CN100588988C - Imaging method for implementing airborne radar refocusing in system short of high-precision motion compensating system - Google Patents

Abstract

The invention relates to a re-focusing and imaging method for an on-board radar without any hi-precision motion compensation system, which comprises two procedures: (1) SAR initial compound image creation based on de-gradient processing; (2) Doppler signal phase error is measured based on improved PGA arithmetic, and compound image re-focusing for phase compensation is carried out. The second procedure comprises the following steps: the compound images are input; the center of the target with the most intensive point goes through cyclic movement; a window is added; Fourier transformation is done for orientation; estimation is made on the phase error; the phase is corrected; inverse Fourier transformation is done to return the compound image area. The invention allows to get on-board L-SARre-focused images with resolution of orientation obviously improved only by original signal data without any hi-accuracy on-board platform movement compensation system to provide the info on movementstate of an airplane; and the higher the contrast in the images, the better the focusing effect will be. As the SAR wavelength of L wave section is longer and penetrating, the invention can play highrole in military reconnaissance and monitoring any disaster in civil application.

Description

Lack and realize airborne radar reunion focusing imaging method under the high-precision motion compensating system

Technical field

The present invention relates to realize airborne radar reunion focusing imaging method under a kind of shortage high-precision motion compensating system, belong to synthetic aperture radar (SAR) signal digital imaging processing technical field.

Background technology

Synthetic-aperture radar is a kind of of imaging radar.It is installed on the motion platforms such as aircraft, airship or satellite usually, handles radar self image information of the electromagnetic wave acquisition terrain object of radiation earthward by receiving.As a kind of means of the active remote sensing of microwave, it has the not available advantage of remote optical sensing means.At first, have round-the-clock, thereby the influence that the ability to work of round-the-clock is not covered by illumination and sexual intercourse; Secondly, because radar signal has the coherent characteristic, the synthetic-aperture radar of particular design can be extracted the three-dimensional information on ground, can draw digital topography map fast and accurately; Moreover, because the electromagnetic wave of lower frequency has certain penetration capacity, the synthetic-aperture radar that is operated in the low section of microwave frequency (as L-band) can be surveyed vegetation and cover target down, can bring into play important role for military affairs investigation and civilian middle disaster monitoring.

Along with improving and development of synthetic aperture radar (SAR) digital signal processing theory, the digital signal processing means have substituted the optical processing method of early stage SAR gradually.Current, even on common personal computer, also can realize non real-time SAR signal imaging processing.Digital imaging technology has the incomparable dirigibility of optical image technology, as long as some parameter of suitable adjustment algorithm just can be handled the radar raw data that obtains under the different condition.Even radar is operated under the nonideal condition, utilize digital imaging method still might obtain desirable radar image, and this need rely on the specific imaging algorithm and makes image focus on the resolution that reaches desirable again.The resolution characteristic of synthetic-aperture radar comprises range resolution and azimuthal resolution, owing to obtain high range resolution comparative maturity technically, so the raising of azimuthal resolution is the most key to synthetic-aperture radar.

The orientation of SAR is to utilize the Doppler frequency variation (phase history) that relative motion produces between flying platform and the ground point target to obtain to resolution, therefore, the stability of SAR signal phase for imaging processing after the orientation of image to the height of resolution fundamental influence is arranged.Platform is under desirable linear uniform motion state, as long as correct calculation goes out the doppler phase course that doppler centroid and these two parameters of Doppler's chirp rate just can complete description target echo signals, realize the orientation to the focusing processing.In theory, the azimuthal resolution of positive side-looking SAR is approximately half of antenna length.In the carried SAR imaging, because airplane motion has very strong randomness, Doppler signal is difficult to accurately represent with linear FM signal under a lot of situations.The orientation will cause the decline of image resolution ratio to nonideal phase history, and the resolution of synthetic-aperture radar will be difficult to improve with the increase of synthetic aperture, even can descend, just along with the blooming effect of the increase image of synthetic aperture time is more obvious.The phase error that general flying platform motion change causes to a certain extent can be by various motion compensating system compensation, calculate Doppler parameter as attitude and the kinematic parameter of accurately measuring platform by the movement measurement system that adopts inertial navigation system (INS), Inertial Measurement Unit (IMU) and GPS compositions such as (GPS), thereby realize that range migration correction and orientation focus on, but compensation effect is subjected to the restriction of system performance.When radar resolution was higher, there was bigger error in the Doppler parameter that the disturbance of aircraft will make the inertial navigation data computation go out.

