation space-variant error can be adapted to.

The objective of the invention is to be achieved through the following technical solutions:

Step 1 is handled in conjunction with the ω-κ algorithm of motion compensation the SAR original echo data utilization that receives, and obtains the two-dimensional frequency data of SAR image;

Step 2 is selected a time-bandwidth product value A, the orientation linear FM signal that frequency modulation rate of structure changes with frequency of distance on two-dimensional frequency:

Step 3, the orientation linear FM signal of utilizing step 2 to obtain is carried out phase perturbation to the two-dimensional frequency data that step 1 obtains, and returns two-dimensional time-domain, obtains the two-dimensional time-domain data:

Step 4, the synthetic aperture of calculating in the two-dimensional time-domain covers the L that counts, and then to obtain with each point be the energy range at center;

Step 5, structure SAR image pixel grid I (mT _m, nT _n), m wherein, n are respectively apart from location of pixels and orientation to location of pixels, T _m, T _nBe respectively distance, orientation at interval to pixel space,

M=1 ..., M, n=1 ..., N, M, N are respectively SAR image distance, orientation to number of pixels, and C is the light velocity, and Fs is the impulse sampling rate, P _RfBe the pulse recurrence interval, v is the carrier aircraft reference velocity;

Step 6 is for the SAR image pixel grid I (mT of step 5 structure _m, nT _n) in each pixel, according to the energy range that step 4 is determined, calculate the position curve of echo in two-dimensional time-domain of each pixel;

Step 7 at the position curve that step 6 obtains, is utilized synthetic-aperture radar space geometry relation, calculates the orientation space-variant kinematic error on the described position curve;

Step 8, the position curve that obtains at step 6, and then learn energy on the described position curve, utilize the advanced line phase correction of BP orientation space-variant Error Compensation Algorithm and then, obtain the plural gray-scale value of SAR image slices vegetarian refreshments at last according to the orientation space-variant kinematic error that step 7 obtains the energy stack.

Beneficial effect of the present invention:

1. compare with existing space-variant kinematic error backoff algorithm, the present invention at first utilizes conventional method to carry out the empty constant error compensation in orientation, carries out remaining space-variant kinematic error compensation according to the BP algorithm then.Utilize the quadratic phase perturbation technique to reduce operand when remaining space-variant kinematic error compensates, saved the time.

2. the phase perturbation by step 3, and return two-dimensional time-domain carries out synthetic aperture in two-dimensional time-domain, not with distance to variation, so the different distance unit object has identical length of synthetic aperture, for follow-up error compensation and accumulation offer convenience.

3. and because the advantage of BP algorithm self, in two-dimensional time-domain single target is compensated separately, avoided the two-dimensional coupled of SAR data, can carry out the correction of distance translation error, therefore can adapt to bigger carrier aircraft kinematic error, have higher compensation precision.

Description of drawings

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

Fig. 2 is the distance translation calibration result after ω-κ algorithm+conventional motion compensation;

Fig. 3 is length of synthetic aperture contrast after original length of synthetic aperture and the linear FM signal disturbance;

Fig. 4 is for to carry out index measurement to the imaging results point target;

Embodiment

In order to understand technical scheme of the present invention better, below in conjunction with drawings and the specific embodiments the present invention is done to describe in further detail.

The present invention proposes a kind of high resolving power pattern-band SAR space-variant kinematic error backoff algorithm---phase perturbation BP space-variant Error Compensation Algorithm based on the BP algorithm, compare with existing orientation space-variant Error Compensation Algorithm, because new method can compensate the distance translation correction error that orientation space-variant error causes, therefore can adapt to bigger orientation space-variant error.As Fig. 1, this method is specially:

Step 1 is handled in conjunction with the ω-κ algorithm of motion compensation the SAR original echo data utilization that receives, and obtains the two-dimensional frequency data of SAR image;

Be specially: SAR original echo data are carried out obtaining two-dimensional frequency data S (f after distance compression, distance translation correction, phase compensation, the Stolt interpolation _τ, f _η), f wherein _τ, f _ηBe respectively frequency of distance coordinate and orientation frequency coordinate;

Step 1 utilize earlier ω-κ algorithm and conventional movement compensating algorithm carry out basic imaging processing and empty constant, apart from the compensation of space-variant kinematic error, can compensate most kinematic error, make the derivation of subsequent compensation step set up.

Step 2 is selected the general selection 100 of a time-bandwidth product value A(), in the two-dimensional frequency data area that obtains of step 1, construct the orientation linear FM signal P (f that a frequency modulation rate changes with frequency of distance _τ, f _η):

$P(f_{\mathrm{\τ}},f_{\mathrm{\η}})=\exp \{j\·\mathrm{\π}\·k_0^{-1}\·f_{\mathrm{\η}}^2\}$

K wherein ₀For the orientation that changes with frequency of distance to the frequency modulation rate:

${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="DEST\_PATH\_HDA00003291510700022.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=\frac{\frac{C}{\mathrm{\&lambda;}}+f_{\mathrm{\&tau;}}}{\frac{\mathrm{CA\&lambda;}}{{4v}^2{\mathrm{\&beta;}}^2}}$

Wherein λ is a system wavelength, and C is the light velocity, and v is the carrier aircraft reference velocity, and β is that antenna bearingt is to field angle.