As coherent radar, the detection error requirements of radial distance reaches the centimetre-sized level, and the bearing accuracy between synthetic-aperture radar and the target must reach below 1/4th of wavelength.On the platform of high-speed motion, want to reach very difficulty of this index.The wavelength that uses as L-band is 22-23cm, and for the aircraft with the above speed flight of 150m/s, it is very difficult to obtain accurate localization information like this.At present, China does not also set up perfect high precision global location navigational system, and the inertial navigation precision of airborne platform inside also can't satisfy the radar requirement.This will cause airborne radar Doppler signal phase history parameter finely not control, and the radar image quality instability of acquisition, fuzzy defocusing with resolution are descended.So, be subjected to the restriction of existing hardware level, the precision of motion compensating system is the imperfect motion of compensating platform fully, must utilize the information of received signal data itself to do to estimate further and compensation that this imaging processing claims auto focusing method usually in actual imaging.For in the patent " a kind of auto-focus method that is applicable to low contrast scene synthetic aperture radar image-forming " of " 03154422.3 ", a kind of auto-focus method that is applicable to the low contrast scene has been proposed in Chinese patent application number, and with the Ku wave band as embodiment.Cause auto-focus method-phase gradient auto-focusing algorithm (PGA) (1. Eichel P H of extensive concern in the world in the radar process field, Wahl D E.Phase gradient algorithms as an optimal estimator of the phase derivative[J] .OptSociety of America, 1989,14 (20): 1101-1103. is WAHL 2., D.E., EICHEL, P.H., GHIGLIA, D.C., et al.Phase gradient autofocus-A robust tool for high resolutionSAR phase correction[J] .IEEE Trans., AES.1994,30 (3): 827-835.), can realize phase error estimation and phase error and compensation preferably, have the advantage lower to the target property degree of dependence.

As mentioned above, the digital focus imaging of synthetic-aperture radar signal has very strong dependence to Position Fixing Navigation System, and is quite high to the positioning accuracy request of motion compensating system.When lacking the high-precision motion compensating system, perhaps carry the motion compensating system catastrophic failure of platform or work when undesired, when in the signal data that receives, not carrying any motion information data of platform, how to only depend on radar original signal data to generate and focus on preferable image, and be fit to the important topic that various object scene are still current research.

Purpose of the present invention just provides a kind of airborne L-SAR high accuracy number reunion focusing imaging method of realizing the complete data driving under the high-precision motion compensating system that lacks.Be implemented under the situation that does not write down any airplane motion condition information in the signal data, begin to handle from the original signal data of airborne original signal register system record, by estimating effectively and the compensation of phase error, obtain finally that signal to noise ratio (S/N ratio) is significantly improved, the orientation is to obviously raising of resolution, steady quality and focus on good airborne L-SAR image.

Summary of the invention

The objective of the invention is to be achieved through the following technical solutions.The invention provides to lack and realize airborne radar L-SAR reunion focusing imaging method under the high-precision motion compensating system, comprise following two big steps:

(1) generates based on the initial complex pattern of the SAR that goes slope to handle.This part originates in the processing of radar original signal, comprises concrete following steps: distance is handled, and Doppler parameter estimates that range migration is corrected, the orientation compression.Wherein distance is handled and the orientation compression is adopted and removed the slope algorithm, range migration correct by distance to the frequency shifts realization.

(2) focus on again based on the go forward side by side complex pattern of line phase compensation of improved phase gradient automatic focus (PGA) algorithm measurement Doppler signal phase error.Phase error estimation and phase error is unescapable in the airborne synthetic aperture radar signal processing, the PGA algorithm is as a kind of phase error estimation method of data-driven, from the ultimate principle of radar imagery and the statistical property of signal, solved the estimation problem of target area point spread function.This step is accepted the complex pattern of (1) generation as input, specifically comprises following step: the complex pattern input; The center ring shift of point of maximum intensity target; Windowing; The orientation to Fourier transform; Phase error estimation and phase error; The orientation is to phase error compensation, and promptly phase place is corrected; Inverse fourier transform returns the complex pattern territory.In this step, the present invention adopts iteration error correcting technique in the information theory to the update method of phase information in the phase error estimation and phase error process, and has effectively avoided overconvergence and two kinds of situations of misconvergence by algorithm reasonable in design.

Advantage of the present invention is:

1) airborne synthetic aperture radar is not owing to have the high-accuracy position system support and be subjected to atmosphere microwave refractometer rate variable effect, and the azimuthal resolution of image does not reach design objective, adopts normal radar image processing method gained image blurring, defocuses serious; Utilize the present invention can recover the resolution of image to a great extent by Jiao that meets again;

2) in the burnt process of meeting again, the original signal sampling point that participates in phase error estimation and phase error blocks at random, without any artificial intervention;

3) degree of dependence to target property is lower, and this method is not only effective to point target, can obtain good effect equally to distribution objectives, can be fit to various object scene.The orientation of burnt image of meeting again obviously improves to resolution, and signal to noise ratio (S/N ratio) obtains significantly to improve.

Description of drawings

Fig. 1 generates the treatment scheme block scheme for the initial complex pattern of the present invention;

Fig. 2 improves the PGA algorithm burnt treatment scheme block scheme of meeting again for the present invention is based on;

Window width and position view that Fig. 3 selects for the present invention is final, among the figure horizontal ordinate be the orientation to, ordinate is an amplitude;

Fig. 4 is an iteration error correction procedure block scheme of the present invention;

Fig. 5 (a) is the orientation diagram of certain point target before focusing on again;

Fig. 5 (b) is the orientation diagram that focuses on certain point target of back again; Among the figure horizontal ordinate be the orientation to, ordinate is an amplitude;

Fig. 6 (a) is L-SAR scene image before focusing on again;

Fig. 6 (b) is for focusing on back L-SAR scene image again.

Embodiment

Below in conjunction with the drawings and specific embodiments the present invention is described in further detail.

Adopt the original signal data of the airborne L-SAR original signal register system record of CAS Electronics Research Institute's development.This system is not enough to carry out motion compensation owing to the inertial navigation system precision, so do not write down the information of any airplane motion in the signal data.Because L-SAR grew up about about 20 seconds whole aperture time very much, aircraft obviously was difficult to guarantee unaccelerated flight in the so long time, lacking under the high-precision Position Fixing Navigation System condition, signal space control sample time is quite coarse, phase stability in the aperture is difficult to guarantee, has had a strong impact on image quality.The random phase errors that caused by the airplane motion instability influences, and the azimuthal resolution that the radar baseband signal is carried out the single-view picture of conventional processing acquisition does not reach and corresponding resolution of synthetic aperture time.Doppler's chirp rate estimation procedure is not restrained, and chirp rate estimated value wave phenomenon is obvious.Radar single-view picture even defocus with the increase of synthetic aperture.In order to obtain to focus on good high-resolution L-SAR image, adopt method of the present invention to carry out focal imaging again.

The original signal of radar record is the time series of plural number.Original signal is handled the complex pattern that can obtain radar through imaging algorithm, and complex pattern can be converted into true picture at an easy rate.The major parameter that handling the synthetic-aperture radar original signal needs comprises radar frequency of operation (carrier frequency), signal waveform parameter (bandwidth, pulse width), the ground velocity of aircraft, pulse repetition rate etc.The L-SAR systematic parameter is: radar work centre frequency: 1250MHz (L-band); Distance sample: 4096 points; Distance samples interval: 2.25m; Echo samples low coverage (apart from the transponder pulse forward position): 15.12252km; Fire pulse width: 33 μ s; Linear frequency modulation bandwidth: 60MHz; Ground velocity and pulse repetition rate (PRF) relation: PRF (Hz)=1.11 * V (m/s); Average ground velocity: 150m/s.Treatment step is:

First step, original signal is through generating initial SAR complex pattern based on the imaging processing of removing slope, as shown in Figure 1.Its course of work is:

1. adopt and go the slope algorithm to carry out the distance processing;

2. matrix transpose;

3. estimate based on the spectral analysis method doppler centroid;

4. estimate based on Doppler's chirp rate of sub-aperture correlation technique;

5. matrix transpose;

6. based on the estimation of the range migration curve of azran relevant (the sub-aperture of distance method), by distance to frequency shifts realize that range migration corrects;

7. matrix transpose;

8. adopt and go the slope algorithm to carry out the orientation compression.

The characteristics of going slope to handle in the above-mentioned initial imaging process are the maximums diameter of hole that has utilized radar, defocusedly can obtain orientation that radar system reaches to limiting resolution so that meet again.