Step 3, the two-dimensional frequency data S (f that step 1 is obtained _τ, f _η) the orientation linear FM signal P (f that utilizes step 2 to obtain _τ, f _η) carry out phase perturbation, and return two-dimensional time-domain, obtain the two-dimensional time-domain data $\hat{s}(\mathrm{\&tau;},\mathrm{\&eta;})={\mathrm{FFT}}^{-1}\{S(f_{\mathrm{\&tau;}},f_{\mathrm{\&eta;}})\&CenterDot;P(f_{\mathrm{\&tau;}},f_{\mathrm{\&eta;}})\}$

FFT wherein ^-1Inverse fourier transform is carried out in { } expression.

Step 4, the two-dimensional time-domain data s that calculation procedure three obtains

In synthetic aperture cover the L that counts, and then obtain energy range η ', ${\mathrm{\&eta;}}^{\&prime;}\&Element;[(n-\frac{L}{2})T_n,(n+\frac{L}{2})T_n]:$

$L=\frac{\mathrm{tg}\left(\frac{\mathrm{\&beta;}}{2}\right)\mathrm{A\&lambda;}\&CenterDot;P_{\mathrm{rf}}}{{\mathrm{\&beta;}}^{2}v}$

P wherein _RfBe pulse recurrence interval (PRF);

L in this step is much smaller than effective aperture length

Step 3 is carried out the frequency modulation rate in two-dimensional frequency and is returned two-dimensional time-domain with the orientation that frequency of distance changes after the linear FM signal disturbance, thereby in step 4 two-dimensional time-domain signal original long length of synthetic aperture is shortened greatly;

Step 5, structure SAR image pixel grid I (mT _m, nT _n), m wherein, n are respectively apart from location of pixels and orientation to location of pixels, T _m, T _nBe respectively distance, orientation at interval to pixel space,

M=1 ..., M, n=1 ..., N, M, N are respectively SAR image distance, orientation to number of pixels, and C is the light velocity, and Fs is the impulse sampling rate, P _RfBe pulse recurrence interval (PRF).

Step 6 is for the SAR image pixel grid I (mT of step 5 structure _m, nT _n) in each pixel, according to the energy range η ' that step 4 is determined, the echo that calculates each pixel is in the two-dimensional time-domain data

In position curve (τ ', η ');

The two-dimensional time-domain data

In only in energy range

The backward energy that comprises this point in the scope calculates this energy and exists

In distance to position τ ':

${\mathrm{\&tau;}}^{\&prime;}={\mathrm{\&tau;}}_0+{\mathrm{mT}}_m+\frac{{2v}^2{\mathrm{\&beta;}}^2{({\mathrm{\&eta;}}^{\&prime;}-{\mathrm{nT}}_n)}^2}{\mathrm{CA\&lambda;}}$

τ wherein ₀Be the corresponding amount of delay of distance samples reference position:

${\mathrm{\&tau;}}_0=\frac{{2R}_{\mathrm{near}}}{C}$

R wherein _NearBe nearest oblique distance.

Step 7, the position curve that obtains according to step 6 (τ ', η ') obtains orientation space-variant kinematic error d _Av(τ ', η ' ', m, n);

This step concerns computer azimuth space-variant kinematic error d according to the SAR space geometry _Av(τ ', η ' ', m, n), wherein in the η ' ' expression SAR space geometry relation orientation corresponding with η ' to the position:

${\mathrm{\&eta;}}^{\&prime;\&prime;}={\mathrm{nT}}_n+({\mathrm{\&eta;}}^{\&prime;}-{\mathrm{nT}}_n)\frac{C({\mathrm{\&tau;}}_0+{\mathrm{mT}}_m){\mathrm{\&beta;}}^2}{\mathrm{A\&lambda;}}$

Step 8, energy on the position curve that step 6 is obtained (τ ', η '), the orientation space-variant kinematic error that obtains according to step 7, utilize the advanced line phase correction of BP orientation space-variant Error Compensation Algorithm and then, obtain SAR image slices vegetarian refreshments I (mT at last the energy stack _m, nT _n) plural gray-scale value;

Be specially:

Right ${\mathrm{\&eta;}}^{\&prime;}\&Element;[(n-\frac{L}{2})T_n,(n+\frac{L}{2})T_n]$ In the scope

In

The energy of position, advanced line phase are proofreaied and correct stack again, obtain SAR image slices vegetarian refreshments I (mT _m, nT _n) plural gray-scale value; Wherein phase correction is operating as:

$\tilde{s}({\mathrm{\&tau;}}^{\&prime;\&prime;},{\mathrm{\&eta;}}^{\&prime;})=\hat{s}({\mathrm{\&tau;}}^{\&prime;\&prime;},{\mathrm{\&eta;}}^{\&prime;})\&CenterDot;\exp \left\{\frac{4\mathrm{\&pi;}}{\mathrm{\&lambda;}}\right[\frac{v^2{\mathrm{\&beta;}}^2{({\mathrm{\&eta;}}^{\&prime;}-n{T}_n)}^2}{\mathrm{A\&lambda;}}+\sqrt{1+\frac{4{\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="DEST\_PATH\_HDA00003291510700022.png"\; state="\{\{state\}\}">v</figure-callout>}}^2{\mathrm{\&beta;}}^4{({\mathrm{\&eta;}}^{\&prime;}-n{T}_n)}^2}{{\mathrm{<figure-callout\; id="2"\; label="A"\; filenames="DEST\_PATH\_HDA00003291510700022.png"\; state="\{\{state\}\}">A</figure-callout>}}^2{\mathrm{\&lambda;}}^2}}\&CenterDot;d_{\mathrm{av}}({\mathrm{\&tau;}}^{\&prime;},{\mathrm{\&eta;}}^{\&prime;\&prime;},m,n)\left]\right\}$

Two-dimensional time-domain signal after length of synthetic aperture shortened carries out remaining space-variant kinematic error compensation according to the BP algorithm, be along distance to changing, therefore can carry out the correction of distance translation error.

Utilize design resolution to carry out emulation experiment for the pattern-band single-point target simulator data of 0.44m, distance/direction is counted and is made as 4096/12288 respectively, adding horizontal/vertical amplitude is respectively the kinematic error of 6m/3m, utilize ω-κ algorithm, ω-κ algorithm+PTA(256 point respectively), ω-κ algorithm+the present invention (time band long-pending be made as 100) carries out imaging processing, verifies validity of the present invention.

Fig. 2 is the distance translation calibration result after ω-κ algorithm+conventional motion compensation, and visible point echo signal energy dissipation is in several range units.

Fig. 3 is length of synthetic aperture contrast after original length of synthetic aperture and the linear FM signal disturbance, and visible length of synthetic aperture shortens greatly, thereby has improved arithmetic speed greatly.

The imaging results point target is carried out index measurement, result such as table 1 and shown in Figure 4.As seen the present invention can recover the orientation to resolution fully under the prerequisite that guarantees peak sidelobe ratio, integration secondary lobe ratio.In addition, time band is long-pending to be provided with greatly more, and compensation precision is high more, but operation efficiency can reduce.In addition, working time of the present invention is slightly longer than PTA.

Table 1

The present invention discusses according to positive side-looking model, also can obtain similar conclusion for strabismus mode.

The present invention is primarily aimed at the orientation space-variant kinematic error compensation of pattern-band SAR, and for the long L-band SAR of synthetic aperture, the present invention is suitable equally.

In sum, more than be preferred embodiment of the present invention only, be not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (2)

1. the high resolving power pattern-band SAR space-variant kinematic error backoff algorithm based on the BP algorithm is characterized in that,

Step 1 is handled in conjunction with the ω-κ algorithm of motion compensation the SAR original echo data utilization that receives, and obtains the two-dimensional frequency data of SAR image;

Step 2 is selected a time-bandwidth product value A, the orientation linear FM signal that frequency modulation rate of structure changes with frequency of distance on two-dimensional frequency:

Step 3, the orientation linear FM signal of utilizing step 2 to obtain is carried out phase perturbation to the two-dimensional frequency data that step 1 obtains, and returns two-dimensional time-domain, obtains the two-dimensional time-domain data:

Step 4, the synthetic aperture of calculating in the two-dimensional time-domain covers the L that counts, and then to obtain with each point be the energy range at center;

Step 5, structure SAR image pixel grid I (mT _m, nT _n), m wherein, n are respectively apart from location of pixels and orientation to location of pixels, T _m, T _nBe respectively distance, orientation at interval to pixel space,

M=1 ..., M, n=1 ..., N, M, N are respectively SAR image distance, orientation to number of pixels, and C is the light velocity, and Fs is the impulse sampling rate, P _RfBe the pulse recurrence interval, v is the carrier aircraft reference velocity;

Step 6 is for the SAR image pixel grid I (mT of step 5 structure _m, nT _n) in each pixel, according to the energy range that step 4 is determined, calculate the position curve of echo in two-dimensional time-domain of each pixel;

Step 7 at the position curve that step 6 obtains, is utilized synthetic-aperture radar space geometry relation, calculates the orientation space-variant kinematic error on the described position curve;

Step 8, the position curve that obtains at step 6, and then learn energy on the described position curve, utilize the advanced line phase correction of BP orientation space-variant Error Compensation Algorithm and then, obtain the plural gray-scale value of SAR image slices vegetarian refreshments at last according to the orientation space-variant kinematic error that step 7 obtains the energy stack.

2. a kind of high resolving power pattern-band SAR space-variant kinematic error backoff algorithm based on the BP algorithm as claimed in claim 1 is characterized in that, the value of time-bandwidth product described in step 2 A selects 100.

Images