Second largest step, the initial complex pattern that first step process is generated carries out based on improving the burnt processing of PGA algorithm reunion.As shown in Figure 2.The course of work is: the complex pattern input; Point of maximum intensity target's center ring shift; Determine window width, windowing; Fourier transform of azimuth; The orientation is to phase error estimation and phase error; The orientation is to phase error compensation, and promptly phase place is corrected; Inverse fourier transform returns the complex pattern territory.At this, its concrete implementation is done one explain.

Complex pattern input is promptly imported in the above-mentioned first step by the initial complex pattern of synthetic-aperture radar that goes after the gradient method imaging.Complex pattern in the orientation to Fourier transform be exactly the orientation to the doppler phase course.If the doppler phase course after distance is handled is:

F _n(s)＝|F _n(s)|exp(j(φ _n(s)+φ _e(s))) (1)

Here n represents range unit, F _n(s), φ _n(s) be amplitude and the phase place that goes up signal apart from resolution element n respectively, the position that the s representative makes progress in the orientation.And φ _e(s) be phase error with the irrelevant needs compensation of range unit.Details are as follows to the processing procedure substep after the complex pattern input below.

The first step is a ring shift.In each range gate, select the strongest scattering point, with its ring shift to 0 position.The result of this operation lines up vertical line with point target the strongest on each range gate on the position of orientation time 0.The purpose in this step is exactly to select the best process points of signal to noise ratio (S/N ratio), no matter these points are real points, or the point in similar clutter background.

Second step was windowing.Be the phase error of estimation point target, must point target be separated from background on every side by windowing.If there is not phase error in the signal, the focusing situation ideal of point target, the width that point occupies is very narrow naturally.But owing to there is a phase error, point is fuzzy to taking place in the orientation, that is to say a little in the orientation to having occupied certain width.Therefore window function must guarantee that certain width just can come out the intercepting of the signal integrity of point target.Here the selection of window width is rule of thumb set, and the 10dB point is a thresholding under the maximal value as setting.

The 3rd step was a phase error estimation and phase error.After displacement of range gate n point of maximum intensity and windowing, need do Fourier transform to signal, picture signal is transformed into the doppler phase course.If the doppler phase course after the conversion is:

G _n(s)＝|G _n(s)|exp(j(φ _e(s)+θ _n(s))) (2)

θ wherein _n(s) be the phase place by target scattering characteristics decision relevant with range unit.Its statistical property: mathematical expectation is 0.Both members is differentiated, can get:

G′ _n(s)＝|G _n(s)|′exp(j(φ _e(s)+θ _n(s)))+|G _n(s)|exp′(j(φ _e(s)+θ _n(s)))

＝|G _n(s)|′exp(j(φ _e(s)+θ _n(s)))+j(φ _e(s)+θ _n(s))′|G _n(s)|exp(j(φ _e(s)+θ _n(s)))

＝(|G _n(s)|′+j(φ _e(s)+θ _n(s))′|G _n(s)|)exp(j(φ _e(s)+θ _n(s)))

(3)

The conjugation of doppler phase course is G _n ^*(s):

$G_n^{*}\left(s\right)=\left|G_n\left(s\right)\right|\exp (-j({\mathrm{\φ}}_e\left(s\right)+{\mathrm{\θ}}_n\left(s\right)))---\left(4\right)$

At this moment,

${G_n}^{\′}\left(s\right)G_n^{*}\left(s\right)={\left|G_n\left(s\right)\right|}^{\′}\left|G_n\left(s\right)\right|+j{({\mathrm{\φ}}_e\left(s\right)+{\mathrm{\θ}}_n\left(s\right))}^{\′}{\left|G_n\left(s\right)\right|}^2---\left(5\right)$

Relevant with phase error in formula (5) is imaginary part, and imaginary part is taken out and can be got:

$\mathrm{Im}\left\{{G_n}^{\′}\left(s\right)G_n^{*}\left(s\right)\right\}={({\mathrm{\φ}}_e\left(s\right)+{\mathrm{\θ}}_n\left(s\right))}^{\′}{\left|G_n\left(s\right)\right|}^2---\left(6\right)$

The adjust the distance unit summation of formula (6) both sides has:

${\mathrm{\Σ}}_n\mathrm{Im}\left\{{G_n}^{\′}\left(s\right)G_n^{*}\left(s\right)\right\}={\mathrm{\Σ}}_n{({\mathrm{\φ}}_e\left(s\right)+{\mathrm{\θ}}_n\left(s\right))}^{\′}{\left|G_n\left(s\right)\right|}^2$

$={\mathrm{\Σ}}_n{{\mathrm{\φ}}_e}^{\′}\left(s\right){\left|G_n\left(s\right)\right|}^2+{\mathrm{\Σ}}_n{{\mathrm{\θ}}_n}^{\′}\left(s\right){\left|G_n\left(s\right)\right|}^2---\left(7\right)$

$={{\mathrm{\φ}}_e}^{\′}\left(s\right){\mathrm{\Σ}}_n{\left|G_n\left(s\right)\right|}^2+{\mathrm{\Σ}}_n{{\mathrm{\θ}}_n}^{\′}\left(s\right){\left|G_n\left(s\right)\right|}^2$

Easily see,

${\widehat{\mathrm{\φ}}}_e^{\′}\left(s\right)=\frac{{\mathrm{\Σ}}_n\mathrm{Im}\left\{{G_n}^{\′}\left(s\right)G_n^{*}\left(s\right)\right\}}{{\mathrm{\Σ}}_n{\left|G_n\left(s\right)\right|}^2}$ (8)

$={{\mathrm{\φ}}_e}^{\′}\left(s\right)+\frac{{\mathrm{\Σ}}_n{{\mathrm{\θ}}_n}^{\′}\left(s\right){\left|G_n\left(s\right)\right|}^2}{{\mathrm{\Σ}}_n{\left|G_n\left(s\right)\right|}^2}$

Here

Be to the phase error estimation of derivative, the evaluated error item is

Because stochastic variable θ _n(s) mathematical expectation is 0, so the mathematical expectation of its linear transformation (derivative is linear transformation) also is 0, so when n is fully big, just participate in average range unit when abundant, error term

To trend towards 0.The estimation of phase error

It is accurate more also just to become.

The 4th step be the orientation to phase error compensation, promptly phase place is corrected.Because whole computing originates in the complex pattern that is subjected to error effect, therefore when selecting the strongest point target, might judge by accident, the choice of location of point is inaccurate, influences the signal to noise ratio (S/N ratio) of whole estimation of error process.After first time estimation of error and compensation, the image focusing situation improves, strong point target is more outstanding, therefore need reselect these point targets, and carry out for the second time estimation and the compensation of phase error, the focusing situation of compensation back image is further improved for the second time, can do circulation for the third time.After the circulation for several times, picture quality will no longer change, and then can stop iteration.

In the said process, the core is by closing on the average of range unit, the derivative of finding the solution error phase.The precision that has two factor affecting phase place derivatives to estimate, one is the correct degree that the point target maximal value is selected, another is the interference component that comprises in the truncated window.Because target defocuses back stack mutually, former should be peaked place, may weaken owing near the reverse superposition of target, and peaked position changes.In fact the false judgment of maximum value position has influenced the selection of signal to noise ratio (S/N ratio) optimum.If the width of truncated window is greater than the width of actual needs, unnecessary interference component can participate in the estimation to phase error, influences focusing effect.The present invention adopts following iterative process to solve, and is described below:

The purpose that at first must clear and definite iteration need reach: real maximum point target is displayed, make and participate in average target best signal to noise ratio (S/N ratio) is arranged; 2. progressively determine the actual window width and the position of window, as far as possible outside the ELIMINATION OF ITS INTERFERENCE of approaching target.What therefore iterative process needed adjustment repeatedly should be the width and the position thereof of maximum value position and truncated window.The original state of iteration is: the maximal value that obtains might not be real maximum value position, the width of setting truncated window fully greater than the width of actual needs (if less than the width of actual needs, target in the image may divide), the position of window is about former point symmetry (because do not know point fuzziness after the distribution situation of energy about initial point).The desirable end-state of iteration is: the maximal value that obtains is the strongest scattering point of reality, the width of window be just point target in the orientation to diffusion breadth, the position of window determines according to the distribution situation of point target energy about initial point, not necessarily about former point symmetry.Final window width of selecting and position as shown in Figure 3, among the figure horizontal ordinate be the orientation to, ordinate is an amplitude.

The error phase that the actual participation phase place is corrected is estimated as:

${\widehat{\mathrm{\φ}}}_e\left(s\right)={\∫\widehat{\mathrm{\φ}}}_e^{\′}\left(s\right)$

$=\∫\frac{{\mathrm{\Σ}}_n\mathrm{Im}\left\{{G_n}^{\′}\left(s\right)G_n^{*}\left(s\right)\right\}}{{\mathrm{\Σ}}_n{\left|G_n\left(s\right)\right|}^2}---\left(9\right)$

$={\mathrm{\φ}}_e\left(s\right)+\∫\frac{{\mathrm{\Σ}}_n{{\mathrm{\θ}}_n}^{\′}\left(s\right){\left|G_n\left(s\right)\right|}^2}{{\mathrm{\Σ}}_n{\left|G_n\left(s\right)\right|}^2}$

Evaluated error is

Because Mathematical expectation convergence 0, so

Value near 0, degree of closeness not only with θ _n(s) statistical property is relevant, and and | G _n(s) | relevant, because | G _n(s) | be denominator term, so | G _n(s) | big more, error

More near 0, Here it is will select the reason of strong scattering point.Even but in first time processing procedure | G _n(s) | not the maximal value in this range unit, because θ _n(s) mathematical expectation is 0 statistical property, phase estimation value

Also be comparison near true error, just degree of closeness is poor.However, exist than after the mistake estimation phase compensation original signal with this, the focusing situation of image can improve, and real point of maximum intensity can be therefore than giving prominence to before the compensation.If at this moment carry out iteration, | G _n(s) | will be than increasing for the first time, the precision of phase error estimation and phase error can continue to improve thus.This process can be carried out repeatedly, and the maximal value on all range units is judged does not have mistake, and continue iteration and will can not improve the effect of focusing this moment.What influence focusing effect is not only position and the width that peaked selection also comprises window, and these two factors are interrelated, needs constantly to adjust in the lump the width and the position of window in iterative process.

Therefore, iterative process is progressively to retrieve apart from maximal value on certain unit, and the process of progressively reducing the truncated window width from about the initial point respectively.Describing below and how avoiding overconvergence and two problems of misconvergence, this process is the present invention's the most key step of burnt method of meeting again.

For fear of overconvergence, the each iteration in the iterative process can only be utilized the new that part of information that produces of last iteration.When searching out new maximal value, after ring shift is blocked, the picture signal contravariant is changed to phase history, and the phase place that the last time is revised removed from phase history, and then carry out new phase estimation, so just can guarantee in a new round is estimated, only to have utilized the more new portion of information, and avoid the mistake of overconvergence.

Avoiding misconvergence mainly is by the determining positions of truncated window, the position of truncated window determines it also is a problem that progressively solves in iterative process, the very wide window of original state progressively reduces in iterative process, but this degree that reduces is asymmetric about to initial point, that is to say that when initial point the left window stops to reduce right side window may also will continue to reduce.Final window width of selecting and position are as shown in Figure 3.

For understanding above-mentioned iterative process better, the iteration error correction procedure is launched to describe, as shown in Figure 4.Raw information obtains updated information after handling through wrong correction algorithm, although updated information is compared with raw information, extent of error reduces, but still can not meet the demands, and therefore needs iteration, further corrects a mistake.In the error correction iterative process, can not can cause the mistake convergence like that very soon directly with the following iteration link of lastest imformation input.Correct method is to extract supplementary from lastest imformation, will import next step wrong correction algorithm after supplementary and the raw information mixing.Therefore the definite and extraction of supplementary is the most important in the error correction iterative process.

With regard to the burnt process of reunion of the present invention, supplementary is each position apart from bright spot on the resolution element.That is to say the position of having only bright spot in iterative process in change, other information all remain unchanged.In case the position of bright spot is no longer upgraded, iterative process has also just stopped.The present invention relies on the iterative process of phase error estimation and phase error just to make the phase error estimation and phase error of distribution objectives and compensation come true just.

The complex pattern territory is returned through inverse fourier transform again in the orientation after phase error compensation is corrected, the L-SAR image after burnt processing that promptly obtain meeting again.

Fig. 5 (a) provides the imaging side bitmap of certain range unit that obtains through conventional imaging processing, among the figure horizontal ordinate be the orientation to, ordinate is an amplitude.There is a tangible point target in figure central authorities.As seen from the figure, this point target seriously defocuses.Blooming effect is caused by the kinetic phase error of aircraft non-stationary.Point target among the figure has two peak values, and the speed of this explanation aircraft is carved at a time significant variation has been taken place.Fig. 5 (b) provides the point target orientation diagram after the present invention focuses on again, and is same, among the figure horizontal ordinate be the orientation to, ordinate is an amplitude.This point target has reached the ideal limit resolution of radar.From Fig. 5 (a) and contrast (b) as seen, handle back point target azimuthal resolution and be greatly improved through focusing on again.

Fig. 6 (a) has provided the L-SAR image that conventional imaging is handled, and is obviously image blurring, seriously defocuses.Image after Fig. 6 (b) provides and focuses on again with the present invention, the visual effect of visual picture is improved quite remarkable, and ground object target is clear easily to be distinguished.

Should be pointed out that the above embodiment can make those skilled in the art more fully understand the present invention, but do not limit the present invention in any way.Therefore, although this instructions has been described in detail the present invention with reference to drawings and embodiments,, it will be appreciated by those skilled in the art that still and can make amendment or be equal to replacement the present invention; And all do not break away from the technical scheme and the improvement thereof of spirit of the present invention and technical spirit, and it all should be encompassed in the middle of the protection domain of patent of the present invention.

Claims (3)

1. one kind lacks realization airborne radar reunion focusing imaging method under the high-precision motion compensating system, and comprise following two big steps: 1. original signal is through generating initial synthetic-aperture radar complex pattern based on the imaging processing of removing slope; 2. be input with the initial complex pattern of synthetic-aperture radar that goes in 1. gradient method to generate, measure the go forward side by side complex pattern of line phase compensation of Doppler signal phase error based on improved phase gradient autofocus algorithm and focus on again,

It is characterized in that 2. step as described in the preamble adopts following details technical characterictic,

When (1) ring shift is handled, in each range gate, select the strongest scattering point,, make point target the strongest on each range gate on the position of orientation time 0, line up vertical line its ring shift to 0 position,

(2) during windowing process, iterative process is progressively retrieved apart from maximal value on certain unit, and progressively reduces the truncated window width respectively from about initial point,

(3) in the iteration error correction procedure, adopt and from lastest imformation, extract supplementary, supplementary and next step wrong method of correcting of raw information mixing back input are avoided overconvergence and misconvergence.

2. according to claim 1ly a kind ofly realize airborne radar reunion focusing imaging method under the high-precision motion compensating system lacking, it is characterized in that:

Feature in the described claim 1 (2), adopt following iterative process, be described below: width and the position thereof of in iterative process, constantly adjusting peaked position and truncated window in the lump, progressively determine actual window width and position thereof, real maximum point target is displayed, the original state of iteration is: the maximal value that obtains might not be real maximum value position, the width of setting truncated window is fully greater than the width of actual needs, and the position of window is about former point symmetry, the desirable end-state of iteration is: the maximal value that obtains is the strongest scattering point of reality, the width of window be just point target in the orientation to diffusion breadth, the position of window is determined according to the distribution situation of point target energy about initial point, not necessarily about former point symmetry

Feature in the described claim 1 (3), adopt following process, be described below: for fear of overconvergence, each iteration in the iterative process can only utilize the new that part of information that produces of last iteration when searching out new maximal value, after ring shift is blocked, the picture signal contravariant is changed to phase history, and the phase place that the last time is revised removed from phase history, and then carry out new phase estimation, thereby guarantee in a new round is estimated, only to utilize the more new portion of information, and avoid the mistake of overconvergence, avoid misconvergence mainly by the determining positions of truncated window, the position of truncated window determines it is a problem that progressively solves in iterative process, the very wide window of original state progressively reduces in iterative process, but this degree that reduces is asymmetric about to initial point, that is to say that when initial point the left window stops to reduce right side window may also will continue to reduce.

3. realize airborne radar reunion focusing imaging method under a kind of shortage high-precision motion compensating according to claim 1 system, it is characterized in that:

Feature in the described claim 1 (3), supplementary is each position apart from bright spot on the resolution element, that is to say that the position of having only bright spot in iterative process is in change, other information all remain unchanged, in case the position of bright spot is no longer upgraded, iterative process also just stops, and the orientation is after phase error compensation is corrected, return the complex pattern territory through inverse fourier transform again, the image after burnt processing that promptly obtain meeting again.

Images